This is an archive of the discontinued LLVM Phabricator instance.

Differential D145441

[AMDGPU] Define data layout entries for buffers
ClosedPublic

Authored by krzysz00 on Mar 6 2023, 3:23 PM.

Details

Reviewers
nhaehnle
arsenm
b-sumner
piotr
sstefan1
jdoerfert
Commits
rGf9c1ede2543b: [AMDGPU] Define data layout entries for buffers
Summary

Per discussion at
https://discourse.llvm.org/t/representing-buffer-descriptors-in-the-amdgpu-target-call-for-suggestions/68798,
we define two new address spaces for AMDGCN targets.

The first is address space 7, a non-integral address space (which was
already in the data layout) that has 160-bit pointers (which are
256-bit aligned) and uses a 32-bit offset. These pointers combine a
128-bit buffer descriptor and a 32-bit offset, and will be usable with
normal LLVM operations (load, store, GEP). However, they will be
rewritten out of existence before code generation.

The second of these is address space 8, the address space for "buffer
resources". These will be used to represent the resource arguments to
buffer instructions, and new buffer intrinsics will be defined that
take them instead of <4 x i32> as resource arguments. ptr
addrspace(8). These pointers are 128-bits long (with the same
alignment). They must not be used as the arguments to getelementptr or
otherwise used in address computations, since they can have
arbitrarily complex inherent addressing semantics that can't be
represented in LLVM. Even though, like their address space 7 cousins,
these pointers have deterministic ptrtoint/inttoptr semantics, they
are defined to be non-integral in order to prevent optimizations that
rely on pointers being a [0, [addr_max]] value from applying to them.

Future work includes:

  • Defining new buffer intrinsics that take ptr addrspace(8) resources.
  • A late rewrite to turn address space 7 operations into buffer

intrinsics and offset computations.

This commit also updates the "fallback address space" for buffer
intrinsics to the buffer resource, and updates the alias analysis
table.

Depends on D143437

Diff Detail

Event Timeline

krzysz00 created this revision.Mar 6 2023, 3:23 PM
Herald added a project: Restricted Project. · View Herald TranscriptMar 6 2023, 3:23 PM
Herald added subscribers: kosarev, jeroen.dobbelaere, foad and 14 others. · View Herald Transcript
krzysz00 requested review of this revision.Mar 6 2023, 3:23 PM
Herald added projects: Restricted Project, Restricted Project. · View Herald TranscriptMar 6 2023, 3:23 PM
Herald added subscribers: llvm-commits, cfe-commits, • pcwang-thead, wdng. · View Herald Transcript
Harbormaster completed remote builds in B217729: Diff 502829.Mar 6 2023, 8:43 PM
foad added a comment.Mar 7 2023, 1:51 AM

Just my 2p: it feels a bit premature to commit patches for this. It feels more like something you could prototype on a branch somewhere and come back when you have more experience with how it all works out in practice.

But I don't actually object to the patch, if the other reviewers are happy and it doens't break anything.

The first is address space 7, a non-integral address space (which was
already in the data layout) that has 160-bit pointers (which are
256-bit aligned)

Any particular reason for choosing 256-bit alignment?

However, they must not be used as the arguments to
getelementptr or otherwise used in address computations

I don't understand what kind of rule this is and how it would be enforced. Is it something that will be written into the IR LangRef?

This commit also updates the "fallback address space" for buffer
intrinsics to the buffer resource,

It's not clear to me that this is any more or less correct, since 7 and 8 behave identically wrt alias analysis don't they?

krzysz00 added a comment.Mar 7 2023, 8:12 AM

@foad I was trying to avoid sending in one mega-patch that has the entire prototype.

As to your comments:

  • Why 256-bit alignment? It's the next power of 2, and using an alignment that isn't a power of 2 causes an assertion failure
  • Re not being allowed to use address space 8 pointers in the usual LLVM operations, I'd call that a target-specific rule about what address space 8 means. I figure this is something that we can check in some sort of AMDGPU-specific validation
  • I changed the fallback address space because the plan is that those intrinsics will be taking address space 8 arguments in the future
nhaehnle added a comment.EditedMar 7 2023, 12:08 PM
In D145441#4175428, @krzysz00 wrote:

@foad I was trying to avoid sending in one mega-patch that has the entire prototype.

The best way to do this is to work on a branch that you rebase regularly. You can push the branch to a personal clone of llvm-project on GitHub and point folks there from e.g. Discourse, and for the purpose of Phabricator you can set up a "Stack" via the "Edit Related Revisions" option.

If you're using arc, then git rebase -i -x "arc diff @^" is one way to update a bunch of patches at once. There are probably other tools that people have written.

These [addrspace(8) pointers] are, however, integral, since inttoptr and ptrtoint behave deterministically and reasonably.

That doesn't make sense to me. I thought "integral" means that

inttoptr(ptrtoint(p) + offset)

is the same (in terms of address, and potentially modulo provenance questions) as

gep i8, p, offset

But that is rather untrue for addrspace(8), isn't it?

llvm/docs/AMDGPUUsage.rst
794

Typo: experimental

804–805

Presumably it is only permitted if the underlying descriptor satisfies all the rules mentioned for addrspace(7).

krzysz00 updated this revision to Diff 507869.Mar 23 2023, 1:47 PM

Fix typo, rebase while making next patch

krzysz00 updated this revision to Diff 507870.Mar 23 2023, 1:50 PM

... actually fix the typo

krzysz00 added a child revision: D146761: [AMDGPU] Make buffer intrinsics pointer-valued.Mar 23 2023, 2:38 PM
krzysz00 updated this revision to Diff 509761.Mar 30 2023, 11:58 AM
krzysz00 edited the summary of this revision. (Show Details)
krzysz00 added a parent revision: D147245: [AMDGPU] Make amdgcn.s.buffer.load a memory-reading intrinsic.

Rebase for splitting s.buffer.load effect change into earlier patch

Harbormaster completed remote builds in B222808: Diff 509761.Mar 30 2023, 1:26 PM
krzysz00 updated this revision to Diff 510056.Mar 31 2023, 9:20 AM
krzysz00 edited the summary of this revision. (Show Details)

Per discussion on the s.buffer.load revision, don't make those changes, and so revert this patch back to what it used to be.

Harbormaster completed remote builds in B223029: Diff 510056.Mar 31 2023, 9:21 AM
krzysz00 updated this revision to Diff 510848.Apr 4 2023, 10:20 AM

Rebase data layout changes

Harbormaster completed remote builds in B223611: Diff 510848.Apr 4 2023, 10:21 AM
krzysz00 updated this revision to Diff 512488.Apr 11 2023, 9:12 AM

Rebase, since we have addrspace 128

Harbormaster completed remote builds in B224802: Diff 512488.Apr 11 2023, 9:13 AM
krzysz00 updated this revision to Diff 514232.Apr 17 2023, 7:50 AM

Fix new merge conflicts

Harbormaster completed remote builds in B226101: Diff 514232.Apr 17 2023, 7:51 AM
krzysz00 removed a parent revision: D147245: [AMDGPU] Make amdgcn.s.buffer.load a memory-reading intrinsic.Apr 21 2023, 8:45 AM
krzysz00 added a child revision: D147547: [AMDGPU] Add buffer intrinsics that take resources as pointers.
krzysz00 updated this revision to Diff 516953.Apr 25 2023, 3:18 PM
krzysz00 edited the summary of this revision. (Show Details)

Make address space 8 non-integral as well because that's a good idea and we'll have an address space cast intrinsic.

arsenm accepted this revision.May 1 2023, 1:26 PM

LGTM. It may not be the end step and may be annoying to make further changes from this point, but on its own this isn't worse than before

llvm/lib/Target/AMDGPU/SIISelLowering.cpp
998

Should this todo be dropped?

This revision is now accepted and ready to land.May 1 2023, 1:26 PM
krzysz00 updated this revision to Diff 518843.May 2 2023, 1:19 PM

Rebase to handle new legalization tests.

Harbormaster completed remote builds in B229517: Diff 518843.May 2 2023, 7:52 PM
This revision was landed with ongoing or failed builds.May 3 2023, 8:26 AM
Closed by commit rGf9c1ede2543b: [AMDGPU] Define data layout entries for buffers (authored by krzysz00). · Explain Why
This revision was automatically updated to reflect the committed changes.
krzysz00 added a commit: rGf9c1ede2543b: [AMDGPU] Define data layout entries for buffers.
jplehr added a subscriber: jplehr.May 3 2023, 9:10 AM

Hey Krzysztof,
I believe this broke the openmp offload buildbot https://lab.llvm.org/buildbot/#/builders/193/builds/30576

krzysz00 added a reverting change: D149758: Revert "[AMDGPU] Define data layout entries for buffers".May 3 2023, 9:10 AM
krzysz00 added a reverting change: rG3f2fbe92d0f4: Revert "[AMDGPU] Define data layout entries for buffers".May 3 2023, 9:11 AM
krzysz00 mentioned this in D149776: Re-land "[AMDGPU] Define data layout entries for buffers"".May 3 2023, 11:52 AM
krzysz00 mentioned this in rGf0415f2a456d: Re-land "[AMDGPU] Define data layout entries for buffers"".May 3 2023, 12:44 PM

Revision Contents

PathSize
clang/
lib/
Basic/
Targets/
2 lines
test/
CodeGen/
4 lines
CodeGenOpenCL/
2 lines
llvm/
docs/
21 lines
lib/
IR/
27 lines
Target/
AMDGPU/
97 lines
27 lines
12 lines
3 lines
11 lines
test/
CodeGen/
AMDGPU/
GlobalISel/
16 lines
16 lines
16 lines
16 lines
8 lines
96 lines
168 lines
96 lines
48 lines
8 lines
8 lines
282 lines
16 lines
12 lines
76 lines
32 lines
40 lines
16 lines
10 lines
2 lines
40 lines
20 lines
12 lines
70 lines
44 lines
24 lines
66 lines
32 lines
18 lines
40 lines
16 lines
50 lines
14 lines
10 lines
2 lines
40 lines
24 lines
20 lines
32 lines
20 lines
12 lines
20 lines
24 lines
14 lines
16 lines
20 lines
10 lines
10 lines
10 lines
2 lines
2 lines
16 lines
16 lines
16 lines
16 lines
8 lines
200 lines
2 lines
4 lines
2 lines
2 lines
2 lines
2 lines
18 lines
15 lines
2 lines
2 lines
2 lines
62 lines
2 lines
2 lines
2 lines
2 lines
MIR/
AMDGPU/
4 lines
Instrumentation/
AddressSanitizer/
AMDGPU/
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
Transforms/
AlignmentFromAssumptions/
2 lines
AtomicExpand/
AMDGPU/
2 lines
EarlyCSE/
AMDGPU/
2 lines
IndVarSimplify/
AMDGPU/
2 lines
InferAddressSpaces/
AMDGPU/
2 lines
2 lines
X86/
2 lines
Inline/
AMDGPU/
2 lines
InstCombine/
AMDGPU/
2 lines
2 lines
LoadStoreVectorizer/
AMDGPU/
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
2 lines
LoopLoadElim/
2 lines
LoopStrengthReduce/
AMDGPU/
2 lines
2 lines
2 lines
2 lines
2 lines
OpenMP/
2 lines
2 lines
2 lines
SLPVectorizer/
AMDGPU/
2 lines
VectorCombine/
AMDGPU/
2 lines
2 lines
unittests/
Bitcode/
28 lines

Diff 507869

clang/lib/Basic/Targets/AMDGPU.cpp

Show All 27 Lines
// getPointerWidthV(). // getPointerWidthV().
static const char *const DataLayoutStringR600 = static const char *const DataLayoutStringR600 =
"e-p:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128" "e-p:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128"
"-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1"; "-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1";
static const char *const DataLayoutStringAMDGCN = static const char *const DataLayoutStringAMDGCN =
"e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32" "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32"
"-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128" "-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128"
"-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1" "-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1"
"-ni:7"; "-ni:7";
const LangASMap AMDGPUTargetInfo::AMDGPUDefIsGenMap = { const LangASMap AMDGPUTargetInfo::AMDGPUDefIsGenMap = {
Generic, // Default Generic, // Default
Global, // opencl_global Global, // opencl_global
Local, // opencl_local Local, // opencl_local
Constant, // opencl_constant Constant, // opencl_constant
▲ Show 20 Lines • Show All 478 Lines • Show Last 20 Lines

clang/test/CodeGen/target-data.c

Show First 20 Lines • Show All 170 Lines • ▼ Show 20 Lines
// R600: target datalayout = "e-p:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1" // R600: target datalayout = "e-p:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1"
// RUN: %clang_cc1 -triple r600-unknown -target-cpu cayman -o - -emit-llvm %s \ // RUN: %clang_cc1 -triple r600-unknown -target-cpu cayman -o - -emit-llvm %s \
// RUN: | FileCheck %s -check-prefix=R600D // RUN: | FileCheck %s -check-prefix=R600D
// R600D: target datalayout = "e-p:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1" // R600D: target datalayout = "e-p:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1"
// RUN: %clang_cc1 -triple amdgcn-unknown -target-cpu hawaii -o - -emit-llvm %s \ // RUN: %clang_cc1 -triple amdgcn-unknown -target-cpu hawaii -o - -emit-llvm %s \
// RUN: | FileCheck %s -check-prefix=R600SI // RUN: | FileCheck %s -check-prefix=R600SI
// R600SI: target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" // R600SI: target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
// Test default -target-cpu // Test default -target-cpu
// RUN: %clang_cc1 -triple amdgcn-unknown -o - -emit-llvm %s \ // RUN: %clang_cc1 -triple amdgcn-unknown -o - -emit-llvm %s \
// RUN: | FileCheck %s -check-prefix=R600SIDefault // RUN: | FileCheck %s -check-prefix=R600SIDefault
// R600SIDefault: target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" // R600SIDefault: target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
// RUN: %clang_cc1 -triple arm64-unknown -o - -emit-llvm %s | \ // RUN: %clang_cc1 -triple arm64-unknown -o - -emit-llvm %s | \
// RUN: FileCheck %s -check-prefix=AARCH64 // RUN: FileCheck %s -check-prefix=AARCH64
// AARCH64: target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128" // AARCH64: target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
// RUN: %clang_cc1 -triple arm64_32-apple-ios7.0 -o - -emit-llvm %s | \ // RUN: %clang_cc1 -triple arm64_32-apple-ios7.0 -o - -emit-llvm %s | \
// RUN: FileCheck %s -check-prefix=AARCH64-ILP32 // RUN: FileCheck %s -check-prefix=AARCH64-ILP32
// AARCH64-ILP32: target datalayout = "e-m:o-p:32:32-i64:64-i128:128-n32:64-S128" // AARCH64-ILP32: target datalayout = "e-m:o-p:32:32-i64:64-i128:128-n32:64-S128"
▲ Show 20 Lines • Show All 78 Lines • Show Last 20 Lines

clang/test/CodeGenOpenCL/amdgpu-env-amdgcn.cl

// RUN: %clang_cc1 %s -O0 -triple amdgcn -emit-llvm -o - | FileCheck %s // RUN: %clang_cc1 %s -O0 -triple amdgcn -emit-llvm -o - | FileCheck %s
// RUN: %clang_cc1 %s -O0 -triple amdgcn---opencl -emit-llvm -o - | FileCheck %s // RUN: %clang_cc1 %s -O0 -triple amdgcn---opencl -emit-llvm -o - | FileCheck %s
// CHECK: target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" // CHECK: target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
void foo(void) {} void foo(void) {}

llvm/docs/AMDGPUUsage.rst

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 669 Lines • ▼ Show 20 Lines .. table:: AMDGPU Address Spaces
Generic 0 flat flat 64 0x0000000000000000 Generic 0 flat flat 64 0x0000000000000000
Global 1 global global 64 0x0000000000000000 Global 1 global global 64 0x0000000000000000
Region 2 N/A GDS 32 *not implemented for AMDHSA* Region 2 N/A GDS 32 *not implemented for AMDHSA*
Local 3 group LDS 32 0xFFFFFFFF Local 3 group LDS 32 0xFFFFFFFF
Constant 4 constant *same as global* 64 0x0000000000000000 Constant 4 constant *same as global* 64 0x0000000000000000
Private 5 private scratch 32 0xFFFFFFFF Private 5 private scratch 32 0xFFFFFFFF
Constant 32-bit 6 *TODO* 0x00000000 Constant 32-bit 6 *TODO* 0x00000000
Buffer Fat Pointer (experimental) 7 *TODO* Buffer Fat Pointer (experimental) 7 *TODO*
Buffer Resource (experimental) 8 *TODO*
================================= =============== =========== ================ ======= ============================ ================================= =============== =========== ================ ======= ============================
**Generic** **Generic**
The generic address space is supported unless the *Target Properties* column The generic address space is supported unless the *Target Properties* column
of :ref:`amdgpu-processor-table` specifies *Does not support generic address of :ref:`amdgpu-processor-table` specifies *Does not support generic address
space*. space*.
The generic address space uses the hardware flat address support for two fixed The generic address space uses the hardware flat address support for two fixed
▲ Show 20 Lines • Show All 92 Lines • ▼ Show 20 Lines**Buffer Fat Pointer**
The buffer fat pointer is an experimental address space that is currently The buffer fat pointer is an experimental address space that is currently
unsupported in the backend. It exposes a non-integral pointer that is in unsupported in the backend. It exposes a non-integral pointer that is in
the future intended to support the modelling of 128-bit buffer descriptors the future intended to support the modelling of 128-bit buffer descriptors
plus a 32-bit offset into the buffer (in total encapsulating a 160-bit plus a 32-bit offset into the buffer (in total encapsulating a 160-bit
*pointer*), allowing normal LLVM load/store/atomic operations to be used to *pointer*), allowing normal LLVM load/store/atomic operations to be used to
model the buffer descriptors used heavily in graphics workloads targeting model the buffer descriptors used heavily in graphics workloads targeting
the backend. the backend.
The buffer descriptor used to construct a buffer fat pointer must be *raw*:
the stride must be 0, the "add tid" flag bust be 0, the swizzle enable bits
must be off, and the extent must be measured in bytes. (On subtargets where
bounds checking may be disabled, buffer fat pointers may choose to enable
it or not).
**Buffer Resource**
The buffer resource is an experemntal address space that is currently unsupported
nhaehnleUnsubmitted
Not Done
ReplyInline Actions

Typo: experimental

nhaehnle: Typo: experimental
in the backend. It exposes a non-integral pointer that will represent a 128-bit
buffer descriptor resource.
Since, in general, a buffer resource supports complex addressing modes that cannot
be easily represented in LLVM (such as implicit swizzled access to structured
buffers), it is **illegal** to perform non-trivial address computations, such as
``getelementptr`` operations, on buffer resources. They may be passed to
AMDGPU buffer intrinsics, and they may be converted to and from ``i128``.
Casting a buffer resource to a bufer fat pointer is permitted and adds an offset
of 0.
nhaehnleUnsubmitted
Not Done
ReplyInline Actions

Presumably it is only permitted if the underlying descriptor satisfies all the rules mentioned for addrspace(7).

nhaehnle: Presumably it is only permitted if the underlying descriptor satisfies all the rules mentioned…
.. _amdgpu-memory-scopes: .. _amdgpu-memory-scopes:
Memory Scopes Memory Scopes
------------- -------------
This section provides LLVM memory synchronization scopes supported by the AMDGPU This section provides LLVM memory synchronization scopes supported by the AMDGPU
backend memory model when the target triple OS is ``amdhsa`` (see backend memory model when the target triple OS is ``amdhsa`` (see
:ref:`amdgpu-amdhsa-memory-model` and :ref:`amdgpu-target-triples`). :ref:`amdgpu-amdhsa-memory-model` and :ref:`amdgpu-target-triples`).
▲ Show 20 Lines • Show All 14,395 Lines • Show Last 20 Lines

llvm/lib/IR/AutoUpgrade.cpp

//===-- AutoUpgrade.cpp - Implement auto-upgrade helper functions ---------===// //===-- AutoUpgrade.cpp - Implement auto-upgrade helper functions ---------===//
// //
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information. // See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
// //
//===----------------------------------------------------------------------===// //===----------------------------------------------------------------------===//
// //
// This file implements the auto-upgrade helper functions. // This file implements the auto-upgrade helper functions.
// This is where deprecated IR intrinsics and other IR features are updated to // This is where deprecated IR intrinsics and other IR features are updated to
// current specifications. // current specifications.
// //
//===----------------------------------------------------------------------===// //===----------------------------------------------------------------------===//
#include "llvm/IR/AutoUpgrade.h" #include "llvm/IR/AutoUpgrade.h"
#include "llvm/ADT/StringRef.h"
#include "llvm/ADT/StringSwitch.h" #include "llvm/ADT/StringSwitch.h"
#include "llvm/BinaryFormat/Dwarf.h" #include "llvm/BinaryFormat/Dwarf.h"
#include "llvm/IR/Constants.h" #include "llvm/IR/Constants.h"
#include "llvm/IR/DebugInfo.h" #include "llvm/IR/DebugInfo.h"
#include "llvm/IR/DebugInfoMetadata.h" #include "llvm/IR/DebugInfoMetadata.h"
#include "llvm/IR/DiagnosticInfo.h" #include "llvm/IR/DiagnosticInfo.h"
#include "llvm/IR/Function.h" #include "llvm/IR/Function.h"
#include "llvm/IR/IRBuilder.h" #include "llvm/IR/IRBuilder.h"
▲ Show 20 Lines • Show All 4,934 Lines • ▼ Show 20 LinesMDNode *llvm::upgradeInstructionLoopAttachment(MDNode &N) {
for (Metadata *MD : T->operands()) for (Metadata *MD : T->operands())
Ops.push_back(upgradeLoopArgument(MD)); Ops.push_back(upgradeLoopArgument(MD));
return MDTuple::get(T->getContext(), Ops); return MDTuple::get(T->getContext(), Ops);
} }
std::string llvm::UpgradeDataLayoutString(StringRef DL, StringRef TT) { std::string llvm::UpgradeDataLayoutString(StringRef DL, StringRef TT) {
Triple T(TT); Triple T(TT);
// For AMDGPU we uprgrade older DataLayouts to include the default globals // The only data layout upgrades needed for pre-GCN are setting the address
// address space of 1. // space of globals to 1.
if (T.isAMDGPU() && !DL.contains("-G") && !DL.startswith("G")) { if (T.isAMDGPU() && !T.isAMDGCN() && !DL.contains("-G") &&
!DL.startswith("G")) {
return DL.empty() ? std::string("G1") : (DL + "-G1").str(); return DL.empty() ? std::string("G1") : (DL + "-G1").str();
} }
if (T.isRISCV64()) { if (T.isRISCV64()) {
// Make i32 a native type for 64-bit RISC-V. // Make i32 a native type for 64-bit RISC-V.
auto I = DL.find("-n64-"); auto I = DL.find("-n64-");
if (I != StringRef::npos) if (I != StringRef::npos)
return (DL.take_front(I) + "-n32:64-" + DL.drop_front(I + 5)).str(); return (DL.take_front(I) + "-n32:64-" + DL.drop_front(I + 5)).str();
return DL.str(); return DL.str();
} }
std::string Res = DL.str(); std::string Res = DL.str();
// AMDGCN data layout upgrades.
if (T.isAMDGCN()) {
// Define address spaces for constants.
if (!DL.contains("-G") && !DL.starts_with("G"))
Res.append(Res.empty() ? "G1" : "-G1");
// Add sizing for address spaces 7 and 8 (fat raw buffers and buffer
// resources) An empty data layout has already been upgraded to G1 by now.
if (!DL.contains("-p7") && !DL.startswith("p7"))
Res.append("-p7:160:256:256:32");
if (!DL.contains("-p8") && !DL.startswith("p8"))
Res.append("-p8:128:128");
// Add missing non-integral declarations.
if (!DL.contains("-ni") && !DL.startswith("ni"))
Res.append("-ni:7");
return Res;
}
if (!T.isX86()) if (!T.isX86())
return Res; return Res;
// If the datalayout matches the expected format, add pointer size address // If the datalayout matches the expected format, add pointer size address
// spaces to the datalayout. // spaces to the datalayout.
std::string AddrSpaces = "-p270:32:32-p271:32:32-p272:64:64"; std::string AddrSpaces = "-p270:32:32-p271:32:32-p272:64:64";
if (!DL.contains(AddrSpaces)) { if (!DL.contains(AddrSpaces)) {
SmallVector<StringRef, 4> Groups; SmallVector<StringRef, 4> Groups;
▲ Show 20 Lines • Show All 55 Lines • Show Last 20 Lines

llvm/lib/Target/AMDGPU/AMDGPU.h

Show First 20 Lines • Show All 363 Lines • ▼ Show 20 Lines
/// OpenCL uses address spaces to differentiate between /// OpenCL uses address spaces to differentiate between
/// various memory regions on the hardware. On the CPU /// various memory regions on the hardware. On the CPU
/// all of the address spaces point to the same memory, /// all of the address spaces point to the same memory,
/// however on the GPU, each address space points to /// however on the GPU, each address space points to
/// a separate piece of memory that is unique from other /// a separate piece of memory that is unique from other
/// memory locations. /// memory locations.
namespace AMDGPUAS { namespace AMDGPUAS {
enum : unsigned { enum : unsigned {
// The maximum value for flat, generic, local, private, constant and region. // The maximum value for flat, generic, local, private, constant and region.
MAX_AMDGPU_ADDRESS = 7, MAX_AMDGPU_ADDRESS = 8,
FLAT_ADDRESS = 0, ///< Address space for flat memory. FLAT_ADDRESS = 0, ///< Address space for flat memory.
GLOBAL_ADDRESS = 1, ///< Address space for global memory (RAT0, VTX0). GLOBAL_ADDRESS = 1, ///< Address space for global memory (RAT0, VTX0).
REGION_ADDRESS = 2, ///< Address space for region memory. (GDS) REGION_ADDRESS = 2, ///< Address space for region memory. (GDS)
CONSTANT_ADDRESS = 4, ///< Address space for constant memory (VTX2). CONSTANT_ADDRESS = 4, ///< Address space for constant memory (VTX2).
LOCAL_ADDRESS = 3, ///< Address space for local memory. LOCAL_ADDRESS = 3, ///< Address space for local memory.
PRIVATE_ADDRESS = 5, ///< Address space for private memory. PRIVATE_ADDRESS = 5, ///< Address space for private memory.
CONSTANT_ADDRESS_32BIT = 6, ///< Address space for 32-bit constant memory. CONSTANT_ADDRESS_32BIT = 6, ///< Address space for 32-bit constant memory.
BUFFER_FAT_POINTER = 7, ///< Address space for 160-bit buffer fat pointers. BUFFER_FAT_POINTER = 7, ///< Address space for 160-bit buffer fat pointers.
///< Not used in backend.
BUFFER_RESOURCE = 8, ///< Address space for 128-bit buffer resources.
/// Address space for direct addressable parameter memory (CONST0). /// Address space for direct addressable parameter memory (CONST0).
PARAM_D_ADDRESS = 6, PARAM_D_ADDRESS = 6,
/// Address space for indirect addressable parameter memory (VTX1). /// Address space for indirect addressable parameter memory (VTX1).
PARAM_I_ADDRESS = 7, PARAM_I_ADDRESS = 7,
// Do not re-order the CONSTANT_BUFFER_* enums. Several places depend on // Do not re-order the CONSTANT_BUFFER_* enums. Several places depend on
// this order to be able to dynamically index a constant buffer, for // this order to be able to dynamically index a constant buffer, for
// example: // example:
// //
// ConstantBufferAS = CONSTANT_BUFFER_0 + CBIdx // ConstantBufferAS = CONSTANT_BUFFER_0 + CBIdx
CONSTANT_BUFFER_0 = 8, CONSTANT_BUFFER_0 = 8,
CONSTANT_BUFFER_1 = 9, CONSTANT_BUFFER_1 = 9,
CONSTANT_BUFFER_2 = 10, CONSTANT_BUFFER_2 = 10,
CONSTANT_BUFFER_3 = 11, CONSTANT_BUFFER_3 = 11,
CONSTANT_BUFFER_4 = 12, CONSTANT_BUFFER_4 = 12,
CONSTANT_BUFFER_5 = 13, CONSTANT_BUFFER_5 = 13,
CONSTANT_BUFFER_6 = 14, CONSTANT_BUFFER_6 = 14,
CONSTANT_BUFFER_7 = 15, CONSTANT_BUFFER_7 = 15,
CONSTANT_BUFFER_8 = 16, CONSTANT_BUFFER_8 = 16,
CONSTANT_BUFFER_9 = 17, CONSTANT_BUFFER_9 = 17,
CONSTANT_BUFFER_10 = 18, CONSTANT_BUFFER_10 = 18,
CONSTANT_BUFFER_11 = 19, CONSTANT_BUFFER_11 = 19,
CONSTANT_BUFFER_12 = 20, CONSTANT_BUFFER_12 = 20,
CONSTANT_BUFFER_13 = 21, CONSTANT_BUFFER_13 = 21,
CONSTANT_BUFFER_14 = 22, CONSTANT_BUFFER_14 = 22,
CONSTANT_BUFFER_15 = 23, CONSTANT_BUFFER_15 = 23,
// Some places use this if the address space can't be determined. // Some places use this if the address space can't be determined.
UNKNOWN_ADDRESS_SPACE = ~0u, UNKNOWN_ADDRESS_SPACE = ~0u,
}; };
} }
namespace AMDGPU { namespace AMDGPU {
// FIXME: Missing constant_32bit // FIXME: Missing constant_32bit
inline bool isFlatGlobalAddrSpace(unsigned AS) { inline bool isFlatGlobalAddrSpace(unsigned AS) {
return AS == AMDGPUAS::GLOBAL_ADDRESS || return AS == AMDGPUAS::GLOBAL_ADDRESS ||
AS == AMDGPUAS::FLAT_ADDRESS || AS == AMDGPUAS::FLAT_ADDRESS ||
Show All 14 Lines

llvm/lib/Target/AMDGPU/AMDGPUAliasAnalysis.cpp

Show First 20 Lines • Show All 41 Lines • ▼ Show 20 LinesAMDGPUAAWrapperPass::AMDGPUAAWrapperPass() : ImmutablePass(ID) {
initializeAMDGPUAAWrapperPassPass(*PassRegistry::getPassRegistry()); initializeAMDGPUAAWrapperPassPass(*PassRegistry::getPassRegistry());
} }
void AMDGPUAAWrapperPass::getAnalysisUsage(AnalysisUsage &AU) const { void AMDGPUAAWrapperPass::getAnalysisUsage(AnalysisUsage &AU) const {
AU.setPreservesAll(); AU.setPreservesAll();
} }
static AliasResult getAliasResult(unsigned AS1, unsigned AS2) { static AliasResult getAliasResult(unsigned AS1, unsigned AS2) {
static_assert(AMDGPUAS::MAX_AMDGPU_ADDRESS <= 7, "Addr space out of range"); static_assert(AMDGPUAS::MAX_AMDGPU_ADDRESS <= 8, "Addr space out of range");
if (AS1 > AMDGPUAS::MAX_AMDGPU_ADDRESS || AS2 > AMDGPUAS::MAX_AMDGPU_ADDRESS) if (AS1 > AMDGPUAS::MAX_AMDGPU_ADDRESS || AS2 > AMDGPUAS::MAX_AMDGPU_ADDRESS)
return AliasResult::MayAlias; return AliasResult::MayAlias;
#define ASMay AliasResult::MayAlias #define ASMay AliasResult::MayAlias
#define ASNo AliasResult::NoAlias #define ASNo AliasResult::NoAlias
// This array is indexed by address space value enum elements 0 ... to 7 // This array is indexed by address space value enum elements 0 ... to 8
static const AliasResult ASAliasRules[8][8] = { // clang-format off
/* Flat Global Region Group Constant Private Const32 Buf Fat Ptr */ static const AliasResult ASAliasRules[9][9] = {
/* Flat */ {ASMay, ASMay, ASNo, ASMay, ASMay, ASMay, ASMay, ASMay}, /* Flat Global Region Group Constant Private Const32 BufFatPtr BufRsrc */
/* Global */ {ASMay, ASMay, ASNo, ASNo, ASMay, ASNo, ASMay, ASMay}, /* Flat */ {ASMay, ASMay, ASNo, ASMay, ASMay, ASMay, ASMay, ASMay, ASMay},
/* Region */ {ASNo, ASNo, ASMay, ASNo, ASNo, ASNo, ASNo, ASNo}, /* Global */ {ASMay, ASMay, ASNo, ASNo, ASMay, ASNo, ASMay, ASMay, ASMay},
/* Group */ {ASMay, ASNo, ASNo, ASMay, ASNo, ASNo, ASNo, ASNo}, /* Region */ {ASNo, ASNo, ASMay, ASNo, ASNo, ASNo, ASNo, ASNo, ASNo},
/* Constant */ {ASMay, ASMay, ASNo, ASNo, ASNo, ASNo, ASMay, ASMay}, /* Group */ {ASMay, ASNo, ASNo, ASMay, ASNo, ASNo, ASNo, ASNo, ASNo},
/* Private */ {ASMay, ASNo, ASNo, ASNo, ASNo, ASMay, ASNo, ASNo}, /* Constant */ {ASMay, ASMay, ASNo, ASNo, ASNo, ASNo, ASMay, ASMay, ASMay},
/* Constant 32-bit */ {ASMay, ASMay, ASNo, ASNo, ASMay, ASNo, ASNo, ASMay}, /* Private */ {ASMay, ASNo, ASNo, ASNo, ASNo, ASMay, ASNo, ASNo, ASNo},
/* Buffer Fat Ptr */ {ASMay, ASMay, ASNo, ASNo, ASMay, ASNo, ASMay, ASMay} /* Constant 32-bit */ {ASMay, ASMay, ASNo, ASNo, ASMay, ASNo, ASNo, ASMay, ASMay},
/* Buffer Fat Ptr */ {ASMay, ASMay, ASNo, ASNo, ASMay, ASNo, ASMay, ASMay, ASMay},
/* Buffer Resource */ {ASMay, ASMay, ASNo, ASNo, ASMay, ASNo, ASMay, ASMay, ASMay},
}; };
// clang-format on
#undef ASMay #undef ASMay
#undef ASNo #undef ASNo
return ASAliasRules[AS1][AS2]; return ASAliasRules[AS1][AS2];
} }
AliasResult AMDGPUAAResult::alias(const MemoryLocation &LocA, AliasResult AMDGPUAAResult::alias(const MemoryLocation &LocA,
const MemoryLocation &LocB, AAQueryInfo &AAQI, const MemoryLocation &LocB, AAQueryInfo &AAQI,
▲ Show 20 Lines • Show All 66 Lines • Show Last 20 Lines

llvm/lib/Target/AMDGPU/AMDGPUTargetMachine.cpp

Show First 20 Lines • Show All 507 Lines • ▼ Show 20 Lines
static StringRef computeDataLayout(const Triple &TT) { static StringRef computeDataLayout(const Triple &TT) {
if (TT.getArch() == Triple::r600) { if (TT.getArch() == Triple::r600) {
// 32-bit pointers. // 32-bit pointers.
return "e-p:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128" return "e-p:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128"
"-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1"; "-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1";
} }
// 32-bit private, local, and region pointers. 64-bit global, constant and // 32-bit private, local, and region pointers. 64-bit global, constant and
// flat, non-integral buffer fat pointers. // flat. 160-bit non-integral fat buffer pointers that include a 128-bit
// buffer descriptor and a 32-bit offset, which are indexed by 32-bit values
// (address space 7), and 128-bit buffer resourcees (address space 8) which
// are integral (so inttoptr usage can be optpmized) but cannot be
// non-trivilally accessed by LLVM memory operations.
return "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32" return "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32"
"-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128" "-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:"
"-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1" "128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-"
"-ni:7"; "G1-ni:7";
} }
LLVM_READNONE LLVM_READNONE
static StringRef getGPUOrDefault(const Triple &TT, StringRef GPU) { static StringRef getGPUOrDefault(const Triple &TT, StringRef GPU) {
if (!GPU.empty()) if (!GPU.empty())
return GPU; return GPU;
// Need to default to a target with flat support for HSA. // Need to default to a target with flat support for HSA.
▲ Show 20 Lines • Show All 1,088 Lines • Show Last 20 Lines

llvm/lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp

Show First 20 Lines • Show All 354 Lines • ▼ Show 20 Linesunsigned GCNTTIImpl::getStoreVectorFactor(unsigned VF, unsigned StoreSize,
return VF; return VF;
} }
unsigned GCNTTIImpl::getLoadStoreVecRegBitWidth(unsigned AddrSpace) const { unsigned GCNTTIImpl::getLoadStoreVecRegBitWidth(unsigned AddrSpace) const {
if (AddrSpace == AMDGPUAS::GLOBAL_ADDRESS || if (AddrSpace == AMDGPUAS::GLOBAL_ADDRESS ||
AddrSpace == AMDGPUAS::CONSTANT_ADDRESS || AddrSpace == AMDGPUAS::CONSTANT_ADDRESS ||
AddrSpace == AMDGPUAS::CONSTANT_ADDRESS_32BIT || AddrSpace == AMDGPUAS::CONSTANT_ADDRESS_32BIT ||
AddrSpace == AMDGPUAS::BUFFER_FAT_POINTER) { AddrSpace == AMDGPUAS::BUFFER_FAT_POINTER ||
AddrSpace == AMDGPUAS::BUFFER_RESOURCE) {
return 512; return 512;
} }
if (AddrSpace == AMDGPUAS::PRIVATE_ADDRESS) if (AddrSpace == AMDGPUAS::PRIVATE_ADDRESS)
return 8 * ST->getMaxPrivateElementSize(); return 8 * ST->getMaxPrivateElementSize();
// Common to flat, global, local and region. Assume for unknown addrspace. // Common to flat, global, local and region. Assume for unknown addrspace.
return 128; return 128;
▲ Show 20 Lines • Show All 910 Lines • Show Last 20 Lines

llvm/lib/Target/AMDGPU/SIISelLowering.cpp

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 989 Lines • ▼ Show 20 Linesbool SITargetLowering::getTgtMemIntrinsic(IntrinsicInfo &Info,
if (const AMDGPU::RsrcIntrinsic *RsrcIntr = if (const AMDGPU::RsrcIntrinsic *RsrcIntr =
AMDGPU::lookupRsrcIntrinsic(IntrID)) { AMDGPU::lookupRsrcIntrinsic(IntrID)) {
AttributeList Attr = Intrinsic::getAttributes(CI.getContext(), AttributeList Attr = Intrinsic::getAttributes(CI.getContext(),
(Intrinsic::ID)IntrID); (Intrinsic::ID)IntrID);
MemoryEffects ME = Attr.getMemoryEffects(); MemoryEffects ME = Attr.getMemoryEffects();
if (ME.doesNotAccessMemory()) if (ME.doesNotAccessMemory())
return false; return false;
// TODO: Should images get their own address space? // TODO: Should images get their own address space?
arsenmUnsubmitted
Not Done
ReplyInline Actions

Should this todo be dropped?

arsenm: Should this todo be dropped?
Info.fallbackAddressSpace = AMDGPUAS::BUFFER_FAT_POINTER; Info.fallbackAddressSpace = AMDGPUAS::BUFFER_RESOURCE;
if (RsrcIntr->IsImage) if (RsrcIntr->IsImage)
Info.align.reset(); Info.align.reset();
Info.flags |= MachineMemOperand::MODereferenceable; Info.flags |= MachineMemOperand::MODereferenceable;
if (ME.onlyReadsMemory()) { if (ME.onlyReadsMemory()) {
unsigned MaxNumLanes = 4; unsigned MaxNumLanes = 4;
▲ Show 20 Lines • Show All 73 Lines • ▼ Show 20 Lines case Intrinsic::amdgcn_ds_fmax: {
if (!Vol->isZero()) if (!Vol->isZero())
Info.flags |= MachineMemOperand::MOVolatile; Info.flags |= MachineMemOperand::MOVolatile;
return true; return true;
} }
case Intrinsic::amdgcn_buffer_atomic_fadd: { case Intrinsic::amdgcn_buffer_atomic_fadd: {
Info.opc = ISD::INTRINSIC_W_CHAIN; Info.opc = ISD::INTRINSIC_W_CHAIN;
Info.memVT = MVT::getVT(CI.getOperand(0)->getType()); Info.memVT = MVT::getVT(CI.getOperand(0)->getType());
Info.fallbackAddressSpace = AMDGPUAS::BUFFER_FAT_POINTER; Info.fallbackAddressSpace = AMDGPUAS::BUFFER_RESOURCE;
Info.align.reset(); Info.align.reset();
Info.flags |= MachineMemOperand::MOLoad | MachineMemOperand::MOStore; Info.flags |= MachineMemOperand::MOLoad | MachineMemOperand::MOStore;
const ConstantInt *Vol = dyn_cast<ConstantInt>(CI.getOperand(4)); const ConstantInt *Vol = dyn_cast<ConstantInt>(CI.getOperand(4));
if (!Vol || !Vol->isZero()) if (!Vol || !Vol->isZero())
Info.flags |= MachineMemOperand::MOVolatile; Info.flags |= MachineMemOperand::MOVolatile;
return true; return true;
Show All 21 Lines Info.flags |= MachineMemOperand::MOLoad |
MachineMemOperand::MOStore | MachineMemOperand::MOStore |
MachineMemOperand::MOVolatile; MachineMemOperand::MOVolatile;
return true; return true;
} }
case Intrinsic::amdgcn_image_bvh_intersect_ray: { case Intrinsic::amdgcn_image_bvh_intersect_ray: {
Info.opc = ISD::INTRINSIC_W_CHAIN; Info.opc = ISD::INTRINSIC_W_CHAIN;
Info.memVT = MVT::getVT(CI.getType()); // XXX: what is correct VT? Info.memVT = MVT::getVT(CI.getType()); // XXX: what is correct VT?
Info.fallbackAddressSpace = AMDGPUAS::BUFFER_FAT_POINTER; Info.fallbackAddressSpace = AMDGPUAS::BUFFER_RESOURCE;
Info.align.reset(); Info.align.reset();
Info.flags |= MachineMemOperand::MOLoad | Info.flags |= MachineMemOperand::MOLoad |
MachineMemOperand::MODereferenceable; MachineMemOperand::MODereferenceable;
return true; return true;
} }
case Intrinsic::amdgcn_global_atomic_fadd: case Intrinsic::amdgcn_global_atomic_fadd:
case Intrinsic::amdgcn_global_atomic_fmin: case Intrinsic::amdgcn_global_atomic_fmin:
case Intrinsic::amdgcn_global_atomic_fmax: case Intrinsic::amdgcn_global_atomic_fmax:
▲ Show 20 Lines • Show All 175 Lines • ▼ Show 20 Linesbool SITargetLowering::isLegalAddressingMode(const DataLayout &DL,
if (AM.BaseGV) if (AM.BaseGV)
return false; return false;
if (AS == AMDGPUAS::GLOBAL_ADDRESS) if (AS == AMDGPUAS::GLOBAL_ADDRESS)
return isLegalGlobalAddressingMode(AM); return isLegalGlobalAddressingMode(AM);
if (AS == AMDGPUAS::CONSTANT_ADDRESS || if (AS == AMDGPUAS::CONSTANT_ADDRESS ||
AS == AMDGPUAS::CONSTANT_ADDRESS_32BIT || AS == AMDGPUAS::CONSTANT_ADDRESS_32BIT ||
AS == AMDGPUAS::BUFFER_FAT_POINTER) { AS == AMDGPUAS::BUFFER_FAT_POINTER || AS == AMDGPUAS::BUFFER_RESOURCE) {
// If the offset isn't a multiple of 4, it probably isn't going to be // If the offset isn't a multiple of 4, it probably isn't going to be
// correctly aligned. // correctly aligned.
// FIXME: Can we get the real alignment here? // FIXME: Can we get the real alignment here?
if (AM.BaseOffs % 4 != 0) if (AM.BaseOffs % 4 != 0)
return isLegalMUBUFAddressingMode(AM); return isLegalMUBUFAddressingMode(AM);
// There are no SMRD extloads, so if we have to do a small type access we // There are no SMRD extloads, so if we have to do a small type access we
// will use a MUBUF load. // will use a MUBUF load.
Show All 23 Lines if (AS == AMDGPUAS::CONSTANT_ADDRESS ||
if (AM.Scale == 1 && AM.HasBaseReg) if (AM.Scale == 1 && AM.HasBaseReg)
return true; return true;
return false; return false;
} else if (AS == AMDGPUAS::PRIVATE_ADDRESS) { } else if (AS == AMDGPUAS::PRIVATE_ADDRESS) {
return isLegalMUBUFAddressingMode(AM); return isLegalMUBUFAddressingMode(AM);
} else if (AS == AMDGPUAS::LOCAL_ADDRESS || } else if (AS == AMDGPUAS::LOCAL_ADDRESS || AS == AMDGPUAS::REGION_ADDRESS) {
AS == AMDGPUAS::REGION_ADDRESS) {
// Basic, single offset DS instructions allow a 16-bit unsigned immediate // Basic, single offset DS instructions allow a 16-bit unsigned immediate
// field. // field.
// XXX - If doing a 4-byte aligned 8-byte type access, we effectively have // XXX - If doing a 4-byte aligned 8-byte type access, we effectively have
// an 8-bit dword offset but we don't know the alignment here. // an 8-bit dword offset but we don't know the alignment here.
if (!isUInt<16>(AM.BaseOffs)) if (!isUInt<16>(AM.BaseOffs))
return false; return false;
if (AM.Scale == 0) // r + i or just i, depending on HasBaseReg. if (AM.Scale == 0) // r + i or just i, depending on HasBaseReg.
▲ Show 20 Lines • Show All 12,093 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/buffer-atomic-fadd.f32-no-rtn.ll

Show All 10 Linesdefine amdgpu_ps void @buffer_atomic_fadd_f32_offset_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX908_GFX11-NEXT: {{ $}} ; GFX908_GFX11-NEXT: {{ $}}
; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908_GFX11-NEXT: S_ENDPGM 0 ; GFX908_GFX11-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offset_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offset_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f32_offen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f32_offen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_offen_no_rtn ; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_offen_no_rtn
; GFX908_GFX11: bb.1 (%ir-block.0): ; GFX908_GFX11: bb.1 (%ir-block.0):
; GFX908_GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX908_GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX908_GFX11-NEXT: {{ $}} ; GFX908_GFX11-NEXT: {{ $}}
; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908_GFX11-NEXT: S_ENDPGM 0 ; GFX908_GFX11-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offen_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f32_idxen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f32_idxen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_idxen_no_rtn ; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_idxen_no_rtn
; GFX908_GFX11: bb.1 (%ir-block.0): ; GFX908_GFX11: bb.1 (%ir-block.0):
; GFX908_GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX908_GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX908_GFX11-NEXT: {{ $}} ; GFX908_GFX11-NEXT: {{ $}}
; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908_GFX11-NEXT: S_ENDPGM 0 ; GFX908_GFX11-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_idxen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_idxen_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f32_bothen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f32_bothen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_bothen_no_rtn ; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_bothen_no_rtn
; GFX908_GFX11: bb.1 (%ir-block.0): ; GFX908_GFX11: bb.1 (%ir-block.0):
; GFX908_GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2 ; GFX908_GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
; GFX908_GFX11-NEXT: {{ $}} ; GFX908_GFX11-NEXT: {{ $}}
; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX908_GFX11-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908_GFX11-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908_GFX11-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX908_GFX11-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908_GFX11-NEXT: S_ENDPGM 0 ; GFX908_GFX11-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_bothen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_bothen_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
ret void ret void
} }
declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg) declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg)
declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg) declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/buffer-atomic-fadd.f32-rtn.ll

Show All 9 Linesdefine amdgpu_ps float @buffer_atomic_fadd_f32_offset_rtn(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFSET_RTN [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFSET_RTN [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_offset_rtn ; GFX11-LABEL: name: buffer_atomic_fadd_f32_offset_rtn
; GFX11: bb.1 (%ir-block.0): ; GFX11: bb.1 (%ir-block.0):
; GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0 ; GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX11-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX11-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFSET_RTN [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFSET_RTN [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN]] ; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret float %ret ret float %ret
} }
define amdgpu_ps float @buffer_atomic_fadd_f32_offen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @buffer_atomic_fadd_f32_offen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offen_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_offen_rtn ; GFX11-LABEL: name: buffer_atomic_fadd_f32_offen_rtn
; GFX11: bb.1 (%ir-block.0): ; GFX11: bb.1 (%ir-block.0):
; GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX11-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX11-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN]] ; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %ret ret float %ret
} }
define amdgpu_ps float @buffer_atomic_fadd_f32_idxen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps float @buffer_atomic_fadd_f32_idxen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_idxen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_idxen_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_IDXEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_IDXEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_idxen_rtn ; GFX11-LABEL: name: buffer_atomic_fadd_f32_idxen_rtn
; GFX11: bb.1 (%ir-block.0): ; GFX11: bb.1 (%ir-block.0):
; GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX11-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX11-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_IDXEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_IDXEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN]] ; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret float %ret ret float %ret
} }
define amdgpu_ps float @buffer_atomic_fadd_f32_bothen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @buffer_atomic_fadd_f32_bothen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_bothen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_bothen_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_bothen_rtn ; GFX11-LABEL: name: buffer_atomic_fadd_f32_bothen_rtn
; GFX11: bb.1 (%ir-block.0): ; GFX11: bb.1 (%ir-block.0):
; GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2 ; GFX11-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX11-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX11-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX11-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX11-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX11-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX11-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX11-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN]] ; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret float %ret ret float %ret
} }
declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg) declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg)
declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg) declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/buffer-atomic-fadd.f64.ll

Show All 10 Linesdefine amdgpu_ps void @buffer_atomic_fadd_f64_offset_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_OFFSET [[REG_SEQUENCE]], [[REG_SEQUENCE1]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_OFFSET [[REG_SEQUENCE]], [[REG_SEQUENCE1]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f64_offen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f64_offen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_offen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_offen_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_OFFEN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_OFFEN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f64_idxen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f64_idxen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_idxen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_idxen_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_IDXEN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_IDXEN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f64_bothen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f64_bothen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_bothen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_bothen_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_BOTHEN [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_BOTHEN [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2) %ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
ret void ret void
} }
define amdgpu_ps double @buffer_atomic_fadd_f64_offset_rtn(double %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps double @buffer_atomic_fadd_f64_offset_rtn(double %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_offset_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_offset_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_OFFSET_RTN [[REG_SEQUENCE]], [[REG_SEQUENCE1]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_OFFSET_RTN [[REG_SEQUENCE]], [[REG_SEQUENCE1]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN]].sub0 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN]].sub0
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN]].sub1 ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN]].sub1
; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY7]], implicit $exec ; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY7]], implicit $exec
; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[V_READFIRSTLANE_B32_]] ; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[V_READFIRSTLANE_B32_]]
; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY8]], implicit $exec ; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY8]], implicit $exec
; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[V_READFIRSTLANE_B32_1]] ; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[V_READFIRSTLANE_B32_1]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $sgpr0, implicit $sgpr1 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $sgpr0, implicit $sgpr1
%ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
Show All 10 Linesdefine amdgpu_ps double @buffer_atomic_fadd_f64_offen_rtn(double %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_OFFEN_RTN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_OFFEN_RTN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN]].sub0 ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN]].sub0
; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN]].sub1 ; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN]].sub1
; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY8]], implicit $exec ; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY8]], implicit $exec
; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[V_READFIRSTLANE_B32_]] ; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[V_READFIRSTLANE_B32_]]
; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY9]], implicit $exec ; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY9]], implicit $exec
; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[V_READFIRSTLANE_B32_1]] ; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[V_READFIRSTLANE_B32_1]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $sgpr0, implicit $sgpr1 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $sgpr0, implicit $sgpr1
%ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
Show All 10 Linesdefine amdgpu_ps double @buffer_atomic_fadd_f64_idxen_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_IDXEN_RTN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_IDXEN_RTN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN]].sub0 ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN]].sub0
; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN]].sub1 ; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN]].sub1
; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY8]], implicit $exec ; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY8]], implicit $exec
; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[V_READFIRSTLANE_B32_]] ; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[V_READFIRSTLANE_B32_]]
; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY9]], implicit $exec ; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY9]], implicit $exec
; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[V_READFIRSTLANE_B32_1]] ; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[V_READFIRSTLANE_B32_1]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $sgpr0, implicit $sgpr1 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $sgpr0, implicit $sgpr1
%ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
Show All 12 Linesdefine amdgpu_ps double @buffer_atomic_fadd_f64_bothen_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN]].sub0 ; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN]].sub0
; GFX90A_GFX940-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN]].sub1 ; GFX90A_GFX940-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN]].sub1
; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY9]], implicit $exec ; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY9]], implicit $exec
; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[V_READFIRSTLANE_B32_]] ; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[V_READFIRSTLANE_B32_]]
; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY10]], implicit $exec ; GFX90A_GFX940-NEXT: [[V_READFIRSTLANE_B32_1:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY10]], implicit $exec
; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[V_READFIRSTLANE_B32_1]] ; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[V_READFIRSTLANE_B32_1]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $sgpr0, implicit $sgpr1 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $sgpr0, implicit $sgpr1
%ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret double %ret ret double %ret
} }
declare double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double, <4 x i32>, i32, i32, i32 immarg) declare double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double, <4 x i32>, i32, i32, i32 immarg)
declare double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double, <4 x i32>, i32, i32, i32, i32 immarg) declare double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/buffer-atomic-fadd.v2f16-no-rtn.ll

Show All 9 Linesdefine amdgpu_ps void @buffer_atomic_fadd_v2f16_offset_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offset_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offset_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_v2f16_offen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_v2f16_offen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_offen_no_rtn ; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_offen_no_rtn
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX908-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offen_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_v2f16_idxen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_v2f16_idxen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_idxen_no_rtn ; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_idxen_no_rtn
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX908-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_idxen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_idxen_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_v2f16_bothen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_v2f16_bothen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_bothen_no_rtn ; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_bothen_no_rtn
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2 ; GFX908-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX908-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_bothen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_bothen_no_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2) %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
ret void ret void
} }
declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg) declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg)
declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg) declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/buffer-atomic-fadd.v2f16-rtn.ll

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -global-isel -march=amdgcn -mcpu=gfx90a -verify-machineinstrs -stop-after=instruction-select < %s | FileCheck -check-prefix=GFX90A_GFX940 %s ; RUN: llc -global-isel -march=amdgcn -mcpu=gfx90a -verify-machineinstrs -stop-after=instruction-select < %s | FileCheck -check-prefix=GFX90A_GFX940 %s
; RUN: llc -global-isel -march=amdgcn -mcpu=gfx940 -verify-machineinstrs -stop-after=instruction-select < %s | FileCheck -check-prefix=GFX90A_GFX940 %s ; RUN: llc -global-isel -march=amdgcn -mcpu=gfx940 -verify-machineinstrs -stop-after=instruction-select < %s | FileCheck -check-prefix=GFX90A_GFX940 %s
define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_offset_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_offset_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offset_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offset_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 1, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 1, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret <2 x half> %ret ret <2 x half> %ret
} }
define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_offen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_offen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offen_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret <2 x half> %ret ret <2 x half> %ret
} }
define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_idxen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_idxen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_idxen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_idxen_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret <2 x half> %ret ret <2 x half> %ret
} }
define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_bothen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_bothen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_bothen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_bothen_rtn
; GFX90A_GFX940: bb.1 (%ir-block.0): ; GFX90A_GFX940: bb.1 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2 ; GFX90A_GFX940-NEXT: liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4, $vgpr0, $vgpr1, $vgpr2
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 3, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 3, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2) %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
ret <2 x half> %ret ret <2 x half> %ret
} }
declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg) declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg)
declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg) declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/irtranslator-non-integral-address-spaces.ll

  • This file was deleted.
; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -global-isel -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 -o - -stop-after=irtranslator %s | FileCheck %s
; Check that the CSEMIRBuilder doesn't fold away the getelementptr during IRTranslator
define ptr addrspace(7) @no_auto_constfold_gep() {
; CHECK-LABEL: name: no_auto_constfold_gep
; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: [[C:%[0-9]+]]:_(p7) = G_CONSTANT i64 0
; CHECK-NEXT: [[C1:%[0-9]+]]:_(s64) = G_CONSTANT i64 123
; CHECK-NEXT: [[PTR_ADD:%[0-9]+]]:_(p7) = G_PTR_ADD [[C]], [[C1]](s64)
; CHECK-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[PTR_ADD]](p7)
; CHECK-NEXT: $vgpr0 = COPY [[UV]](s32)
; CHECK-NEXT: $vgpr1 = COPY [[UV1]](s32)
; CHECK-NEXT: SI_RETURN implicit $vgpr0, implicit $vgpr1
%gep = getelementptr i8, ptr addrspace(7) null, i64 123
ret ptr addrspace(7) %gep
}
; Check that the CSEMIRBuilder doesn't fold away the getelementptr during IRTranslator
define <2 x ptr addrspace(7)> @no_auto_constfold_gep_vector() {
; CHECK-LABEL: name: no_auto_constfold_gep_vector
; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: [[C:%[0-9]+]]:_(p7) = G_CONSTANT i64 0
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x p7>) = G_BUILD_VECTOR [[C]](p7), [[C]](p7)
; CHECK-NEXT: [[C1:%[0-9]+]]:_(s64) = G_CONSTANT i64 123
; CHECK-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[C1]](s64), [[C1]](s64)
; CHECK-NEXT: [[PTR_ADD:%[0-9]+]]:_(<2 x p7>) = G_PTR_ADD [[BUILD_VECTOR]], [[BUILD_VECTOR1]](<2 x s64>)
; CHECK-NEXT: [[COPY:%[0-9]+]]:_(<2 x p7>) = COPY [[PTR_ADD]](<2 x p7>)
; CHECK-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x p7>)
; CHECK-NEXT: $vgpr0 = COPY [[UV]](s32)
; CHECK-NEXT: $vgpr1 = COPY [[UV1]](s32)
; CHECK-NEXT: $vgpr2 = COPY [[UV2]](s32)
; CHECK-NEXT: $vgpr3 = COPY [[UV3]](s32)
; CHECK-NEXT: SI_RETURN implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
%gep = getelementptr i8, <2 x ptr addrspace(7)> zeroinitializer, <2 x i64> <i64 123, i64 123>
ret <2 x ptr addrspace(7)> %gep
}

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.atomic.dim.a16.ll

Show All 14 Linesdefine amdgpu_ps float @atomic_swap_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.swap.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.swap.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_swap_1d ; GFX10NSA-LABEL: name: atomic_swap_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.swap.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.swap.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.swap.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.swap.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_add_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_add_1d ; GFX10NSA-LABEL: name: atomic_add_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.add.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.add.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_sub_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.sub.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.sub.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_sub_1d ; GFX10NSA-LABEL: name: atomic_sub_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.sub.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.sub.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.sub.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.sub.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_smin_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.smin.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.smin.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_smin_1d ; GFX10NSA-LABEL: name: atomic_smin_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.smin.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.smin.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.smin.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.smin.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 10 Linesdefine amdgpu_ps float @atomic_umin_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.umin.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.umin.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_umin_1d ; GFX10NSA-LABEL: name: atomic_umin_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.umin.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.umin.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.umin.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.umin.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_smax_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.smax.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.smax.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_smax_1d ; GFX10NSA-LABEL: name: atomic_smax_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.smax.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.smax.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.smax.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.smax.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_umax_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.umax.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.umax.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_umax_1d ; GFX10NSA-LABEL: name: atomic_umax_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.umax.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.umax.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.umax.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.umax.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_and_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.and.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.and.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_and_1d ; GFX10NSA-LABEL: name: atomic_and_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.and.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.and.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.and.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.and.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_or_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.or.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.or.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_or_1d ; GFX10NSA-LABEL: name: atomic_or_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.or.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.or.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.or.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.or.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_xor_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.xor.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.xor.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_xor_1d ; GFX10NSA-LABEL: name: atomic_xor_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.xor.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.xor.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.xor.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.xor.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_inc_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.inc.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.inc.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_inc_1d ; GFX10NSA-LABEL: name: atomic_inc_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.inc.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.inc.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.inc.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.inc.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_dec_1d(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.dec.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.dec.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_dec_1d ; GFX10NSA-LABEL: name: atomic_dec_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.dec.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.dec.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.dec.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.dec.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 11 Linesdefine amdgpu_ps float @atomic_cmpswap_1d(<8 x i32> inreg %rsrc, i32 %cmp, i32 %swap, i16 %s) {
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_cmpswap_1d ; GFX10NSA-LABEL: name: atomic_cmpswap_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.cmpswap.1d.i32.i16(i32 %cmp, i32 %swap, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.cmpswap.1d.i32.i16(i32 %cmp, i32 %swap, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 12 Linesdefine amdgpu_ps float @atomic_add_2d(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t) {
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2d), [[COPY8]](s32), [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2d), [[COPY8]](s32), [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_add_2d ; GFX10NSA-LABEL: name: atomic_add_2d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2d), [[COPY8]](s32), [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2d), [[COPY8]](s32), [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.add.2d.i32.i16(i32 %data, i16 %s, i16 %t, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.add.2d.i32.i16(i32 %data, i16 %s, i16 %t, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 17 Linesdefine amdgpu_ps float @atomic_add_3d(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t, i16 %r) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.3d), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.3d), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_add_3d ; GFX10NSA-LABEL: name: atomic_add_3d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
Show All 10 Linesdefine amdgpu_ps float @atomic_add_3d(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t, i16 %r) {
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.3d), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.3d), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.add.3d.i32.i16(i32 %data, i16 %s, i16 %t, i16 %r, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.add.3d.i32.i16(i32 %data, i16 %s, i16 %t, i16 %r, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 17 Linesdefine amdgpu_ps float @atomic_add_cube(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t, i16 %face) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.cube), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.cube), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_add_cube ; GFX10NSA-LABEL: name: atomic_add_cube
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
Show All 10 Linesdefine amdgpu_ps float @atomic_add_cube(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t, i16 %face) {
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.cube), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.cube), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.add.cube.i32.i16(i32 %data, i16 %s, i16 %t, i16 %face, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.add.cube.i32.i16(i32 %data, i16 %s, i16 %t, i16 %face, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 12 Linesdefine amdgpu_ps float @atomic_add_1darray(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %slice) {
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1darray), [[COPY8]](s32), [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1darray), [[COPY8]](s32), [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_add_1darray ; GFX10NSA-LABEL: name: atomic_add_1darray
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1darray), [[COPY8]](s32), [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1darray), [[COPY8]](s32), [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.add.1darray.i32.i16(i32 %data, i16 %s, i16 %slice, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.add.1darray.i32.i16(i32 %data, i16 %s, i16 %slice, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 17 Linesdefine amdgpu_ps float @atomic_add_2darray(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t, i16 %slice) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2darray), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2darray), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_add_2darray ; GFX10NSA-LABEL: name: atomic_add_2darray
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
Show All 10 Linesdefine amdgpu_ps float @atomic_add_2darray(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t, i16 %slice) {
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2darray), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2darray), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.add.2darray.i32.i16(i32 %data, i16 %s, i16 %t, i16 %slice, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.add.2darray.i32.i16(i32 %data, i16 %s, i16 %t, i16 %slice, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 17 Linesdefine amdgpu_ps float @atomic_add_2dmsaa(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t, i16 %fragid) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2dmsaa), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2dmsaa), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_add_2dmsaa ; GFX10NSA-LABEL: name: atomic_add_2dmsaa
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
Show All 10 Linesdefine amdgpu_ps float @atomic_add_2dmsaa(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t, i16 %fragid) {
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2dmsaa), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2dmsaa), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.add.2dmsaa.i32.i16(i32 %data, i16 %s, i16 %t, i16 %fragid, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.add.2dmsaa.i32.i16(i32 %data, i16 %s, i16 %t, i16 %fragid, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 18 Linesdefine amdgpu_ps float @atomic_add_2darraymsaa(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t, i16 %slice, i16 %fragid) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2darraymsaa), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2darraymsaa), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_add_2darraymsaa ; GFX10NSA-LABEL: name: atomic_add_2darraymsaa
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
Show All 11 Linesdefine amdgpu_ps float @atomic_add_2darraymsaa(<8 x i32> inreg %rsrc, i32 %data, i16 %s, i16 %t, i16 %slice, i16 %fragid) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2darraymsaa), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.2darraymsaa), [[COPY8]](s32), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.add.2darraymsaa.i32.i16(i32 %data, i16 %s, i16 %t, i16 %slice, i16 %fragid, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.add.2darraymsaa.i32.i16(i32 %data, i16 %s, i16 %t, i16 %slice, i16 %fragid, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 9 Linesdefine amdgpu_ps float @atomic_add_1d_slc(<8 x i32> inreg %rsrc, i32 %data, i16 %s) {
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_add_1d_slc ; GFX10NSA-LABEL: name: atomic_add_1d_slc
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY9]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.add.1d), [[COPY8]](s32), [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.add.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 2) %v = call i32 @llvm.amdgcn.image.atomic.add.1d.i32.i16(i32 %data, i16 %s, <8 x i32> %rsrc, i32 0, i32 2)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 14 Linesdefine amdgpu_ps float @atomic_cmpswap_2d(<8 x i32> inreg %rsrc, i32 %cmp, i32 %swap, i16 %s, i16 %t) {
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.2d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.2d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_cmpswap_2d ; GFX10NSA-LABEL: name: atomic_cmpswap_2d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.2d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.2d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.cmpswap.2d.i32.i16(i32 %cmp, i32 %swap, i16 %s, i16 %t, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.cmpswap.2d.i32.i16(i32 %cmp, i32 %swap, i16 %s, i16 %t, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 19 Linesdefine amdgpu_ps float @atomic_cmpswap_3d(<8 x i32> inreg %rsrc, i32 %cmp, i32 %swap, i16 %s, i16 %t, i16 %r) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.3d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.3d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_cmpswap_3d ; GFX10NSA-LABEL: name: atomic_cmpswap_3d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
Show All 12 Linesdefine amdgpu_ps float @atomic_cmpswap_3d(<8 x i32> inreg %rsrc, i32 %cmp, i32 %swap, i16 %s, i16 %t, i16 %r) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY11]](s32)
; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.3d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.3d), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.cmpswap.3d.i32.i16(i32 %cmp, i32 %swap, i16 %s, i16 %t, i16 %r, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.cmpswap.3d.i32.i16(i32 %cmp, i32 %swap, i16 %s, i16 %t, i16 %r, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 20 Linesdefine amdgpu_ps float @atomic_cmpswap_2darraymsaa(<8 x i32> inreg %rsrc, i32 %cmp, i32 %swap, i16 %s, i16 %t, i16 %slice, i16 %fragid) {
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.2darraymsaa), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.2darraymsaa), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: atomic_cmpswap_2darraymsaa ; GFX10NSA-LABEL: name: atomic_cmpswap_2darraymsaa
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
Show All 13 Linesdefine amdgpu_ps float @atomic_cmpswap_2darraymsaa(<8 x i32> inreg %rsrc, i32 %cmp, i32 %swap, i16 %s, i16 %t, i16 %slice, i16 %fragid) {
; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10NSA-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10NSA-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.2darraymsaa), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.2darraymsaa), [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call i32 @llvm.amdgcn.image.atomic.cmpswap.2darraymsaa.i32.i16(i32 %cmp, i32 %swap, i16 %s, i16 %t, i16 %slice, i16 %fragid, <8 x i32> %rsrc, i32 0, i32 0) %v = call i32 @llvm.amdgcn.image.atomic.cmpswap.2darraymsaa.i32.i16(i32 %cmp, i32 %swap, i16 %s, i16 %t, i16 %slice, i16 %fragid, <8 x i32> %rsrc, i32 0, i32 0)
%out = bitcast i32 %v to float %out = bitcast i32 %v to float
ret float %out ret float %out
} }
Show All 29 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.dim.a16.ll

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show All 16 Linesdefine amdgpu_ps <4 x float> @load_1d(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_1d ; GFX10NSA-LABEL: name: load_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
Show All 20 Linesdefine amdgpu_ps <4 x float> @load_2d(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_2d ; GFX10NSA-LABEL: name: load_2d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @load_2d(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
Show All 28 Linesdefine amdgpu_ps <4 x float> @load_3d(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_3d ; GFX10NSA-LABEL: name: load_3d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @load_3d(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
Show All 29 Linesdefine amdgpu_ps <4 x float> @load_cube(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_cube ; GFX10NSA-LABEL: name: load_cube
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @load_cube(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
Show All 22 Linesdefine amdgpu_ps <4 x float> @load_1darray(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1darray), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1darray), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_1darray ; GFX10NSA-LABEL: name: load_1darray
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @load_1darray(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1darray), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1darray), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
Show All 28 Linesdefine amdgpu_ps <4 x float> @load_2darray(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_2darray ; GFX10NSA-LABEL: name: load_2darray
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @load_2darray(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
Show All 29 Linesdefine amdgpu_ps <4 x float> @load_2dmsaa(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2dmsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2dmsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_2dmsaa ; GFX10NSA-LABEL: name: load_2dmsaa
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @load_2dmsaa(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2dmsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2dmsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
Show All 31 Linesdefine amdgpu_ps <4 x float> @load_2darraymsaa(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_2darraymsaa ; GFX10NSA-LABEL: name: load_2darraymsaa
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @load_2darraymsaa(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
Show All 23 Linesdefine amdgpu_ps <4 x float> @load_mip_1d(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.1d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.1d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_mip_1d ; GFX10NSA-LABEL: name: load_mip_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @load_mip_1d(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.1d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.1d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
Show All 28 Linesdefine amdgpu_ps <4 x float> @load_mip_2d(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_mip_2d ; GFX10NSA-LABEL: name: load_mip_2d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @load_mip_2d(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
Show All 31 Linesdefine amdgpu_ps <4 x float> @load_mip_3d(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_mip_3d ; GFX10NSA-LABEL: name: load_mip_3d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @load_mip_3d(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
Show All 32 Linesdefine amdgpu_ps <4 x float> @load_mip_cube(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_mip_cube ; GFX10NSA-LABEL: name: load_mip_cube
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @load_mip_cube(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
Show All 30 Linesdefine amdgpu_ps <4 x float> @load_mip_1darray(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.1darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.1darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_mip_1darray ; GFX10NSA-LABEL: name: load_mip_1darray
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @load_mip_1darray(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.1darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.1darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
Show All 31 Linesdefine amdgpu_ps <4 x float> @load_mip_2darray(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_mip_2darray ; GFX10NSA-LABEL: name: load_mip_2darray
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @load_mip_2darray(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.mip.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
Show All 23 Linesdefine amdgpu_ps void @store_1d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_1d ; GFX10NSA-LABEL: name: store_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4 ; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
call void @llvm.amdgcn.image.store.1d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.1d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
define amdgpu_ps void @store_2d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) { define amdgpu_ps void @store_2d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
Show All 20 Linesdefine amdgpu_ps void @store_2d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_2d ; GFX10NSA-LABEL: name: store_2d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 13 Linesdefine amdgpu_ps void @store_2d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
%t = extractelement <2 x i16> %coords, i32 1 %t = extractelement <2 x i16> %coords, i32 1
call void @llvm.amdgcn.image.store.2d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.2d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
Show All 28 Linesdefine amdgpu_ps void @store_3d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.3d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.3d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_3d ; GFX10NSA-LABEL: name: store_3d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 20 Linesdefine amdgpu_ps void @store_3d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.3d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.3d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
%t = extractelement <2 x i16> %coords_lo, i32 1 %t = extractelement <2 x i16> %coords_lo, i32 1
%r = extractelement <2 x i16> %coords_hi, i32 0 %r = extractelement <2 x i16> %coords_hi, i32 0
call void @llvm.amdgcn.image.store.3d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %r, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.3d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %r, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
Show All 29 Linesdefine amdgpu_ps void @store_cube(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.cube), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.cube), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_cube ; GFX10NSA-LABEL: name: store_cube
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 20 Linesdefine amdgpu_ps void @store_cube(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.cube), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.cube), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
%t = extractelement <2 x i16> %coords_lo, i32 1 %t = extractelement <2 x i16> %coords_lo, i32 1
%slice = extractelement <2 x i16> %coords_hi, i32 0 %slice = extractelement <2 x i16> %coords_hi, i32 0
call void @llvm.amdgcn.image.store.cube.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %slice, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.cube.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %slice, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
Show All 22 Linesdefine amdgpu_ps void @store_1darray(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_1darray ; GFX10NSA-LABEL: name: store_1darray
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 13 Linesdefine amdgpu_ps void @store_1darray(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
%slice = extractelement <2 x i16> %coords, i32 1 %slice = extractelement <2 x i16> %coords, i32 1
call void @llvm.amdgcn.image.store.1darray.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %slice, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.1darray.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %slice, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
Show All 28 Linesdefine amdgpu_ps void @store_2darray(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_2darray ; GFX10NSA-LABEL: name: store_2darray
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 20 Linesdefine amdgpu_ps void @store_2darray(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
%t = extractelement <2 x i16> %coords_lo, i32 1 %t = extractelement <2 x i16> %coords_lo, i32 1
%slice = extractelement <2 x i16> %coords_hi, i32 0 %slice = extractelement <2 x i16> %coords_hi, i32 0
call void @llvm.amdgcn.image.store.2darray.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %slice, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.2darray.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %slice, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
Show All 29 Linesdefine amdgpu_ps void @store_2dmsaa(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2dmsaa), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2dmsaa), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_2dmsaa ; GFX10NSA-LABEL: name: store_2dmsaa
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 20 Linesdefine amdgpu_ps void @store_2dmsaa(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2dmsaa), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2dmsaa), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
%t = extractelement <2 x i16> %coords_lo, i32 1 %t = extractelement <2 x i16> %coords_lo, i32 1
%fragid = extractelement <2 x i16> %coords_hi, i32 0 %fragid = extractelement <2 x i16> %coords_hi, i32 0
call void @llvm.amdgcn.image.store.2dmsaa.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %fragid, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.2dmsaa.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %fragid, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
Show All 31 Linesdefine amdgpu_ps void @store_2darraymsaa(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2darraymsaa), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2darraymsaa), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_2darraymsaa ; GFX10NSA-LABEL: name: store_2darraymsaa
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 22 Linesdefine amdgpu_ps void @store_2darraymsaa(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2darraymsaa), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.2darraymsaa), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
%t = extractelement <2 x i16> %coords_lo, i32 1 %t = extractelement <2 x i16> %coords_lo, i32 1
%slice = extractelement <2 x i16> %coords_hi, i32 0 %slice = extractelement <2 x i16> %coords_hi, i32 0
%fragid = extractelement <2 x i16> %coords_hi, i32 1 %fragid = extractelement <2 x i16> %coords_hi, i32 1
call void @llvm.amdgcn.image.store.2darraymsaa.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %slice, i16 %fragid, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.2darraymsaa.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %slice, i16 %fragid, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
Show All 23 Linesdefine amdgpu_ps void @store_mip_1d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_mip_1d ; GFX10NSA-LABEL: name: store_mip_1d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 13 Linesdefine amdgpu_ps void @store_mip_1d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
%mip = extractelement <2 x i16> %coords, i32 1 %mip = extractelement <2 x i16> %coords, i32 1
call void @llvm.amdgcn.image.store.mip.1d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.mip.1d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
Show All 28 Linesdefine amdgpu_ps void @store_mip_2d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.2d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.2d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_mip_2d ; GFX10NSA-LABEL: name: store_mip_2d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 20 Linesdefine amdgpu_ps void @store_mip_2d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.2d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.2d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
%t = extractelement <2 x i16> %coords_lo, i32 1 %t = extractelement <2 x i16> %coords_lo, i32 1
%mip = extractelement <2 x i16> %coords_hi, i32 0 %mip = extractelement <2 x i16> %coords_hi, i32 0
call void @llvm.amdgcn.image.store.mip.2d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.mip.2d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
Show All 31 Linesdefine amdgpu_ps void @store_mip_3d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.3d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.3d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_mip_3d ; GFX10NSA-LABEL: name: store_mip_3d
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 22 Linesdefine amdgpu_ps void @store_mip_3d(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.3d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.3d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
%t = extractelement <2 x i16> %coords_lo, i32 1 %t = extractelement <2 x i16> %coords_lo, i32 1
%r = extractelement <2 x i16> %coords_hi, i32 0 %r = extractelement <2 x i16> %coords_hi, i32 0
%mip = extractelement <2 x i16> %coords_hi, i32 1 %mip = extractelement <2 x i16> %coords_hi, i32 1
call void @llvm.amdgcn.image.store.mip.3d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %r, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.mip.3d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %r, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
Show All 32 Linesdefine amdgpu_ps void @store_mip_cube(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.cube), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.cube), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_mip_cube ; GFX10NSA-LABEL: name: store_mip_cube
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 22 Linesdefine amdgpu_ps void @store_mip_cube(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.cube), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.cube), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
%t = extractelement <2 x i16> %coords_lo, i32 1 %t = extractelement <2 x i16> %coords_lo, i32 1
%slice = extractelement <2 x i16> %coords_hi, i32 0 %slice = extractelement <2 x i16> %coords_hi, i32 0
%mip = extractelement <2 x i16> %coords_hi, i32 1 %mip = extractelement <2 x i16> %coords_hi, i32 1
call void @llvm.amdgcn.image.store.mip.cube.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %slice, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.mip.cube.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %slice, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
Show All 30 Linesdefine amdgpu_ps void @store_mip_1darray(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.1darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.1darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_mip_1darray ; GFX10NSA-LABEL: name: store_mip_1darray
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 20 Linesdefine amdgpu_ps void @store_mip_1darray(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.1darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.1darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
%slice = extractelement <2 x i16> %coords_lo, i32 1 %slice = extractelement <2 x i16> %coords_lo, i32 1
%mip = extractelement <2 x i16> %coords_hi, i32 0 %mip = extractelement <2 x i16> %coords_hi, i32 0
call void @llvm.amdgcn.image.store.mip.1darray.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %slice, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.mip.1darray.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %slice, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
Show All 31 Linesdefine amdgpu_ps void @store_mip_2darray(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.2darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.2darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_mip_2darray ; GFX10NSA-LABEL: name: store_mip_2darray
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 22 Linesdefine amdgpu_ps void @store_mip_2darray(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY13]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.2darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.mip.2darray), [[BUILD_VECTOR1]](<4 x s32>), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords_lo, i32 0 %s = extractelement <2 x i16> %coords_lo, i32 0
%t = extractelement <2 x i16> %coords_lo, i32 1 %t = extractelement <2 x i16> %coords_lo, i32 1
%slice = extractelement <2 x i16> %coords_hi, i32 0 %slice = extractelement <2 x i16> %coords_hi, i32 0
%mip = extractelement <2 x i16> %coords_hi, i32 1 %mip = extractelement <2 x i16> %coords_hi, i32 1
call void @llvm.amdgcn.image.store.mip.2darray.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %slice, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.mip.2darray.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, i16 %t, i16 %slice, i16 %mip, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
▲ Show 20 Lines • Show All 469 Lines • ▼ Show 20 Linesdefine amdgpu_ps float @load_1d_V1(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 8, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 8, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: load_1d_V1 ; GFX10NSA-LABEL: name: load_1d_V1
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 8, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 8, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
%v = call float @llvm.amdgcn.image.load.1d.f32.i16(i32 8, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call float @llvm.amdgcn.image.load.1d.f32.i16(i32 8, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
ret float %v ret float %v
} }
Show All 11 Linesdefine amdgpu_ps <2 x float> @load_1d_V2(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 9, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 9, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
; GFX10NSA-LABEL: name: load_1d_V2 ; GFX10NSA-LABEL: name: load_1d_V2
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 9, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 9, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
%v = call <2 x float> @llvm.amdgcn.image.load.1d.v2f32.i16(i32 9, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call <2 x float> @llvm.amdgcn.image.load.1d.v2f32.i16(i32 9, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
ret <2 x float> %v ret <2 x float> %v
Show All 14 Linesdefine amdgpu_ps void @store_1d_V1(<8 x i32> inreg %rsrc, float %vdata, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[COPY8]](s32), 2, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (s32), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[COPY8]](s32), 2, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (s32), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_1d_V1 ; GFX10NSA-LABEL: name: store_1d_V1
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[COPY8]](s32), 2, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (s32), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[COPY8]](s32), 2, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (s32), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
call void @llvm.amdgcn.image.store.1d.f32.i16(float %vdata, i32 2, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.1d.f32.i16(float %vdata, i32 2, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
define amdgpu_ps void @store_1d_V2(<8 x i32> inreg %rsrc, <2 x float> %vdata, <2 x i16> %coords) { define amdgpu_ps void @store_1d_V2(<8 x i32> inreg %rsrc, <2 x float> %vdata, <2 x i16> %coords) {
Show All 13 Linesdefine amdgpu_ps void @store_1d_V2(<8 x i32> inreg %rsrc, <2 x float> %vdata, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<2 x s32>), 12, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<2 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<2 x s32>), 12, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 3 :: (dereferenceable store (<2 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_1d_V2 ; GFX10NSA-LABEL: name: store_1d_V2
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<2 x s32>), 12, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<2 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<2 x s32>), 12, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 1 :: (dereferenceable store (<2 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
call void @llvm.amdgcn.image.store.1d.v2f32.i16(<2 x float> %vdata, i32 12, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.1d.v2f32.i16(<2 x float> %vdata, i32 12, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
define amdgpu_ps <4 x float> @load_1d_glc(<8 x i32> inreg %rsrc, <2 x i16> %coords) { define amdgpu_ps <4 x float> @load_1d_glc(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
Show All 10 Linesdefine amdgpu_ps <4 x float> @load_1d_glc(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 1, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 1, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_1d_glc ; GFX10NSA-LABEL: name: load_1d_glc
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 1, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 1, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
Show All 15 Linesdefine amdgpu_ps <4 x float> @load_1d_slc(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_1d_slc ; GFX10NSA-LABEL: name: load_1d_slc
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
Show All 15 Linesdefine amdgpu_ps <4 x float> @load_1d_glc_slc(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 3, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 3, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_1d_glc_slc ; GFX10NSA-LABEL: name: load_1d_glc_slc
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 3, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 3, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
Show All 20 Linesdefine amdgpu_ps void @store_1d_glc(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 1, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 1, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_1d_glc ; GFX10NSA-LABEL: name: store_1d_glc
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4 ; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 1, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 1, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
call void @llvm.amdgcn.image.store.1d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, <8 x i32> %rsrc, i32 0, i32 1) call void @llvm.amdgcn.image.store.1d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, <8 x i32> %rsrc, i32 0, i32 1)
ret void ret void
} }
define amdgpu_ps void @store_1d_slc(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) { define amdgpu_ps void @store_1d_slc(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
Show All 15 Linesdefine amdgpu_ps void @store_1d_slc(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_1d_slc ; GFX10NSA-LABEL: name: store_1d_slc
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4 ; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 2, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
call void @llvm.amdgcn.image.store.1d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, <8 x i32> %rsrc, i32 0, i32 2) call void @llvm.amdgcn.image.store.1d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, <8 x i32> %rsrc, i32 0, i32 2)
ret void ret void
} }
define amdgpu_ps void @store_1d_glc_slc(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) { define amdgpu_ps void @store_1d_glc_slc(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
Show All 15 Linesdefine amdgpu_ps void @store_1d_glc_slc(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 3, 3 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 3, 3 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10NSA-LABEL: name: store_1d_glc_slc ; GFX10NSA-LABEL: name: store_1d_glc_slc
; GFX10NSA: bb.1.main_body: ; GFX10NSA: bb.1.main_body:
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4 ; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr4
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY12]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 3, 1 :: (dereferenceable store (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE intrinsic(@llvm.amdgcn.image.store.1d), [[BUILD_VECTOR1]](<4 x s32>), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 0, 3, 1 :: (dereferenceable store (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: S_ENDPGM 0 ; GFX10NSA-NEXT: S_ENDPGM 0
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
call void @llvm.amdgcn.image.store.1d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, <8 x i32> %rsrc, i32 0, i32 3) call void @llvm.amdgcn.image.store.1d.v4f32.i16(<4 x float> %vdata, i32 15, i16 %s, <8 x i32> %rsrc, i32 0, i32 3)
ret void ret void
} }
define amdgpu_ps <4 x float> @getresinfo_dmask0(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) { define amdgpu_ps <4 x float> @getresinfo_dmask0(<8 x i32> inreg %rsrc, <4 x float> %vdata, <2 x i16> %coords) {
Show All 40 Linesdefine amdgpu_ps <4 x float> @load_1d_tfe(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX9-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 1, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 1, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GFX9-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GFX9-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_1d_tfe ; GFX10NSA-LABEL: name: load_1d_tfe
Show All 10 Linesdefine amdgpu_ps <4 x float> @load_1d_tfe(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX10NSA-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 1, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), 1, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GFX10NSA-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GFX10NSA-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
Show All 25 Linesdefine amdgpu_ps <4 x float> @load_2d_tfe(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX9-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX9-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX9-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GFX9-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GFX9-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_2d_tfe ; GFX10NSA-LABEL: name: load_2d_tfe
Show All 15 Linesdefine amdgpu_ps <4 x float> @load_2d_tfe(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10NSA-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY8]](<2 x s16>)
; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; GFX10NSA-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GFX10NSA-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GFX10NSA-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
Show All 33 Linesdefine amdgpu_ps <4 x float> @load_3d_tfe(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF1:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF1:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF1]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GFX9-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GFX9-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_3d_tfe ; GFX10NSA-LABEL: name: load_3d_tfe
Show All 22 Linesdefine amdgpu_ps <4 x float> @load_3d_tfe(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32) ; GFX10NSA-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[DEF1:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF1:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF1]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GFX10NSA-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GFX10NSA-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
Show All 36 Linesdefine amdgpu_ps <4 x float> @load_2darraymsaa_tfe(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX9-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX9-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX9-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GFX9-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GFX9-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_2darraymsaa_tfe ; GFX10NSA-LABEL: name: load_2darraymsaa_tfe
Show All 24 Linesdefine amdgpu_ps <4 x float> @load_2darraymsaa_tfe(<8 x i32> inreg %rsrc, <2 x i16> %coords_lo, <2 x i16> %coords_hi) {
; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32) ; GFX10NSA-NEXT: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32) ; GFX10NSA-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>) ; GFX10NSA-NEXT: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[COPY9]](<2 x s16>)
; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32) ; GFX10NSA-NEXT: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C1]](s32)
; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32) ; GFX10NSA-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10NSA-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10NSA-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GFX10NSA-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GFX10NSA-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
▲ Show 20 Lines • Show All 61 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.load.2d.d16.ll

Show All 13 Linesdefine amdgpu_ps half @image_load_f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(s16) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(s16) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](s16) ; UNPACKED-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](s16)
; UNPACKED-NEXT: $vgpr0 = COPY [[ANYEXT]](s32) ; UNPACKED-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: image_load_f16 ; PACKED-LABEL: name: image_load_f16
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(s16) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(s16) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](s16) ; PACKED-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](s16)
; PACKED-NEXT: $vgpr0 = COPY [[ANYEXT]](s32) ; PACKED-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%tex = call half @llvm.amdgcn.image.load.2d.f16.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call half @llvm.amdgcn.image.load.2d.f16.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret half %tex ret half %tex
} }
define amdgpu_ps <2 x half> @image_load_v2f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) { define amdgpu_ps <2 x half> @image_load_v2f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-LABEL: name: image_load_v2f16 ; UNPACKED-LABEL: name: image_load_v2f16
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 10 Linesdefine amdgpu_ps <2 x half> @image_load_v2f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%tex = call <2 x half> @llvm.amdgcn.image.load.2d.v2f16.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <2 x half> @llvm.amdgcn.image.load.2d.v2f16.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <2 x half> %tex ret <2 x half> %tex
} }
define amdgpu_ps <3 x half> @image_load_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) { define amdgpu_ps <3 x half> @image_load_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-LABEL: name: image_load_v3f16 ; UNPACKED-LABEL: name: image_load_v3f16
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 16 Linesdefine amdgpu_ps <3 x half> @image_load_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s16>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s16>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[UV1]](<2 x s16>) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[UV1]](<2 x s16>)
; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; PACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[BITCAST]], [[C1]] ; PACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[BITCAST]], [[C1]]
; PACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; PACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C2]], [[C]](s32) ; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C2]], [[C]](s32)
; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
Show All 17 Linesdefine amdgpu_ps <4 x half> @image_load_v4f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s16>), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s16>), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s32>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 16 Linesdefine amdgpu_ps <4 x half> @image_load_v4f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s16>), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s16>), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s16>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s16>)
; PACKED-NEXT: $vgpr0 = COPY [[UV]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[UV]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[UV1]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[UV1]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%tex = call <4 x half> @llvm.amdgcn.image.load.2d.v4f16.i32(i32 15, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <4 x half> @llvm.amdgcn.image.load.2d.v4f16.i32(i32 15, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <4 x half> %tex ret <4 x half> %tex
} }
Show All 10 Linesdefine amdgpu_ps half @image_load_tfe_f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: $vgpr0 = COPY [[UV]](s32) ; UNPACKED-NEXT: $vgpr0 = COPY [[UV]](s32)
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: image_load_tfe_f16 ; PACKED-LABEL: name: image_load_tfe_f16
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: $vgpr0 = COPY [[UV]](s32) ; PACKED-NEXT: $vgpr0 = COPY [[UV]](s32)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%res = call { half, i32 } @llvm.amdgcn.image.load.2d.sl_f16i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { half, i32 } @llvm.amdgcn.image.load.2d.sl_f16i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { half, i32 } %res, 0 %tex = extractvalue { half, i32 } %res, 0
%tfe = extractvalue { half, i32 } %res, 1 %tfe = extractvalue { half, i32 } %res, 1
store i32 %tfe, ptr addrspace(1) undef store i32 %tfe, ptr addrspace(1) undef
Show All 13 Linesdefine amdgpu_ps <2 x half> @image_load_tfe_v2f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 12 Linesdefine amdgpu_ps <2 x half> @image_load_tfe_v2f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%res = call { <2 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f16i32s.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <2 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f16i32s.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { <2 x half>, i32 } %res, 0 %tex = extractvalue { <2 x half>, i32 } %res, 0
%tfe = extractvalue { <2 x half>, i32 } %res, 1 %tfe = extractvalue { <2 x half>, i32 } %res, 1
Show All 14 Linesdefine amdgpu_ps <3 x half> @image_load_tfe_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s32>)
; UNPACKED-NEXT: G_STORE [[UV3]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV3]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
Show All 18 Linesdefine amdgpu_ps <3 x half> @image_load_tfe_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV1]](s32) ; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV1]](s32)
; PACKED-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[BITCAST1]](<2 x s16>) ; PACKED-NEXT: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[BITCAST1]](<2 x s16>)
; PACKED-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; PACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[BITCAST2]], [[C]] ; PACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[BITCAST2]], [[C]]
; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
Show All 24 Linesdefine amdgpu_ps <4 x half> @image_load_tfe_v4f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<4 x s16>), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<4 x s16>), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<5 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<5 x s32>)
; UNPACKED-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
Show All 18 Linesdefine amdgpu_ps <4 x half> @image_load_tfe_v4f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<4 x s16>), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<4 x s16>), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV1]](s32) ; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV1]](s32)
; PACKED-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[BITCAST1]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[BITCAST1]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%res = call { <4 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f16i32s.i32(i32 15, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <4 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f16i32s.i32(i32 15, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
Show All 38 Linesdefine amdgpu_ps <2 x half> @image_load_v2f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
; UNPACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>) ; UNPACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>)
Show All 9 Linesdefine amdgpu_ps <2 x half> @image_load_v2f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%tex = call <2 x half> @llvm.amdgcn.image.load.2d.v2f16.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <2 x half> @llvm.amdgcn.image.load.2d.v2f16.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <2 x half> %tex ret <2 x half> %tex
} }
define amdgpu_ps <2 x half> @image_load_v2f16_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) { define amdgpu_ps <2 x half> @image_load_v2f16_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-LABEL: name: image_load_v2f16_dmask_0000 ; UNPACKED-LABEL: name: image_load_v2f16_dmask_0000
Show All 30 Linesdefine amdgpu_ps <3 x half> @image_load_v3f16_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 15 Linesdefine amdgpu_ps <3 x half> @image_load_v3f16_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C]], [[C1]](s32) ; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C]], [[C1]](s32)
; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[C]], [[SHL]] ; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[C]], [[SHL]]
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[BITCAST]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[BITCAST]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
Show All 13 Linesdefine amdgpu_ps <3 x half> @image_load_v3f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
; UNPACKED-NEXT: [[OR1:%[0-9]+]]:_(s32) = G_OR [[C1]], [[SHL]] ; UNPACKED-NEXT: [[OR1:%[0-9]+]]:_(s32) = G_OR [[C1]], [[SHL]]
Show All 12 Linesdefine amdgpu_ps <3 x half> @image_load_v3f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C]], [[C1]](s32) ; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C]], [[C1]](s32)
; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[C]], [[SHL]] ; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[C]], [[SHL]]
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[BITCAST]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[BITCAST]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Linesdefine amdgpu_ps <4 x half> @image_load_v4f16_dmask_1110(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 16 Linesdefine amdgpu_ps <4 x half> @image_load_v4f16_dmask_1110(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s16>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s16>)
; PACKED-NEXT: $vgpr0 = COPY [[UV]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[UV]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[UV1]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[UV1]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%tex = call <4 x half> @llvm.amdgcn.image.load.2d.v4f16.i32(i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <4 x half> @llvm.amdgcn.image.load.2d.v4f16.i32(i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <4 x half> %tex ret <4 x half> %tex
} }
Show All 9 Linesdefine amdgpu_ps <4 x half> @image_load_v4f16_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 15 Linesdefine amdgpu_ps <4 x half> @image_load_v4f16_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF
; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[DEF]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[DEF]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%tex = call <4 x half> @llvm.amdgcn.image.load.2d.v4f16.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <4 x half> @llvm.amdgcn.image.load.2d.v4f16.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <4 x half> %tex ret <4 x half> %tex
} }
Show All 9 Linesdefine amdgpu_ps <4 x half> @image_load_v4f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
; UNPACKED-NEXT: [[OR1:%[0-9]+]]:_(s32) = G_OR [[C1]], [[SHL]] ; UNPACKED-NEXT: [[OR1:%[0-9]+]]:_(s32) = G_OR [[C1]], [[SHL]]
Show All 12 Linesdefine amdgpu_ps <4 x half> @image_load_v4f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s16>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF
; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[DEF]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[DEF]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%tex = call <4 x half> @llvm.amdgcn.image.load.2d.v4f16.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <4 x half> @llvm.amdgcn.image.load.2d.v4f16.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <4 x half> %tex ret <4 x half> %tex
} }
Show All 37 Linesdefine amdgpu_ps half @image_load_tfe_f16_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: $vgpr0 = COPY [[UV]](s32) ; UNPACKED-NEXT: $vgpr0 = COPY [[UV]](s32)
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: image_load_tfe_f16_dmask_0000 ; PACKED-LABEL: name: image_load_tfe_f16_dmask_0000
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: $vgpr0 = COPY [[UV]](s32) ; PACKED-NEXT: $vgpr0 = COPY [[UV]](s32)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%res = call { half, i32 } @llvm.amdgcn.image.load.2d.sl_f16i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { half, i32 } @llvm.amdgcn.image.load.2d.sl_f16i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { half, i32 } %res, 0 %tex = extractvalue { half, i32 } %res, 0
%tfe = extractvalue { half, i32 } %res, 1 %tfe = extractvalue { half, i32 } %res, 1
store i32 %tfe, ptr addrspace(1) undef store i32 %tfe, ptr addrspace(1) undef
Show All 13 Linesdefine amdgpu_ps <2 x half> @image_load_tfe_v2f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 12 Linesdefine amdgpu_ps <2 x half> @image_load_tfe_v2f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%res = call { <2 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f16i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <2 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f16i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { <2 x half>, i32 } %res, 0 %tex = extractvalue { <2 x half>, i32 } %res, 0
%tfe = extractvalue { <2 x half>, i32 } %res, 1 %tfe = extractvalue { <2 x half>, i32 } %res, 1
Show All 14 Linesdefine amdgpu_ps <2 x half> @image_load_tfe_v2f16_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 12 Linesdefine amdgpu_ps <2 x half> @image_load_tfe_v2f16_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%res = call { <2 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f16i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <2 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f16i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { <2 x half>, i32 } %res, 0 %tex = extractvalue { <2 x half>, i32 } %res, 0
%tfe = extractvalue { <2 x half>, i32 } %res, 1 %tfe = extractvalue { <2 x half>, i32 } %res, 1
Show All 14 Linesdefine amdgpu_ps <3 x half> @image_load_tfe_v3f16_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>)
; UNPACKED-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
Show All 17 Linesdefine amdgpu_ps <3 x half> @image_load_tfe_v3f16_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C]], [[C1]](s32) ; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C]], [[C1]](s32)
; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[C]], [[SHL]] ; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[C]], [[SHL]]
; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 20 Linesdefine amdgpu_ps <3 x half> @image_load_tfe_v3f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
Show All 15 Linesdefine amdgpu_ps <3 x half> @image_load_tfe_v3f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C]], [[C1]](s32) ; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C]], [[C1]](s32)
; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[C]], [[SHL]] ; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[C]], [[SHL]]
; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 20 Linesdefine amdgpu_ps <3 x half> @image_load_tfe_v3f16_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
Show All 15 Linesdefine amdgpu_ps <3 x half> @image_load_tfe_v3f16_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; PACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; PACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C]], [[C1]](s32) ; PACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C]], [[C1]](s32)
; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[C]], [[SHL]] ; PACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[C]], [[SHL]]
; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32) ; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
Show All 20 Linesdefine amdgpu_ps <4 x half> @image_load_tfe_v4f16_dmask_1110(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<4 x s32>)
; UNPACKED-NEXT: G_STORE [[UV3]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV3]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
Show All 18 Linesdefine amdgpu_ps <4 x half> @image_load_tfe_v4f16_dmask_1110(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s16>), align 8, addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV1]](s32) ; PACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV1]](s32)
; PACKED-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[BITCAST1]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[BITCAST1]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%res = call { <4 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f16i32s.i32(i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <4 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f16i32s.i32(i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
Show All 16 Linesdefine amdgpu_ps <4 x half> @image_load_tfe_v4f16_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<3 x s32>)
; UNPACKED-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]] ; UNPACKED-NEXT: [[AND1:%[0-9]+]]:_(s32) = G_AND [[UV1]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
Show All 17 Linesdefine amdgpu_ps <4 x half> @image_load_tfe_v4f16_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s16>), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: [[DEF1:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF1:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[DEF1]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[DEF1]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%res = call { <4 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f16i32s.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <4 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f16i32s.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
Show All 16 Linesdefine amdgpu_ps <4 x half> @image_load_tfe_v4f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
Show All 15 Linesdefine amdgpu_ps <4 x half> @image_load_tfe_v4f16_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: [[DEF1:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF1:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[DEF1]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[DEF1]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%res = call { <4 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f16i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <4 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f16i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
Show All 16 Linesdefine amdgpu_ps <4 x half> @image_load_tfe_v4f16_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; UNPACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; UNPACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; UNPACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; UNPACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]] ; UNPACKED-NEXT: [[AND:%[0-9]+]]:_(s32) = G_AND [[UV]], [[C]]
; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; UNPACKED-NEXT: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32) ; UNPACKED-NEXT: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[C1]], [[C2]](s32)
; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]] ; UNPACKED-NEXT: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
Show All 15 Linesdefine amdgpu_ps <4 x half> @image_load_tfe_v4f16_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; PACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; PACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; PACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 7) ; PACKED-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD_D16_:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD_D16 intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s16), addrspace 8)
; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>) ; PACKED-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD_D16_]](<2 x s32>)
; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32) ; PACKED-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[UV]](s32)
; PACKED-NEXT: [[DEF1:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF ; PACKED-NEXT: [[DEF1:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF
; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; PACKED-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>) ; PACKED-NEXT: $vgpr0 = COPY [[BITCAST]](<2 x s16>)
; PACKED-NEXT: $vgpr1 = COPY [[DEF1]](<2 x s16>) ; PACKED-NEXT: $vgpr1 = COPY [[DEF1]](<2 x s16>)
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%res = call { <4 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f16i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <4 x half>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f16i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
Show All 16 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.load.2d.ll

Show All 12 Linesdefine amdgpu_ps float @image_load_f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GCN-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%tex = call float @llvm.amdgcn.image.load.2d.f32.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call float @llvm.amdgcn.image.load.2d.f32.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret float %tex ret float %tex
} }
define amdgpu_ps <2 x float> @image_load_v2f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t) { define amdgpu_ps <2 x float> @image_load_v2f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-LABEL: name: image_load_v2f32 ; GCN-LABEL: name: image_load_v2f32
; GCN: bb.1 (%ir-block.0): ; GCN: bb.1 (%ir-block.0):
; GCN-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1 ; GCN-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1
; GCN-NEXT: {{ $}} ; GCN-NEXT: {{ $}}
; GCN-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GCN-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GCN-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GCN-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GCN-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GCN-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GCN-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GCN-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%tex = call <2 x float> @llvm.amdgcn.image.load.2d.v2f32.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <2 x float> @llvm.amdgcn.image.load.2d.v2f32.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <2 x float> %tex ret <2 x float> %tex
} }
Show All 9 Linesdefine amdgpu_ps <3 x float> @image_load_v3f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s32>), align 16, addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s32>), align 16, addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<3 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<3 x s32>)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%tex = call <3 x float> @llvm.amdgcn.image.load.2d.v3f32.i32(i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <3 x float> @llvm.amdgcn.image.load.2d.v3f32.i32(i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <3 x float> %tex ret <3 x float> %tex
} }
Show All 10 Linesdefine amdgpu_ps <4 x float> @image_load_v4f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GCN-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GCN-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
%tex = call <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32 15, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32 15, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <4 x float> %tex ret <4 x float> %tex
Show All 12 Linesdefine amdgpu_ps float @image_load_tfe_f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%res = call { float, i32 } @llvm.amdgcn.image.load.2d.sl_f32i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { float, i32 } @llvm.amdgcn.image.load.2d.sl_f32i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { float, i32 } %res, 0 %tex = extractvalue { float, i32 } %res, 0
%tfe = extractvalue { float, i32 } %res, 1 %tfe = extractvalue { float, i32 } %res, 1
store i32 %tfe, ptr addrspace(1) undef store i32 %tfe, ptr addrspace(1) undef
Show All 13 Linesdefine amdgpu_ps <2 x float> @image_load_tfe_v2f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<3 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<3 x s32>)
; GCN-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%res = call { <2 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f32i32s.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <2 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f32i32s.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { <2 x float>, i32 } %res, 0 %tex = extractvalue { <2 x float>, i32 } %res, 0
%tfe = extractvalue { <2 x float>, i32 } %res, 1 %tfe = extractvalue { <2 x float>, i32 } %res, 1
Show All 14 Linesdefine amdgpu_ps <3 x float> @image_load_tfe_v3f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s32>), align 16, addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s32>), align 16, addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GCN-NEXT: G_STORE [[UV3]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV3]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%res = call { <3 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v3f32i32s.i32(i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <3 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v3f32i32s.i32(i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { <3 x float>, i32 } %res, 0 %tex = extractvalue { <3 x float>, i32 } %res, 0
Show All 15 Linesdefine amdgpu_ps <4 x float> @image_load_tfe_v4f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GCN-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV4]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GCN-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GCN-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
%res = call { <4 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f32i32s.i32(i32 15, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <4 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v4f32i32s.i32(i32 15, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
Show All 29 Linesdefine amdgpu_ps <2 x float> @image_load_v2f32_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GCN-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GCN-NEXT: $vgpr1 = COPY [[DEF]](s32) ; GCN-NEXT: $vgpr1 = COPY [[DEF]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%tex = call <2 x float> @llvm.amdgcn.image.load.2d.v2f32.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <2 x float> @llvm.amdgcn.image.load.2d.v2f32.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <2 x float> %tex ret <2 x float> %tex
} }
Show All 25 Linesdefine amdgpu_ps <3 x float> @image_load_v3f32_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[DEF]](s32) ; GCN-NEXT: $vgpr2 = COPY [[DEF]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%tex = call <3 x float> @llvm.amdgcn.image.load.2d.v3f32.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <3 x float> @llvm.amdgcn.image.load.2d.v3f32.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <3 x float> %tex ret <3 x float> %tex
Show All 11 Linesdefine amdgpu_ps <3 x float> @image_load_v3f32_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GCN-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GCN-NEXT: $vgpr1 = COPY [[DEF]](s32) ; GCN-NEXT: $vgpr1 = COPY [[DEF]](s32)
; GCN-NEXT: $vgpr2 = COPY [[DEF]](s32) ; GCN-NEXT: $vgpr2 = COPY [[DEF]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%tex = call <3 x float> @llvm.amdgcn.image.load.2d.v3f32.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <3 x float> @llvm.amdgcn.image.load.2d.v3f32.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <3 x float> %tex ret <3 x float> %tex
} }
Show All 27 Linesdefine amdgpu_ps <4 x float> @image_load_v4f32_dmask_1110(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s32>), align 16, addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<3 x s32>), align 16, addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<3 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<3 x s32>)
; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GCN-NEXT: $vgpr3 = COPY [[DEF]](s32) ; GCN-NEXT: $vgpr3 = COPY [[DEF]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
%tex = call <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
Show All 12 Linesdefine amdgpu_ps <4 x float> @image_load_v4f32_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[DEF]](s32) ; GCN-NEXT: $vgpr2 = COPY [[DEF]](s32)
; GCN-NEXT: $vgpr3 = COPY [[DEF]](s32) ; GCN-NEXT: $vgpr3 = COPY [[DEF]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
%tex = call <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
Show All 12 Linesdefine amdgpu_ps <4 x float> @image_load_v4f32_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GCN-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GCN-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GCN-NEXT: $vgpr1 = COPY [[DEF]](s32) ; GCN-NEXT: $vgpr1 = COPY [[DEF]](s32)
; GCN-NEXT: $vgpr2 = COPY [[DEF]](s32) ; GCN-NEXT: $vgpr2 = COPY [[DEF]](s32)
; GCN-NEXT: $vgpr3 = COPY [[DEF]](s32) ; GCN-NEXT: $vgpr3 = COPY [[DEF]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
%tex = call <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) %tex = call <4 x float> @llvm.amdgcn.image.load.2d.v4f32.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret <4 x float> %tex ret <4 x float> %tex
Show All 30 Linesdefine amdgpu_ps float @image_load_tfe_f32_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%res = call { float, i32 } @llvm.amdgcn.image.load.2d.sl_f32i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { float, i32 } @llvm.amdgcn.image.load.2d.sl_f32i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { float, i32 } %res, 0 %tex = extractvalue { float, i32 } %res, 0
%tfe = extractvalue { float, i32 } %res, 1 %tfe = extractvalue { float, i32 } %res, 1
store i32 %tfe, ptr addrspace(1) undef store i32 %tfe, ptr addrspace(1) undef
Show All 13 Linesdefine amdgpu_ps <2 x float> @image_load_tfe_v2f32_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%res = call { <2 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f32i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <2 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f32i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { <2 x float>, i32 } %res, 0 %tex = extractvalue { <2 x float>, i32 } %res, 0
Show All 15 Linesdefine amdgpu_ps <2 x float> @image_load_tfe_v2f32_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%res = call { <2 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f32i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <2 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v2f32i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { <2 x float>, i32 } %res, 0 %tex = extractvalue { <2 x float>, i32 } %res, 0
Show All 15 Linesdefine amdgpu_ps <3 x float> @image_load_tfe_v3f32_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<3 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<3 x s32>)
; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%res = call { <3 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v3f32i32s.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <3 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v3f32i32s.i32(i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
Show All 16 Linesdefine amdgpu_ps <3 x float> @image_load_tfe_v3f32_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%res = call { <3 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v3f32i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <3 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v3f32i32s.i32(i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
Show All 16 Linesdefine amdgpu_ps <3 x float> @image_load_tfe_v3f32_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%res = call { <3 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v3f32i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0) %res = call { <3 x float>, i32 } @llvm.amdgcn.image.load.2d.sl_v3f32i32s.i32(i32 0, i32 %s, i32 %t, <8 x i32> %rsrc, i32 1, i32 0)
Show All 16 Linesdefine amdgpu_ps <4 x float> @image_load_tfe_v4f32_dmask_1110(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s32>), align 16, addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<3 x s32>), align 16, addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: G_STORE [[UV3]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV3]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GCN-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GCN-NEXT: $vgpr3 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr3 = COPY [[DEF1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
Show All 17 Linesdefine amdgpu_ps <4 x float> @image_load_tfe_v4f32_dmask_1100(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<3 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<3 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<3 x s32>)
; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV2]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32)
; GCN-NEXT: $vgpr3 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr3 = COPY [[DEF1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
Show All 17 Linesdefine amdgpu_ps <4 x float> @image_load_tfe_v4f32_dmask_1000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32)
; GCN-NEXT: $vgpr3 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr3 = COPY [[DEF1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
Show All 17 Linesdefine amdgpu_ps <4 x float> @image_load_tfe_v4f32_dmask_0000(<8 x i32> inreg %rsrc, i32 %s, i32 %t) {
; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GCN-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GCN-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GCN-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GCN-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GCN-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GCN-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GCN-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GCN-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2d), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GCN-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GCN-NEXT: [[DEF1:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GCN-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GCN-NEXT: $vgpr0 = COPY [[UV]](s32) ; GCN-NEXT: $vgpr0 = COPY [[UV]](s32)
; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr1 = COPY [[DEF1]](s32)
; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr2 = COPY [[DEF1]](s32)
; GCN-NEXT: $vgpr3 = COPY [[DEF1]](s32) ; GCN-NEXT: $vgpr3 = COPY [[DEF1]](s32)
; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GCN-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
Show All 17 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.load.2darraymsaa.ll

Show All 16 Linesdefine amdgpu_ps <4 x float> @load_2darraymsaa(<8 x i32> inreg %rsrc, i32 %s, i32 %t, i32 %slice, i32 %fragid) {
; GFX6-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX6-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX6-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX6-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX6-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX6-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX6-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX6-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX6-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX6-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX6-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX6-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX6-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX6-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX6-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX6-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX6-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[BUILD_VECTOR1]](<4 x s32>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX6-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[BUILD_VECTOR1]](<4 x s32>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX6-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX6-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX6-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX6-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX6-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX6-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX6-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX6-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX6-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX6-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_2darraymsaa ; GFX10NSA-LABEL: name: load_2darraymsaa
; GFX10NSA: bb.1 (%ir-block.0): ; GFX10NSA: bb.1 (%ir-block.0):
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
%v = call <4 x float> @llvm.amdgcn.image.load.2darraymsaa.v4f32.i32(i32 15, i32 %s, i32 %t, i32 %slice, i32 %fragid, <8 x i32> %rsrc, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.load.2darraymsaa.v4f32.i32(i32 15, i32 %s, i32 %t, i32 %slice, i32 %fragid, <8 x i32> %rsrc, i32 0, i32 0)
ret <4 x float> %v ret <4 x float> %v
Show All 16 Linesdefine amdgpu_ps <4 x float> @load_2darraymsaa_tfe(<8 x i32> inreg %rsrc, ptr addrspace(1) inreg %out, i32 %s, i32 %t, i32 %slice, i32 %fragid) {
; GFX6-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10 ; GFX6-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
; GFX6-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11 ; GFX6-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
; GFX6-NEXT: [[MV:%[0-9]+]]:_(p1) = G_MERGE_VALUES [[COPY8]](s32), [[COPY9]](s32) ; GFX6-NEXT: [[MV:%[0-9]+]]:_(p1) = G_MERGE_VALUES [[COPY8]](s32), [[COPY9]](s32)
; GFX6-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX6-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX6-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX6-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX6-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX6-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX6-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX6-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX6-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY10]](s32), [[COPY11]](s32), [[COPY12]](s32), [[COPY13]](s32) ; GFX6-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY10]](s32), [[COPY11]](s32), [[COPY12]](s32), [[COPY13]](s32)
; GFX6-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[BUILD_VECTOR1]](<4 x s32>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX6-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[BUILD_VECTOR1]](<4 x s32>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX6-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GFX6-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GFX6-NEXT: G_STORE [[UV4]](s32), [[MV]](p1) :: (store (s32) into %ir.out, addrspace 1) ; GFX6-NEXT: G_STORE [[UV4]](s32), [[MV]](p1) :: (store (s32) into %ir.out, addrspace 1)
; GFX6-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX6-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX6-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX6-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX6-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX6-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX6-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX6-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10NSA-LABEL: name: load_2darraymsaa_tfe ; GFX10NSA-LABEL: name: load_2darraymsaa_tfe
Show All 11 Linesdefine amdgpu_ps <4 x float> @load_2darraymsaa_tfe(<8 x i32> inreg %rsrc, ptr addrspace(1) inreg %out, i32 %s, i32 %t, i32 %slice, i32 %fragid) {
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
; GFX10NSA-NEXT: [[MV:%[0-9]+]]:_(p1) = G_MERGE_VALUES [[COPY8]](s32), [[COPY9]](s32) ; GFX10NSA-NEXT: [[MV:%[0-9]+]]:_(p1) = G_MERGE_VALUES [[COPY8]](s32), [[COPY9]](s32)
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10NSA-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[COPY10]](s32), [[COPY11]](s32), [[COPY12]](s32), [[COPY13]](s32), [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<5 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.2darraymsaa), 15, [[COPY10]](s32), [[COPY11]](s32), [[COPY12]](s32), [[COPY13]](s32), [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<5 x s32>)
; GFX10NSA-NEXT: G_STORE [[UV4]](s32), [[MV]](p1) :: (store (s32) into %ir.out, addrspace 1) ; GFX10NSA-NEXT: G_STORE [[UV4]](s32), [[MV]](p1) :: (store (s32) into %ir.out, addrspace 1)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10NSA-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10NSA-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10NSA-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
%v = call { <4 x float>, i32 } @llvm.amdgcn.image.load.2darraymsaa.sl_v4f32i32s.i32(i32 15, i32 %s, i32 %t, i32 %slice, i32 %fragid, <8 x i32> %rsrc, i32 1, i32 0) %v = call { <4 x float>, i32 } @llvm.amdgcn.image.load.2darraymsaa.sl_v4f32i32s.i32(i32 15, i32 %s, i32 %t, i32 %slice, i32 %fragid, <8 x i32> %rsrc, i32 1, i32 0)
Show All 10 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.load.3d.ll

Show All 15 Linesdefine amdgpu_ps float @image_load_3d_f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t, i32 %r) {
; GFX6-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX6-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX6-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX6-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX6-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX6-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX6-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX6-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX6-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX6-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX6-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX6-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX6-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX6-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX6-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32) ; GFX6-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32)
; GFX6-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 1, [[BUILD_VECTOR1]](<3 x s32>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GFX6-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 1, [[BUILD_VECTOR1]](<3 x s32>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GFX6-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX6-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: image_load_3d_f32 ; GFX10NSA-LABEL: name: image_load_3d_f32
; GFX10NSA: bb.1 (%ir-block.0): ; GFX10NSA: bb.1 (%ir-block.0):
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 1, [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 1, [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%tex = call float @llvm.amdgcn.image.load.3d.f32.i32(i32 1, i32 %s, i32 %t, i32 %r, <8 x i32> %rsrc, i32 0, i32 0) %tex = call float @llvm.amdgcn.image.load.3d.f32.i32(i32 1, i32 %s, i32 %t, i32 %r, <8 x i32> %rsrc, i32 0, i32 0)
ret float %tex ret float %tex
} }
define amdgpu_ps float @image_load_3d_tfe_f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t, i32 %r) { define amdgpu_ps float @image_load_3d_tfe_f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t, i32 %r) {
; GFX6-LABEL: name: image_load_3d_tfe_f32 ; GFX6-LABEL: name: image_load_3d_tfe_f32
Show All 9 Linesdefine amdgpu_ps float @image_load_3d_tfe_f32(<8 x i32> inreg %rsrc, i32 %s, i32 %t, i32 %r) {
; GFX6-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX6-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX6-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX6-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX6-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX6-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX6-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX6-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX6-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX6-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX6-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX6-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX6-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GFX6-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GFX6-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32) ; GFX6-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32)
; GFX6-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 1, [[BUILD_VECTOR1]](<3 x s32>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GFX6-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 1, [[BUILD_VECTOR1]](<3 x s32>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GFX6-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GFX6-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GFX6-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GFX6-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GFX6-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX6-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10NSA-LABEL: name: image_load_3d_tfe_f32 ; GFX10NSA-LABEL: name: image_load_3d_tfe_f32
; GFX10NSA: bb.1 (%ir-block.0): ; GFX10NSA: bb.1 (%ir-block.0):
; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX10NSA-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX10NSA-NEXT: {{ $}} ; GFX10NSA-NEXT: {{ $}}
; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10NSA-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10NSA-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10NSA-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10NSA-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10NSA-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10NSA-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10NSA-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10NSA-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10NSA-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10NSA-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10NSA-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10NSA-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF ; GFX10NSA-NEXT: [[DEF:%[0-9]+]]:_(p1) = G_IMPLICIT_DEF
; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 1, [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 7) ; GFX10NSA-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.3d), 1, [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[BUILD_VECTOR]](<8 x s32>), 1, 0, 0 :: (dereferenceable load (s32), addrspace 8)
; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GFX10NSA-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GFX10NSA-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1) ; GFX10NSA-NEXT: G_STORE [[UV1]](s32), [[DEF]](p1) :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10NSA-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10NSA-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call { float, i32 } @llvm.amdgcn.image.load.3d.sl_f32i32s.i32(i32 1, i32 %s, i32 %t, i32 %r, <8 x i32> %rsrc, i32 1, i32 0) %val = call { float, i32 } @llvm.amdgcn.image.load.3d.sl_f32i32s.i32(i32 1, i32 %s, i32 %t, i32 %r, <8 x i32> %rsrc, i32 1, i32 0)
%tex = extractvalue { float, i32 } %val, 0 %tex = extractvalue { float, i32 } %val, 0
%tfe = extractvalue { float, i32 } %val, 1 %tfe = extractvalue { float, i32 } %val, 1
store i32 %tfe, ptr addrspace(1) undef store i32 %tfe, ptr addrspace(1) undef
ret float %tex ret float %tex
} }
declare float @llvm.amdgcn.image.load.3d.f32.i32(i32 immarg, i32, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0 declare float @llvm.amdgcn.image.load.3d.f32.i32(i32 immarg, i32, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0
declare { float, i32 } @llvm.amdgcn.image.load.3d.sl_f32i32s.i32(i32 immarg, i32, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0 declare { float, i32 } @llvm.amdgcn.image.load.3d.sl_f32i32s.i32(i32 immarg, i32, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0
attributes #0 = { nounwind readonly } attributes #0 = { nounwind readonly }

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.sample.a16.ll

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s) {
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_1d ; GFX10-LABEL: name: sample_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 10 Linesdefine amdgpu_ps <4 x float> @sample_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s) {
; GFX10-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10 ; GFX10-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
; GFX10-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11 ; GFX10-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_1d ; GFX11-LABEL: name: sample_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 10 Linesdefine amdgpu_ps <4 x float> @sample_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s) {
; GFX11-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX11-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX11-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10 ; GFX11-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
; GFX11-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11 ; GFX11-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f16(i32 15, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f16(i32 15, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_2d ; GFX10-LABEL: name: sample_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @sample_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t) {
; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_2d ; GFX11-LABEL: name: sample_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @sample_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t) {
; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.2d.v4f32.f16(i32 15, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.2d.v4f32.f16(i32 15, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %r) {
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_3d ; GFX10-LABEL: name: sample_3d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %r) {
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_3d ; GFX11-LABEL: name: sample_3d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %r) {
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.3d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.3d.v4f32.f16(i32 15, half %s, half %t, half %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.3d.v4f32.f16(i32 15, half %s, half %t, half %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_cube(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %face) {
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_cube ; GFX10-LABEL: name: sample_cube
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_cube(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %face) {
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cube ; GFX11-LABEL: name: sample_cube
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_cube(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %face) {
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cube), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cube.v4f32.f16(i32 15, half %s, half %t, half %face, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cube.v4f32.f16(i32 15, half %s, half %t, half %face, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_1darray(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %slice) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1darray), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1darray), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_1darray ; GFX10-LABEL: name: sample_1darray
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @sample_1darray(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %slice) {
; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1darray), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1darray), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_1darray ; GFX11-LABEL: name: sample_1darray
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @sample_1darray(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %slice) {
; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1darray), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1darray), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.1darray.v4f32.f16(i32 15, half %s, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.1darray.v4f32.f16(i32 15, half %s, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_2darray(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %slice) {
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_2darray ; GFX10-LABEL: name: sample_2darray
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_2darray(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %slice) {
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_2darray ; GFX11-LABEL: name: sample_2darray
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_2darray(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %slice) {
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.2darray), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.2darray.v4f32.f16(i32 15, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.2darray.v4f32.f16(i32 15, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 21 Linesdefine amdgpu_ps <4 x float> @sample_c_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s) {
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_1d ; GFX10-LABEL: name: sample_c_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 15 Linesdefine amdgpu_ps <4 x float> @sample_c_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s) {
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_1d ; GFX11-LABEL: name: sample_c_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 15 Linesdefine amdgpu_ps <4 x float> @sample_c_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s) {
; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.1d.v4f32.f16(i32 15, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.1d.v4f32.f16(i32 15, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_c_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t) {
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_2d ; GFX10-LABEL: name: sample_c_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_c_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t) {
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_2d ; GFX11-LABEL: name: sample_c_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_c_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t) {
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %clamp) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_cl_1d ; GFX10-LABEL: name: sample_cl_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @sample_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %clamp) {
; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cl_1d ; GFX11-LABEL: name: sample_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @sample_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %clamp) {
; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cl.1d.v4f32.f16(i32 15, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cl.1d.v4f32.f16(i32 15, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %clamp) {
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_cl_2d ; GFX10-LABEL: name: sample_cl_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %clamp) {
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cl_2d ; GFX11-LABEL: name: sample_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %clamp) {
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cl.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cl.2d.v4f32.f16(i32 15, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cl.2d.v4f32.f16(i32 15, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_c_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %clamp) {
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_cl_1d ; GFX10-LABEL: name: sample_c_cl_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_c_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %clamp) {
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_cl_1d ; GFX11-LABEL: name: sample_c_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_c_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %clamp) {
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.cl.1d.v4f32.f16(i32 15, float %zcompare, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.cl.1d.v4f32.f16(i32 15, float %zcompare, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 26 Linesdefine amdgpu_ps <4 x float> @sample_c_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t, half %clamp) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_cl_2d ; GFX10-LABEL: name: sample_c_cl_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t, half %clamp) {
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_cl_2d ; GFX11-LABEL: name: sample_c_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t, half %clamp) {
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.cl.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.cl.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_b_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s) {
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_b_1d ; GFX10-LABEL: name: sample_b_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_b_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s) {
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_b_1d ; GFX11-LABEL: name: sample_b_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_b_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s) {
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.b.1d.v4f32.f16.f16(i32 15, half %bias, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.b.1d.v4f32.f16.f16(i32 15, half %bias, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_b_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %t) {
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_b_2d ; GFX10-LABEL: name: sample_b_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_b_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %t) {
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_b_2d ; GFX11-LABEL: name: sample_b_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_b_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %t) {
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.b.2d.v4f32.f16.f16(i32 15, half %bias, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.b.2d.v4f32.f16.f16(i32 15, half %bias, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_c_b_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s) {
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_b_1d ; GFX10-LABEL: name: sample_c_b_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 17 Linesdefine amdgpu_ps <4 x float> @sample_c_b_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s) {
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_b_1d ; GFX11-LABEL: name: sample_c_b_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 17 Linesdefine amdgpu_ps <4 x float> @sample_c_b_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s) {
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.b.1d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.1d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 26 Linesdefine amdgpu_ps <4 x float> @sample_c_b_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %t) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_b_2d ; GFX10-LABEL: name: sample_c_b_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_b_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %t) {
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_b_2d ; GFX11-LABEL: name: sample_c_b_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_b_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %t) {
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.b.2d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.2d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_b_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %clamp) {
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_b_cl_1d ; GFX10-LABEL: name: sample_b_cl_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_b_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %clamp) {
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_b_cl_1d ; GFX11-LABEL: name: sample_b_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_b_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %clamp) {
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.b.cl.1d.v4f32.f16.f16(i32 15, half %bias, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.b.cl.1d.v4f32.f16.f16(i32 15, half %bias, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_b_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %t, half %clamp) {
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_b_cl_2d ; GFX10-LABEL: name: sample_b_cl_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_b_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %t, half %clamp) {
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_b_cl_2d ; GFX11-LABEL: name: sample_b_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_b_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, half %s, half %t, half %clamp) {
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.b.cl.2d.v4f32.f16.f16(i32 15, half %bias, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.b.cl.2d.v4f32.f16.f16(i32 15, half %bias, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 26 Linesdefine amdgpu_ps <4 x float> @sample_c_b_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %clamp) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_b_cl_1d ; GFX10-LABEL: name: sample_c_b_cl_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_b_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %clamp) {
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_b_cl_1d ; GFX11-LABEL: name: sample_c_b_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_b_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %clamp) {
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.b.cl.1d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.cl.1d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 29 Linesdefine amdgpu_ps <4 x float> @sample_c_b_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %t, half %clamp) {
; GFX9-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX9-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_b_cl_2d ; GFX10-LABEL: name: sample_c_b_cl_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_c_b_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %t, half %clamp) {
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_b_cl_2d ; GFX11-LABEL: name: sample_c_b_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_c_b_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %bias, float %zcompare, half %s, half %t, half %clamp) {
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[TRUNC2]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.b.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.b.cl.2d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.b.cl.2d.v4f32.f16.f16(i32 15, half %bias, float %zcompare, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_d_1d ; GFX10-LABEL: name: sample_d_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) {
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_1d ; GFX11-LABEL: name: sample_d_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) {
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 30 Linesdefine amdgpu_ps <4 x float> @sample_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX9-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX9-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX9-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX9-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX9-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX9-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX9-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX9-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_d_2d ; GFX10-LABEL: name: sample_d_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 23 Linesdefine amdgpu_ps <4 x float> @sample_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX10-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_2d ; GFX11-LABEL: name: sample_d_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 23 Linesdefine amdgpu_ps <4 x float> @sample_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX11-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 40 Linesdefine amdgpu_ps <4 x float> @sample_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r) {
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16) ; GFX9-NEXT: [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16)
; GFX9-NEXT: [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_d_3d ; GFX10-LABEL: name: sample_d_3d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 34 Linesdefine amdgpu_ps <4 x float> @sample_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r) {
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16) ; GFX10-NEXT: [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16)
; GFX10-NEXT: [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_3d ; GFX11-LABEL: name: sample_d_3d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 34 Linesdefine amdgpu_ps <4 x float> @sample_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r) {
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16) ; GFX11-NEXT: [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16)
; GFX11-NEXT: [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_c_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s) {
; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_d_1d ; GFX10-LABEL: name: sample_c_d_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_c_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s) {
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_1d ; GFX11-LABEL: name: sample_c_d_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_c_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s) {
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 32 Linesdefine amdgpu_ps <4 x float> @sample_c_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX9-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX9-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX9-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX9-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX9-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX9-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_d_2d ; GFX10-LABEL: name: sample_c_d_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_c_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX10-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_2d ; GFX11-LABEL: name: sample_c_d_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_c_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX11-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_d_cl_1d ; GFX10-LABEL: name: sample_d_cl_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_cl_1d ; GFX11-LABEL: name: sample_d_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 34 Linesdefine amdgpu_ps <4 x float> @sample_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX9-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX9-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX9-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX9-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_d_cl_2d ; GFX10-LABEL: name: sample_d_cl_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX10-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_cl_2d ; GFX11-LABEL: name: sample_d_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX11-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 29 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX9-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX9-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_d_cl_1d ; GFX10-LABEL: name: sample_c_d_cl_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_cl_1d ; GFX11-LABEL: name: sample_c_d_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 36 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX9-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32) ; GFX9-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<10 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<10 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[CONCAT_VECTORS]](<10 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[CONCAT_VECTORS]](<10 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_d_cl_2d ; GFX10-LABEL: name: sample_c_d_cl_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 29 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX10-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32) ; GFX10-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_cl_2d ; GFX11-LABEL: name: sample_c_d_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 29 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX11-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32) ; GFX11-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_cd_1d ; GFX10-LABEL: name: sample_cd_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) {
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cd_1d ; GFX11-LABEL: name: sample_cd_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) {
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cd.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cd.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 30 Linesdefine amdgpu_ps <4 x float> @sample_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX9-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX9-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX9-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX9-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX9-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX9-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX9-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX9-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_cd_2d ; GFX10-LABEL: name: sample_cd_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 23 Linesdefine amdgpu_ps <4 x float> @sample_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX10-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cd_2d ; GFX11-LABEL: name: sample_cd_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 23 Linesdefine amdgpu_ps <4 x float> @sample_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX11-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cd.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cd.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s) {
; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_cd_1d ; GFX10-LABEL: name: sample_c_cd_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s) {
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_cd_1d ; GFX11-LABEL: name: sample_c_cd_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s) {
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 32 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX9-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX9-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX9-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX9-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX9-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX9-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_cd_2d ; GFX10-LABEL: name: sample_c_cd_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX10-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_cd_2d ; GFX11-LABEL: name: sample_c_cd_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX11-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_cd_cl_1d ; GFX10-LABEL: name: sample_cd_cl_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cd_cl_1d ; GFX11-LABEL: name: sample_cd_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 34 Linesdefine amdgpu_ps <4 x float> @sample_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX9-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX9-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX9-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX9-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_cd_cl_2d ; GFX10-LABEL: name: sample_cd_cl_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX10-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cd_cl_2d ; GFX11-LABEL: name: sample_cd_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32) ; GFX11-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY18]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 29 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX9-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX9-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX9-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[CONCAT_VECTORS]](<8 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_cd_cl_1d ; GFX10-LABEL: name: sample_c_cd_cl_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_cd_cl_1d ; GFX11-LABEL: name: sample_c_cd_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp) {
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.1d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dsdv, half %s, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 36 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX9-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32) ; GFX9-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<10 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<10 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[CONCAT_VECTORS]](<10 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[CONCAT_VECTORS]](<10 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_cd_cl_2d ; GFX10-LABEL: name: sample_c_cd_cl_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 29 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX10-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32) ; GFX10-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_cd_cl_2d ; GFX11-LABEL: name: sample_c_cd_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 29 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp) {
; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX11-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32) ; GFX11-NEXT: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[COPY19]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.2d.v4f32.f32.f16(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_l_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %lod) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_l_1d ; GFX10-LABEL: name: sample_l_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @sample_l_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %lod) {
; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_l_1d ; GFX11-LABEL: name: sample_l_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @sample_l_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %lod) {
; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.l.1d.v4f32.f16(i32 15, half %s, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.l.1d.v4f32.f16(i32 15, half %s, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_l_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %lod) {
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_l_2d ; GFX10-LABEL: name: sample_l_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_l_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %lod) {
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_l_2d ; GFX11-LABEL: name: sample_l_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_l_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t, half %lod) {
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.l.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.l.2d.v4f32.f16(i32 15, half %s, half %t, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.l.2d.v4f32.f16(i32 15, half %s, half %t, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_c_l_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %lod) {
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_l_1d ; GFX10-LABEL: name: sample_c_l_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_c_l_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %lod) {
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_l_1d ; GFX11-LABEL: name: sample_c_l_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_c_l_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %lod) {
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.l.1d.v4f32.f16(i32 15, float %zcompare, half %s, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.l.1d.v4f32.f16(i32 15, float %zcompare, half %s, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 26 Linesdefine amdgpu_ps <4 x float> @sample_c_l_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t, half %lod) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX9-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.2d), 15, [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_l_2d ; GFX10-LABEL: name: sample_c_l_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_l_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t, half %lod) {
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_l_2d ; GFX11-LABEL: name: sample_c_l_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_l_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t, half %lod) {
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.l.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.l.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.l.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, half %lod, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_lz_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s) {
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_lz_1d ; GFX10-LABEL: name: sample_lz_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 10 Linesdefine amdgpu_ps <4 x float> @sample_lz_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s) {
; GFX10-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10 ; GFX10-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
; GFX10-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11 ; GFX10-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_lz_1d ; GFX11-LABEL: name: sample_lz_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 10 Linesdefine amdgpu_ps <4 x float> @sample_lz_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s) {
; GFX11-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX11-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX11-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10 ; GFX11-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $sgpr10
; GFX11-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11 ; GFX11-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $sgpr11
; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.1d), 15, [[TRUNC]](s16), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.lz.1d.v4f32.f16(i32 15, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.lz.1d.v4f32.f16(i32 15, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_lz_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t) {
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_lz_2d ; GFX10-LABEL: name: sample_lz_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @sample_lz_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t) {
; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_lz_2d ; GFX11-LABEL: name: sample_lz_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 13 Linesdefine amdgpu_ps <4 x float> @sample_lz_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %s, half %t) {
; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12 ; GFX11-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $sgpr12
; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13 ; GFX11-NEXT: [[COPY11:%[0-9]+]]:_(s32) = COPY $sgpr13
; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.lz.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.lz.2d.v4f32.f16(i32 15, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.lz.2d.v4f32.f16(i32 15, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 21 Linesdefine amdgpu_ps <4 x float> @sample_c_lz_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s) {
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_lz_1d ; GFX10-LABEL: name: sample_c_lz_1d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 15 Linesdefine amdgpu_ps <4 x float> @sample_c_lz_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s) {
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_lz_1d ; GFX11-LABEL: name: sample_c_lz_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 15 Linesdefine amdgpu_ps <4 x float> @sample_c_lz_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s) {
; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GFX11-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.1d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.lz.1d.v4f32.f16(i32 15, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.lz.1d.v4f32.f16(i32 15, float %zcompare, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_c_lz_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t) {
; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX9-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX9-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX10-LABEL: name: sample_c_lz_2d ; GFX10-LABEL: name: sample_c_lz_2d
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_c_lz_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t) {
; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_lz_2d ; GFX11-LABEL: name: sample_c_lz_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 16 Linesdefine amdgpu_ps <4 x float> @sample_c_lz_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %s, half %t) {
; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX11-NEXT: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.lz.2d), 15, [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 1 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.lz.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.lz.2d.v4f32.f16(i32 15, float %zcompare, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 38 Linesdefine amdgpu_ps float @sample_c_d_o_2darray_V1(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) {
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (s32), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (s32), addrspace 8)
; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10-LABEL: name: sample_c_d_o_2darray_V1 ; GFX10-LABEL: name: sample_c_d_o_2darray_V1
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8 ; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
; GFX10-NEXT: {{ $}} ; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
Show All 28 Linesdefine amdgpu_ps float @sample_c_d_o_2darray_V1(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) {
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (s32), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (s32), addrspace 8)
; GFX10-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX11-LABEL: name: sample_c_d_o_2darray_V1 ; GFX11-LABEL: name: sample_c_d_o_2darray_V1
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
; GFX11-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8 ; GFX11-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX11-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX11-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
Show All 28 Linesdefine amdgpu_ps float @sample_c_d_o_2darray_V1(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) {
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (s32), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (s32), addrspace 8)
; GFX11-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call float @llvm.amdgcn.image.sample.c.d.o.2darray.f32.f16.f16(i32 4, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call float @llvm.amdgcn.image.sample.c.d.o.2darray.f32.f16.f16(i32 4, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
ret float %v ret float %v
} }
define amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) { define amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) {
Show All 34 Linesdefine amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) {
; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX9-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX9-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GFX9-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GFX9-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX9-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX9-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GFX9-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
; GFX10-LABEL: name: sample_c_d_o_2darray_V2 ; GFX10-LABEL: name: sample_c_d_o_2darray_V2
; GFX10: bb.1.main_body: ; GFX10: bb.1.main_body:
; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8 ; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
; GFX10-NEXT: {{ $}} ; GFX10-NEXT: {{ $}}
Show All 30 Linesdefine amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) {
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
; GFX11-LABEL: name: sample_c_d_o_2darray_V2 ; GFX11-LABEL: name: sample_c_d_o_2darray_V2
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
; GFX11-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8 ; GFX11-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
Show All 30 Linesdefine amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice) {
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
main_body: main_body:
%v = call <2 x float> @llvm.amdgcn.image.sample.c.d.o.2darray.v2f32.f32.f16(i32 6, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <2 x float> @llvm.amdgcn.image.sample.c.d.o.2darray.v2f32.f32.f16(i32 6, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, half %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
ret <2 x float> %v ret <2 x float> %v
} }
▲ Show 20 Lines • Show All 60 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.sample.d.ll

Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r) {
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX10-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8 ; GFX10-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<9 x s32>) = G_BUILD_VECTOR [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<9 x s32>) = G_BUILD_VECTOR [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[BUILD_VECTOR2]](<9 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[BUILD_VECTOR2]](<9 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_3d ; GFX11-LABEL: name: sample_d_3d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r) {
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX11-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8 ; GFX11-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<5 x s32>) = G_BUILD_VECTOR [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<5 x s32>) = G_BUILD_VECTOR [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[BUILD_VECTOR2]](<5 x s32>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[BUILD_VECTOR2]](<5 x s32>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f32.f32(i32 15, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f32.f32(i32 15, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_c_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r) {
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX10-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8 ; GFX10-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
; GFX10-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9 ; GFX10-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<10 x s32>) = G_BUILD_VECTOR [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<10 x s32>) = G_BUILD_VECTOR [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.3d), 15, [[BUILD_VECTOR2]](<10 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.3d), 15, [[BUILD_VECTOR2]](<10 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_3d ; GFX11-LABEL: name: sample_c_d_3d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r) {
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX11-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8 ; GFX11-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
; GFX11-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9 ; GFX11-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<6 x s32>) = G_BUILD_VECTOR [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<6 x s32>) = G_BUILD_VECTOR [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.3d), 15, [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[BUILD_VECTOR2]](<6 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.3d), 15, [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[BUILD_VECTOR2]](<6 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.3d.v4f32.f32.f32(i32 15, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.3d.v4f32.f32.f32(i32 15, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 26 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, float %clamp) {
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX10-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8 ; GFX10-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
; GFX10-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9 ; GFX10-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9
; GFX10-NEXT: [[COPY22:%[0-9]+]]:_(s32) = COPY $vgpr10 ; GFX10-NEXT: [[COPY22:%[0-9]+]]:_(s32) = COPY $vgpr10
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<11 x s32>) = G_BUILD_VECTOR [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32), [[COPY22]](s32) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<11 x s32>) = G_BUILD_VECTOR [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32), [[COPY22]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.3d), 15, [[BUILD_VECTOR2]](<11 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.3d), 15, [[BUILD_VECTOR2]](<11 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_cl_3d ; GFX11-LABEL: name: sample_c_d_cl_3d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 20 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, float %clamp) {
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX11-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8 ; GFX11-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
; GFX11-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9 ; GFX11-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9
; GFX11-NEXT: [[COPY22:%[0-9]+]]:_(s32) = COPY $vgpr10 ; GFX11-NEXT: [[COPY22:%[0-9]+]]:_(s32) = COPY $vgpr10
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<7 x s32>) = G_BUILD_VECTOR [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32), [[COPY22]](s32) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<7 x s32>) = G_BUILD_VECTOR [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32), [[COPY22]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.3d), 15, [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[BUILD_VECTOR2]](<7 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.3d), 15, [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[BUILD_VECTOR2]](<7 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.3d.v4f32.f32.f32(i32 15, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.3d.v4f32.f32.f32(i32 15, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_o_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, float %clamp) {
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX10-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8 ; GFX10-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
; GFX10-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9 ; GFX10-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9
; GFX10-NEXT: [[COPY22:%[0-9]+]]:_(s32) = COPY $vgpr10 ; GFX10-NEXT: [[COPY22:%[0-9]+]]:_(s32) = COPY $vgpr10
; GFX10-NEXT: [[COPY23:%[0-9]+]]:_(s32) = COPY $vgpr11 ; GFX10-NEXT: [[COPY23:%[0-9]+]]:_(s32) = COPY $vgpr11
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<12 x s32>) = G_BUILD_VECTOR [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32), [[COPY22]](s32), [[COPY23]](s32) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<12 x s32>) = G_BUILD_VECTOR [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32), [[COPY22]](s32), [[COPY23]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.o.3d), 15, [[BUILD_VECTOR2]](<12 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.o.3d), 15, [[BUILD_VECTOR2]](<12 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_cl_o_3d ; GFX11-LABEL: name: sample_c_d_cl_o_3d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 21 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_o_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, float %clamp) {
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX11-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8 ; GFX11-NEXT: [[COPY20:%[0-9]+]]:_(s32) = COPY $vgpr8
; GFX11-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9 ; GFX11-NEXT: [[COPY21:%[0-9]+]]:_(s32) = COPY $vgpr9
; GFX11-NEXT: [[COPY22:%[0-9]+]]:_(s32) = COPY $vgpr10 ; GFX11-NEXT: [[COPY22:%[0-9]+]]:_(s32) = COPY $vgpr10
; GFX11-NEXT: [[COPY23:%[0-9]+]]:_(s32) = COPY $vgpr11 ; GFX11-NEXT: [[COPY23:%[0-9]+]]:_(s32) = COPY $vgpr11
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32), [[COPY22]](s32), [[COPY23]](s32) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY16]](s32), [[COPY17]](s32), [[COPY18]](s32), [[COPY19]](s32), [[COPY20]](s32), [[COPY21]](s32), [[COPY22]](s32), [[COPY23]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.o.3d), 15, [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[BUILD_VECTOR2]](<8 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.o.3d), 15, [[COPY12]](s32), [[COPY13]](s32), [[COPY14]](s32), [[COPY15]](s32), [[BUILD_VECTOR2]](<8 x s32>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.o.3d.v4f32.f32.f32(i32 15, i32 %offset, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.o.3d.v4f32.f32.f32(i32 15, i32 %offset, float %zcompare, float %dsdh, float %dtdh, float %drdh, float %dsdv, float %dtdv, float %drdv, float %s, float %t, float %r, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
ret <4 x float> %v ret <4 x float> %v
} }
declare <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f32.f32(i32, float, float, float, float, float, float, float, float, float, <8 x i32>, <4 x i32>, i1, i32, i32) declare <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f32.f32(i32, float, float, float, float, float, float, float, float, float, <8 x i32>, <4 x i32>, i1, i32, i32)
declare <4 x float> @llvm.amdgcn.image.sample.c.d.3d.v4f32.f32.f32(i32, float, float, float, float, float, float, float, float, float, float, <8 x i32>, <4 x i32>, i1, i32, i32) declare <4 x float> @llvm.amdgcn.image.sample.c.d.3d.v4f32.f32.f32(i32, float, float, float, float, float, float, float, float, float, float, <8 x i32>, <4 x i32>, i1, i32, i32)
declare <4 x float> @llvm.amdgcn.image.sample.c.d.cl.3d.v4f32.f32.f32(i32, float, float, float, float, float, float, float, float, float, float, float, <8 x i32>, <4 x i32>, i1, i32, i32) declare <4 x float> @llvm.amdgcn.image.sample.c.d.cl.3d.v4f32.f32.f32(i32, float, float, float, float, float, float, float, float, float, float, float, <8 x i32>, <4 x i32>, i1, i32, i32)
declare <4 x float> @llvm.amdgcn.image.sample.c.d.cl.o.3d.v4f32.f32.f32(i32, i32, float, float, float, float, float, float, float, float, float, float, float, <8 x i32>, <4 x i32>, i1, i32, i32) declare <4 x float> @llvm.amdgcn.image.sample.c.d.cl.o.3d.v4f32.f32.f32(i32, i32, float, float, float, float, float, float, float, float, float, float, float, <8 x i32>, <4 x i32>, i1, i32, i32)

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.sample.g16.a16.ll

Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_d_1d_g16_a16(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) {
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_1d_g16_a16 ; GFX11-LABEL: name: sample_d_1d_g16_a16
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 18 Linesdefine amdgpu_ps <4 x float> @sample_d_1d_g16_a16(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, half %s) {
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f16(i32 15, half %dsdh, half %dsdv, half %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 29 Linesdefine amdgpu_ps <4 x float> @sample_d_2d_g16_a16(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX10-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX10-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_2d_g16_a16 ; GFX11-LABEL: name: sample_d_2d_g16_a16
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 23 Linesdefine amdgpu_ps <4 x float> @sample_d_2d_g16_a16(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t) {
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX11-NEXT: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32) ; GFX11-NEXT: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[COPY17]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC4]](s16), [[TRUNC5]](s16)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, half %s, half %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 40 Linesdefine amdgpu_ps <4 x float> @sample_d_3d_g16_a16(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r) {
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16) ; GFX10-NEXT: [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16)
; GFX10-NEXT: [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_3d_g16_a16 ; GFX11-LABEL: name: sample_d_3d_g16_a16
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 34 Linesdefine amdgpu_ps <4 x float> @sample_d_3d_g16_a16(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r) {
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16) ; GFX11-NEXT: [[BUILD_VECTOR6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC6]](s16), [[TRUNC7]](s16)
; GFX11-NEXT: [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC8]](s16), [[DEF]](s16)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR6]](<2 x s16>), [[BUILD_VECTOR7]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 3 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f16(i32 15, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, half %s, half %t, half %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
ret <4 x float> %v ret <4 x float> %v
} }
declare <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) declare <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f16(i32, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32)
declare <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f16(i32, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) declare <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f16(i32, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32)
declare <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f16(i32, half, half, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32) declare <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f16(i32, half, half, half, half, half, half, half, half, half, <8 x i32>, <4 x i32>, i1, i32, i32)

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.sample.g16.ll

Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, float %s) {
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_1d ; GFX11-LABEL: name: sample_d_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 17 Linesdefine amdgpu_ps <4 x float> @sample_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, float %s) {
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f32(i32 15, half %dsdh, half %dsdv, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f32(i32 15, half %dsdh, half %dsdv, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 28 Linesdefine amdgpu_ps <4 x float> @sample_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t) {
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_2d ; GFX11-LABEL: name: sample_d_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t) {
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f32(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.2d.v4f32.f16.f32(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 38 Linesdefine amdgpu_ps <4 x float> @sample_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, float %s, float %t, float %r) {
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16) ; GFX10-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<14 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<14 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[CONCAT_VECTORS]](<14 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[CONCAT_VECTORS]](<14 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_3d ; GFX11-LABEL: name: sample_d_3d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 32 Linesdefine amdgpu_ps <4 x float> @sample_d_3d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, float %s, float %t, float %r) {
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16) ; GFX11-NEXT: [[BUILD_VECTOR4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC3]](s16), [[TRUNC4]](s16)
; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC5]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.3d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BUILD_VECTOR4]](<2 x s16>), [[BUILD_VECTOR5]](<2 x s16>), [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f32(i32 15, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, float %s, float %t, float %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.3d.v4f32.f16.f32(i32 15, half %dsdh, half %dtdh, half %drdh, half %dsdv, half %dtdv, half %drdv, float %s, float %t, float %r, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_c_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, float %s) {
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_1d ; GFX11-LABEL: name: sample_c_d_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, float %s) {
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.1d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dsdv, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.1d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dsdv, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 30 Linesdefine amdgpu_ps <4 x float> @sample_c_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t) {
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_2d ; GFX11-LABEL: name: sample_c_d_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_c_d_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t) {
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.2d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.2d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, float %s, float %clamp) {
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_cl_1d ; GFX11-LABEL: name: sample_d_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, float %s, float %clamp) {
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.1d.v4f32.f16.f32(i32 15, half %dsdh, half %dsdv, float %s, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.1d.v4f32.f16.f32(i32 15, half %dsdh, half %dsdv, float %s, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 30 Linesdefine amdgpu_ps <4 x float> @sample_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp) {
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_d_cl_2d ; GFX11-LABEL: name: sample_d_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp) {
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.2d.v4f32.f16.f32(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.cl.2d.v4f32.f16.f32(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, float %s, float %clamp) {
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_cl_1d ; GFX11-LABEL: name: sample_c_d_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 21 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, float %s, float %clamp) {
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.1d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dsdv, float %s, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.1d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dsdv, float %s, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 33 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp) {
; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX10-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32) ; GFX10-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_d_cl_2d ; GFX11-LABEL: name: sample_c_d_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_c_d_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp) {
; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX11-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32) ; GFX11-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.2d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.d.cl.2d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 23 Linesdefine amdgpu_ps <4 x float> @sample_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, float %s) {
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cd_1d ; GFX11-LABEL: name: sample_cd_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 17 Linesdefine amdgpu_ps <4 x float> @sample_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, float %s) {
; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32) ; GFX11-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY12]](s32)
; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX11-NEXT: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cd.1d.v4f32.f16.f32(i32 15, half %dsdh, half %dsdv, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cd.1d.v4f32.f16.f32(i32 15, half %dsdh, half %dsdv, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 28 Linesdefine amdgpu_ps <4 x float> @sample_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t) {
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cd_2d ; GFX11-LABEL: name: sample_cd_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 22 Linesdefine amdgpu_ps <4 x float> @sample_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t) {
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY15]](s32)
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cd.2d.v4f32.f16.f32(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cd.2d.v4f32.f16.f32(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, float %s) {
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_cd_1d ; GFX11-LABEL: name: sample_c_cd_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, float %s) {
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY14]](s32)
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.1d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dsdv, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.1d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dsdv, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 30 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t) {
; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX10-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_cd_2d ; GFX11-LABEL: name: sample_c_cd_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t) {
; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32) ; GFX11-NEXT: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[COPY16]](s32)
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.2d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.2d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 25 Linesdefine amdgpu_ps <4 x float> @sample_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, float %s, float %clamp) {
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cd_cl_1d ; GFX11-LABEL: name: sample_cd_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 19 Linesdefine amdgpu_ps <4 x float> @sample_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, float %s, float %clamp) {
; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32) ; GFX11-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY13]](s32)
; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX11-NEXT: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY14]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.1d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.1d.v4f32.f16.f32(i32 15, half %dsdh, half %dsdv, float %s, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.1d.v4f32.f16.f32(i32 15, half %dsdh, half %dsdv, float %s, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 30 Linesdefine amdgpu_ps <4 x float> @sample_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp) {
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX10-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX10-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_cd_cl_2d ; GFX11-LABEL: name: sample_cd_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 24 Linesdefine amdgpu_ps <4 x float> @sample_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp) {
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5 ; GFX11-NEXT: [[COPY17:%[0-9]+]]:_(s32) = COPY $vgpr5
; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6 ; GFX11-NEXT: [[COPY18:%[0-9]+]]:_(s32) = COPY $vgpr6
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.cd.cl.2d), 15, [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.2d.v4f32.f16.f32(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.cd.cl.2d.v4f32.f16.f32(i32 15, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, float %s, float %clamp) {
; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX10-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX10-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_cd_cl_1d ; GFX11-LABEL: name: sample_c_cd_cl_1d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 21 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_cl_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dsdv, float %s, float %clamp) {
; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3 ; GFX11-NEXT: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr3
; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4 ; GFX11-NEXT: [[COPY16:%[0-9]+]]:_(s32) = COPY $vgpr4
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX11-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC1]](s16), [[DEF]](s16)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY15]](s32)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY16]](s32)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.1d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.1d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dsdv, float %s, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.1d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dsdv, float %s, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 33 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp) {
; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX10-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX10-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32) ; GFX10-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<12 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[CONCAT_VECTORS]](<12 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX10-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX10-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
; GFX11-LABEL: name: sample_c_cd_cl_2d ; GFX11-LABEL: name: sample_c_cd_cl_2d
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
Show All 27 Linesdefine amdgpu_ps <4 x float> @sample_c_cd_cl_2d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp) {
; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7 ; GFX11-NEXT: [[COPY19:%[0-9]+]]:_(s32) = COPY $vgpr7
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY17]](s32)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX11-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32) ; GFX11-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.cd.cl.2d), 15, [[BITCAST]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[CONCAT_VECTORS]](<4 x s16>), $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32) ; GFX11-NEXT: $vgpr2 = COPY [[UV2]](s32)
; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32) ; GFX11-NEXT: $vgpr3 = COPY [[UV3]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2, implicit $vgpr3
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.2d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.c.cd.cl.2d.v4f32.f16.f32(i32 15, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %clamp, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
Show All 35 Linesdefine amdgpu_ps float @sample_c_d_o_2darray_V1(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %slice) {
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX10-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32) ; GFX10-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32)
; GFX10-NEXT: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32) ; GFX10-NEXT: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<14 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<14 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[CONCAT_VECTORS]](<14 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (s32), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[CONCAT_VECTORS]](<14 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (s32), addrspace 8)
; GFX10-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX11-LABEL: name: sample_c_d_o_2darray_V1 ; GFX11-LABEL: name: sample_c_d_o_2darray_V1
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
; GFX11-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8 ; GFX11-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX11-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX11-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
Show All 25 Linesdefine amdgpu_ps float @sample_c_d_o_2darray_V1(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %slice) {
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX11-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32) ; GFX11-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32)
; GFX11-NEXT: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32) ; GFX11-NEXT: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (s32), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 4, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (s32), addrspace 8)
; GFX11-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[AMDGPU_INTRIN_IMAGE_LOAD]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
main_body: main_body:
%v = call float @llvm.amdgcn.image.sample.c.d.o.2darray.f16.f32.f32(i32 4, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call float @llvm.amdgcn.image.sample.c.d.o.2darray.f16.f32.f32(i32 4, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
ret float %v ret float %v
} }
define amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %slice) { define amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %slice) {
Show All 31 Linesdefine amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %slice) {
; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX10-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX10-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX10-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32) ; GFX10-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32)
; GFX10-NEXT: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32) ; GFX10-NEXT: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<14 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<14 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>)
; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[CONCAT_VECTORS]](<14 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GFX10-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[CONCAT_VECTORS]](<14 x s16>), $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GFX10-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX10-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX10-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
; GFX11-LABEL: name: sample_c_d_o_2darray_V2 ; GFX11-LABEL: name: sample_c_d_o_2darray_V2
; GFX11: bb.1.main_body: ; GFX11: bb.1.main_body:
; GFX11-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8 ; GFX11-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
Show All 27 Linesdefine amdgpu_ps <2 x float> @sample_c_d_o_2darray_V2(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %slice) {
; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32) ; GFX11-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY12]](s32)
; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32) ; GFX11-NEXT: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY13]](s32)
; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX11-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16) ; GFX11-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[TRUNC3]](s16)
; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32) ; GFX11-NEXT: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY18]](s32)
; GFX11-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32) ; GFX11-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY19]](s32)
; GFX11-NEXT: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32) ; GFX11-NEXT: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[COPY20]](s32)
; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>) ; GFX11-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>)
; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<2 x s32>), addrspace 7) ; GFX11-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:_(<2 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.c.d.o.2darray), 6, [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>), [[CONCAT_VECTORS]](<6 x s16>), $noreg, $noreg, $noreg, $noreg, [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 2 :: (dereferenceable load (<2 x s32>), addrspace 8)
; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>) ; GFX11-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[AMDGPU_INTRIN_IMAGE_LOAD]](<2 x s32>)
; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32) ; GFX11-NEXT: $vgpr0 = COPY [[UV]](s32)
; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32) ; GFX11-NEXT: $vgpr1 = COPY [[UV1]](s32)
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
main_body: main_body:
%v = call <2 x float> @llvm.amdgcn.image.sample.c.d.o.2darray.v2f32.f16.f32(i32 6, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <2 x float> @llvm.amdgcn.image.sample.c.d.o.2darray.v2f32.f16.f32(i32 6, i32 %offset, float %zcompare, half %dsdh, half %dtdh, half %dsdv, half %dtdv, float %s, float %t, float %slice, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
ret <2 x float> %v ret <2 x float> %v
} }
Show All 26 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-llvm.amdgcn.image.store.2d.d16.ll

Show All 18 Linesdefine amdgpu_ps void @image_store_f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, half %data) {
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; UNPACKED-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; UNPACKED-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[TRUNC]](s16), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (s16), addrspace 7) ; UNPACKED-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[TRUNC]](s16), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (s16), addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; GFX81-LABEL: name: image_store_f16 ; GFX81-LABEL: name: image_store_f16
; GFX81: bb.1 (%ir-block.0): ; GFX81: bb.1 (%ir-block.0):
; GFX81-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX81-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX81-NEXT: {{ $}} ; GFX81-NEXT: {{ $}}
; GFX81-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX81-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX81-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX81-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX81-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX81-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX81-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX81-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX81-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX81-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX81-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX81-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX81-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX81-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX81-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX81-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX81-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX81-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX81-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX81-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX81-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX81-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX81-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX81-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX81-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX81-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX81-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX81-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX81-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[TRUNC]](s16), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (s16), addrspace 7) ; GFX81-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[TRUNC]](s16), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (s16), addrspace 8)
; GFX81-NEXT: S_ENDPGM 0 ; GFX81-NEXT: S_ENDPGM 0
; GFX9-LABEL: name: image_store_f16 ; GFX9-LABEL: name: image_store_f16
; GFX9: bb.1 (%ir-block.0): ; GFX9: bb.1 (%ir-block.0):
; GFX9-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX9-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX9-NEXT: {{ $}} ; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX9-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX9-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX9-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX9-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX9-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[TRUNC]](s16), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (s16), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[TRUNC]](s16), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (s16), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10-LABEL: name: image_store_f16 ; GFX10-LABEL: name: image_store_f16
; GFX10: bb.1 (%ir-block.0): ; GFX10: bb.1 (%ir-block.0):
; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX10-NEXT: {{ $}} ; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2 ; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr2
; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32) ; GFX10-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY10]](s32)
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX10-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[TRUNC]](s16), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (s16), addrspace 7) ; GFX10-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[TRUNC]](s16), 1, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (s16), addrspace 8)
; GFX10-NEXT: S_ENDPGM 0 ; GFX10-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.image.store.2d.f16.i32(half %data, i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.2d.f16.i32(half %data, i32 1, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
define amdgpu_ps void @image_store_v2f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <2 x half> %in) { define amdgpu_ps void @image_store_v2f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <2 x half> %in) {
; UNPACKED-LABEL: name: image_store_v2f16 ; UNPACKED-LABEL: name: image_store_v2f16
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
Show All 11 Linesdefine amdgpu_ps void @image_store_v2f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <2 x half> %in) {
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2 ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; UNPACKED-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; UNPACKED-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[BITCAST]](s32), [[LSHR]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[BITCAST]](s32), [[LSHR]](s32)
; UNPACKED-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR2]](<2 x s32>), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<2 x s16>), addrspace 7) ; UNPACKED-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR2]](<2 x s32>), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<2 x s16>), addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; GFX81-LABEL: name: image_store_v2f16 ; GFX81-LABEL: name: image_store_v2f16
; GFX81: bb.1 (%ir-block.0): ; GFX81: bb.1 (%ir-block.0):
; GFX81-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX81-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX81-NEXT: {{ $}} ; GFX81-NEXT: {{ $}}
; GFX81-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX81-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX81-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX81-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX81-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX81-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX81-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX81-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX81-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX81-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX81-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX81-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX81-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX81-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX81-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX81-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX81-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX81-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX81-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX81-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX81-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX81-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX81-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2 ; GFX81-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2
; GFX81-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX81-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX81-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>) ; GFX81-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>)
; GFX81-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GFX81-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GFX81-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[BITCAST]](s32), [[DEF]](s32) ; GFX81-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[BITCAST]](s32), [[DEF]](s32)
; GFX81-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR2]](<2 x s32>), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<2 x s16>), addrspace 7) ; GFX81-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR2]](<2 x s32>), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<2 x s16>), addrspace 8)
; GFX81-NEXT: S_ENDPGM 0 ; GFX81-NEXT: S_ENDPGM 0
; GFX9-LABEL: name: image_store_v2f16 ; GFX9-LABEL: name: image_store_v2f16
; GFX9: bb.1 (%ir-block.0): ; GFX9: bb.1 (%ir-block.0):
; GFX9-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX9-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX9-NEXT: {{ $}} ; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX9-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX9-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX9-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX9-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[COPY10]](<2 x s16>), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<2 x s16>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[COPY10]](<2 x s16>), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<2 x s16>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10-LABEL: name: image_store_v2f16 ; GFX10-LABEL: name: image_store_v2f16
; GFX10: bb.1 (%ir-block.0): ; GFX10: bb.1 (%ir-block.0):
; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2 ; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2
; GFX10-NEXT: {{ $}} ; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2 ; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX10-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[COPY10]](<2 x s16>), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<2 x s16>), addrspace 7) ; GFX10-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[COPY10]](<2 x s16>), 3, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<2 x s16>), addrspace 8)
; GFX10-NEXT: S_ENDPGM 0 ; GFX10-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.image.store.2d.v2f16.i32(<2 x half> %in, i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.2d.v2f16.i32(<2 x half> %in, i32 3, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
define amdgpu_ps void @image_store_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <3 x half> %in) { define amdgpu_ps void @image_store_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <3 x half> %in) {
; UNPACKED-LABEL: name: image_store_v3f16 ; UNPACKED-LABEL: name: image_store_v3f16
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
Show All 13 Linesdefine amdgpu_ps void @image_store_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <3 x half> %in) {
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2 ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2
; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr3 ; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr3
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; UNPACKED-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; UNPACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY11]](<2 x s16>) ; UNPACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY11]](<2 x s16>)
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[BITCAST]](s32), [[LSHR]](s32), [[BITCAST1]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[BITCAST]](s32), [[LSHR]](s32), [[BITCAST1]](s32)
; UNPACKED-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR2]](<3 x s32>), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<3 x s16>), align 8, addrspace 7) ; UNPACKED-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR2]](<3 x s32>), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<3 x s16>), align 8, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; GFX81-LABEL: name: image_store_v3f16 ; GFX81-LABEL: name: image_store_v3f16
; GFX81: bb.1 (%ir-block.0): ; GFX81: bb.1 (%ir-block.0):
; GFX81-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX81-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX81-NEXT: {{ $}} ; GFX81-NEXT: {{ $}}
; GFX81-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX81-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX81-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX81-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX81-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX81-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 22 Linesdefine amdgpu_ps void @image_store_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <3 x half> %in) {
; GFX81-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 0 ; GFX81-NEXT: [[C2:%[0-9]+]]:_(s32) = G_CONSTANT i32 0
; GFX81-NEXT: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[C2]], [[C]](s32) ; GFX81-NEXT: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[C2]], [[C]](s32)
; GFX81-NEXT: [[OR1:%[0-9]+]]:_(s32) = G_OR [[AND2]], [[SHL1]] ; GFX81-NEXT: [[OR1:%[0-9]+]]:_(s32) = G_OR [[AND2]], [[SHL1]]
; GFX81-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32) ; GFX81-NEXT: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
; GFX81-NEXT: [[OR2:%[0-9]+]]:_(s32) = G_OR [[C2]], [[SHL1]] ; GFX81-NEXT: [[OR2:%[0-9]+]]:_(s32) = G_OR [[C2]], [[SHL1]]
; GFX81-NEXT: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR2]](s32) ; GFX81-NEXT: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR2]](s32)
; GFX81-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>) ; GFX81-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>)
; GFX81-NEXT: [[BITCAST5:%[0-9]+]]:_(<3 x s32>) = G_BITCAST [[CONCAT_VECTORS]](<6 x s16>) ; GFX81-NEXT: [[BITCAST5:%[0-9]+]]:_(<3 x s32>) = G_BITCAST [[CONCAT_VECTORS]](<6 x s16>)
; GFX81-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BITCAST5]](<3 x s32>), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<3 x s16>), align 8, addrspace 7) ; GFX81-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BITCAST5]](<3 x s32>), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<3 x s16>), align 8, addrspace 8)
; GFX81-NEXT: S_ENDPGM 0 ; GFX81-NEXT: S_ENDPGM 0
; GFX9-LABEL: name: image_store_v3f16 ; GFX9-LABEL: name: image_store_v3f16
; GFX9: bb.1 (%ir-block.0): ; GFX9: bb.1 (%ir-block.0):
; GFX9-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX9-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX9-NEXT: {{ $}} ; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX9-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX9-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 14 Linesdefine amdgpu_ps void @image_store_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <3 x half> %in) {
; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX9-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY11]](<2 x s16>) ; GFX9-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY11]](<2 x s16>)
; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32) ; GFX9-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX9-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX9-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX9-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[CONCAT_VECTORS]](<4 x s16>), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<3 x s16>), align 8, addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[CONCAT_VECTORS]](<4 x s16>), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<3 x s16>), align 8, addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10-LABEL: name: image_store_v3f16 ; GFX10-LABEL: name: image_store_v3f16
; GFX10: bb.1 (%ir-block.0): ; GFX10: bb.1 (%ir-block.0):
; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX10-NEXT: {{ $}} ; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
Show All 14 Linesdefine amdgpu_ps void @image_store_v3f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <3 x half> %in) {
; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32) ; GFX10-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY11]](<2 x s16>) ; GFX10-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY11]](<2 x s16>)
; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32) ; GFX10-NEXT: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF ; GFX10-NEXT: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16) ; GFX10-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC]](s16), [[TRUNC1]](s16)
; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16) ; GFX10-NEXT: [[BUILD_VECTOR3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR [[TRUNC2]](s16), [[DEF]](s16)
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR2]](<2 x s16>), [[BUILD_VECTOR3]](<2 x s16>)
; GFX10-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[CONCAT_VECTORS]](<4 x s16>), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<3 x s16>), align 8, addrspace 7) ; GFX10-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[CONCAT_VECTORS]](<4 x s16>), 7, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<3 x s16>), align 8, addrspace 8)
; GFX10-NEXT: S_ENDPGM 0 ; GFX10-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.image.store.2d.v3f16.i32(<3 x half> %in, i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.2d.v3f16.i32(<3 x half> %in, i32 7, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
define amdgpu_ps void @image_store_v4f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <4 x half> %in) { define amdgpu_ps void @image_store_v4f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <4 x half> %in) {
; UNPACKED-LABEL: name: image_store_v4f16 ; UNPACKED-LABEL: name: image_store_v4f16
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
Show All 14 Linesdefine amdgpu_ps void @image_store_v4f16(<8 x i32> inreg %rsrc, i32 %s, i32 %t, <4 x half> %in) {
; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr3 ; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr3
; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>) ; UNPACKED-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY10]](<2 x s16>)
; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16 ; UNPACKED-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
; UNPACKED-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32) ; UNPACKED-NEXT: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
; UNPACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY11]](<2 x s16>) ; UNPACKED-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[COPY11]](<2 x s16>)
; UNPACKED-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32) ; UNPACKED-NEXT: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
; UNPACKED-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[BITCAST]](s32), [[LSHR]](s32), [[BITCAST1]](s32), [[LSHR1]](s32) ; UNPACKED-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[BITCAST]](s32), [[LSHR]](s32), [[BITCAST1]](s32), [[LSHR1]](s32)
; UNPACKED-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR2]](<4 x s32>), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<4 x s16>), addrspace 7) ; UNPACKED-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR2]](<4 x s32>), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<4 x s16>), addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; GFX81-LABEL: name: image_store_v4f16 ; GFX81-LABEL: name: image_store_v4f16
; GFX81: bb.1 (%ir-block.0): ; GFX81: bb.1 (%ir-block.0):
; GFX81-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX81-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX81-NEXT: {{ $}} ; GFX81-NEXT: {{ $}}
; GFX81-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX81-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX81-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX81-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX81-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX81-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX81-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX81-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX81-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX81-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX81-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX81-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX81-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX81-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX81-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX81-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX81-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX81-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX81-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX81-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX81-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX81-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX81-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2 ; GFX81-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2
; GFX81-NEXT: [[COPY11:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr3 ; GFX81-NEXT: [[COPY11:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr3
; GFX81-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[COPY10]](<2 x s16>), [[COPY11]](<2 x s16>) ; GFX81-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[COPY10]](<2 x s16>), [[COPY11]](<2 x s16>)
; GFX81-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX81-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX81-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s32>) = G_BITCAST [[CONCAT_VECTORS]](<4 x s16>) ; GFX81-NEXT: [[BITCAST:%[0-9]+]]:_(<2 x s32>) = G_BITCAST [[CONCAT_VECTORS]](<4 x s16>)
; GFX81-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[BITCAST]](<2 x s32>) ; GFX81-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[BITCAST]](<2 x s32>)
; GFX81-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF ; GFX81-NEXT: [[DEF:%[0-9]+]]:_(s32) = G_IMPLICIT_DEF
; GFX81-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[UV]](s32), [[UV1]](s32), [[DEF]](s32), [[DEF]](s32) ; GFX81-NEXT: [[BUILD_VECTOR2:%[0-9]+]]:_(<4 x s32>) = G_BUILD_VECTOR [[UV]](s32), [[UV1]](s32), [[DEF]](s32), [[DEF]](s32)
; GFX81-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR2]](<4 x s32>), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<4 x s16>), addrspace 7) ; GFX81-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[BUILD_VECTOR2]](<4 x s32>), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<4 x s16>), addrspace 8)
; GFX81-NEXT: S_ENDPGM 0 ; GFX81-NEXT: S_ENDPGM 0
; GFX9-LABEL: name: image_store_v4f16 ; GFX9-LABEL: name: image_store_v4f16
; GFX9: bb.1 (%ir-block.0): ; GFX9: bb.1 (%ir-block.0):
; GFX9-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX9-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX9-NEXT: {{ $}} ; GFX9-NEXT: {{ $}}
; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX9-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX9-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX9-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX9-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX9-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX9-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX9-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX9-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX9-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX9-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX9-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2 ; GFX9-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2
; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr3 ; GFX9-NEXT: [[COPY11:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr3
; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[COPY10]](<2 x s16>), [[COPY11]](<2 x s16>) ; GFX9-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[COPY10]](<2 x s16>), [[COPY11]](<2 x s16>)
; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX9-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[CONCAT_VECTORS]](<4 x s16>), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<4 x s16>), addrspace 7) ; GFX9-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[CONCAT_VECTORS]](<4 x s16>), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<4 x s16>), addrspace 8)
; GFX9-NEXT: S_ENDPGM 0 ; GFX9-NEXT: S_ENDPGM 0
; GFX10-LABEL: name: image_store_v4f16 ; GFX10-LABEL: name: image_store_v4f16
; GFX10: bb.1 (%ir-block.0): ; GFX10: bb.1 (%ir-block.0):
; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; GFX10-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; GFX10-NEXT: {{ $}} ; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2 ; GFX10-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $sgpr2
; GFX10-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3 ; GFX10-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $sgpr3
; GFX10-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4 ; GFX10-NEXT: [[COPY2:%[0-9]+]]:_(s32) = COPY $sgpr4
; GFX10-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5 ; GFX10-NEXT: [[COPY3:%[0-9]+]]:_(s32) = COPY $sgpr5
; GFX10-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6 ; GFX10-NEXT: [[COPY4:%[0-9]+]]:_(s32) = COPY $sgpr6
; GFX10-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7 ; GFX10-NEXT: [[COPY5:%[0-9]+]]:_(s32) = COPY $sgpr7
; GFX10-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8 ; GFX10-NEXT: [[COPY6:%[0-9]+]]:_(s32) = COPY $sgpr8
; GFX10-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9 ; GFX10-NEXT: [[COPY7:%[0-9]+]]:_(s32) = COPY $sgpr9
; GFX10-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GFX10-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GFX10-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0 ; GFX10-NEXT: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr0
; GFX10-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1 ; GFX10-NEXT: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2 ; GFX10-NEXT: [[COPY10:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr2
; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr3 ; GFX10-NEXT: [[COPY11:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr3
; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[COPY10]](<2 x s16>), [[COPY11]](<2 x s16>) ; GFX10-NEXT: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[COPY10]](<2 x s16>), [[COPY11]](<2 x s16>)
; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32) ; GFX10-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32)
; GFX10-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[CONCAT_VECTORS]](<4 x s16>), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<4 x s16>), addrspace 7) ; GFX10-NEXT: G_AMDGPU_INTRIN_IMAGE_STORE_D16 intrinsic(@llvm.amdgcn.image.store.2d), [[CONCAT_VECTORS]](<4 x s16>), 15, [[BUILD_VECTOR1]](<2 x s32>), $noreg, [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable store (<4 x s16>), addrspace 8)
; GFX10-NEXT: S_ENDPGM 0 ; GFX10-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.image.store.2d.v4f16.i32(<4 x half> %in, i32 15, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0) call void @llvm.amdgcn.image.store.2d.v4f16.i32(<4 x half> %in, i32 15, i32 %s, i32 %t, <8 x i32> %rsrc, i32 0, i32 0)
ret void ret void
} }
declare void @llvm.amdgcn.image.store.2d.f16.i32(half, i32 immarg, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0 declare void @llvm.amdgcn.image.store.2d.f16.i32(half, i32 immarg, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0
declare void @llvm.amdgcn.image.store.2d.v2f16.i32(<2 x half>, i32 immarg, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0 declare void @llvm.amdgcn.image.store.2d.v2f16.i32(<2 x half>, i32 immarg, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0
declare void @llvm.amdgcn.image.store.2d.v3f16.i32(<3 x half>, i32 immarg, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0 declare void @llvm.amdgcn.image.store.2d.v3f16.i32(<3 x half>, i32 immarg, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0
declare void @llvm.amdgcn.image.store.2d.v4f16.i32(<4 x half>, i32 immarg, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0 declare void @llvm.amdgcn.image.store.2d.v4f16.i32(<4 x half>, i32 immarg, i32, i32, <8 x i32>, i32 immarg, i32 immarg) #0
attributes #0 = { nounwind writeonly } attributes #0 = { nounwind writeonly }

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.image.atomic.dim.mir

Show All 14 Lines bb.0:
liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2 liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2
; GFX6-LABEL: name: atomic_cmpswap_i32_1d ; GFX6-LABEL: name: atomic_cmpswap_i32_1d
; GFX6: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2 ; GFX6: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2
; GFX6-NEXT: {{ $}} ; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX6-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX6-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 ; GFX6-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
; GFX6-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX6-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX6-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_si:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_si [[COPY1]], [[COPY2]], [[COPY]], 3, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX6-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_si:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_si [[COPY1]], [[COPY2]], [[COPY]], 3, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX6-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V1_V1_si]].sub0 ; GFX6-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V1_V1_si]].sub0
; GFX6-NEXT: $vgpr0 = COPY [[COPY3]] ; GFX6-NEXT: $vgpr0 = COPY [[COPY3]]
; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX8-LABEL: name: atomic_cmpswap_i32_1d ; GFX8-LABEL: name: atomic_cmpswap_i32_1d
; GFX8: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2 ; GFX8: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2
; GFX8-NEXT: {{ $}} ; GFX8-NEXT: {{ $}}
; GFX8-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX8-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX8-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 ; GFX8-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
; GFX8-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX8-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX8-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_vi:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_vi [[COPY1]], [[COPY2]], [[COPY]], 3, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX8-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_vi:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_vi [[COPY1]], [[COPY2]], [[COPY]], 3, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX8-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V1_V1_vi]].sub0 ; GFX8-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V1_V1_vi]].sub0
; GFX8-NEXT: $vgpr0 = COPY [[COPY3]] ; GFX8-NEXT: $vgpr0 = COPY [[COPY3]]
; GFX8-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX8-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX10-LABEL: name: atomic_cmpswap_i32_1d ; GFX10-LABEL: name: atomic_cmpswap_i32_1d
; GFX10: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2 ; GFX10: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2
; GFX10-NEXT: {{ $}} ; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX10-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX10-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 ; GFX10-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
; GFX10-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX10-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX10-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx10_:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx10 [[COPY1]], [[COPY2]], [[COPY]], 3, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx10_:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx10 [[COPY1]], [[COPY2]], [[COPY]], 3, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx10_]].sub0 ; GFX10-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx10_]].sub0
; GFX10-NEXT: $vgpr0 = COPY [[COPY3]] ; GFX10-NEXT: $vgpr0 = COPY [[COPY3]]
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; GFX11-LABEL: name: atomic_cmpswap_i32_1d ; GFX11-LABEL: name: atomic_cmpswap_i32_1d
; GFX11: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2 ; GFX11: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX11-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX11-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx11_:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx11 [[COPY1]], [[COPY2]], [[COPY]], 3, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX11-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx11_:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx11 [[COPY1]], [[COPY2]], [[COPY]], 3, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX11-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx11_]].sub0 ; GFX11-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx11_]].sub0
; GFX11-NEXT: $vgpr0 = COPY [[COPY3]] ; GFX11-NEXT: $vgpr0 = COPY [[COPY3]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%0:sgpr(<8 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 %0:sgpr(<8 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
%1:vgpr(<2 x s32>) = COPY $vgpr0_vgpr1 %1:vgpr(<2 x s32>) = COPY $vgpr0_vgpr1
%2:vgpr(s32) = COPY $vgpr2 %2:vgpr(s32) = COPY $vgpr2
%3:vgpr(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), %1(<2 x s32>), $noreg, %2(s32), %0(<8 x s32>), 0, 0, 0 :: (volatile dereferenceable load store (s32), addrspace 7) %3:vgpr(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), %1(<2 x s32>), $noreg, %2(s32), %0(<8 x s32>), 0, 0, 0 :: (volatile dereferenceable load store (s32), addrspace 8)
$vgpr0 = COPY %3(s32) $vgpr0 = COPY %3(s32)
SI_RETURN_TO_EPILOG implicit $vgpr0 SI_RETURN_TO_EPILOG implicit $vgpr0
... ...
--- ---
name: atomic_cmpswap_i32_1d_no_return name: atomic_cmpswap_i32_1d_no_return
legalized: true legalized: true
regBankSelected: true regBankSelected: true
tracksRegLiveness: true tracksRegLiveness: true
body: | body: |
bb.0: bb.0:
liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2 liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2
; GFX6-LABEL: name: atomic_cmpswap_i32_1d_no_return ; GFX6-LABEL: name: atomic_cmpswap_i32_1d_no_return
; GFX6: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2 ; GFX6: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2
; GFX6-NEXT: {{ $}} ; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX6-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX6-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 ; GFX6-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
; GFX6-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX6-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX6-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_si:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_si [[COPY1]], [[COPY2]], [[COPY]], 3, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX6-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_si:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_si [[COPY1]], [[COPY2]], [[COPY]], 3, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX6-NEXT: S_ENDPGM 0 ; GFX6-NEXT: S_ENDPGM 0
; GFX8-LABEL: name: atomic_cmpswap_i32_1d_no_return ; GFX8-LABEL: name: atomic_cmpswap_i32_1d_no_return
; GFX8: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2 ; GFX8: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2
; GFX8-NEXT: {{ $}} ; GFX8-NEXT: {{ $}}
; GFX8-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX8-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX8-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 ; GFX8-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
; GFX8-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX8-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX8-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_vi:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_vi [[COPY1]], [[COPY2]], [[COPY]], 3, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX8-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_vi:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_vi [[COPY1]], [[COPY2]], [[COPY]], 3, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX8-NEXT: S_ENDPGM 0 ; GFX8-NEXT: S_ENDPGM 0
; GFX10-LABEL: name: atomic_cmpswap_i32_1d_no_return ; GFX10-LABEL: name: atomic_cmpswap_i32_1d_no_return
; GFX10: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2 ; GFX10: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2
; GFX10-NEXT: {{ $}} ; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX10-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX10-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 ; GFX10-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
; GFX10-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX10-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX10-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx10_:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx10 [[COPY1]], [[COPY2]], [[COPY]], 3, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX10-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx10_:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx10 [[COPY1]], [[COPY2]], [[COPY]], 3, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX10-NEXT: S_ENDPGM 0 ; GFX10-NEXT: S_ENDPGM 0
; GFX11-LABEL: name: atomic_cmpswap_i32_1d_no_return ; GFX11-LABEL: name: atomic_cmpswap_i32_1d_no_return
; GFX11: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2 ; GFX11: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1, $vgpr2
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX11-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX11-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx11_:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx11 [[COPY1]], [[COPY2]], [[COPY]], 3, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 7) ; GFX11-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx11_:%[0-9]+]]:vreg_64 = IMAGE_ATOMIC_CMPSWAP_V1_V1_gfx11 [[COPY1]], [[COPY2]], [[COPY]], 3, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), addrspace 8)
; GFX11-NEXT: S_ENDPGM 0 ; GFX11-NEXT: S_ENDPGM 0
%0:sgpr(<8 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 %0:sgpr(<8 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
%1:vgpr(<2 x s32>) = COPY $vgpr0_vgpr1 %1:vgpr(<2 x s32>) = COPY $vgpr0_vgpr1
%2:vgpr(s32) = COPY $vgpr2 %2:vgpr(s32) = COPY $vgpr2
%3:vgpr(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), %1(<2 x s32>), $noreg, %2(s32), %0(<8 x s32>), 0, 0, 0 :: (volatile dereferenceable load store (s32), addrspace 7) %3:vgpr(s32) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), %1(<2 x s32>), $noreg, %2(s32), %0(<8 x s32>), 0, 0, 0 :: (volatile dereferenceable load store (s32), addrspace 8)
S_ENDPGM 0 S_ENDPGM 0
... ...
--- ---
name: atomic_cmpswap_i64_1d name: atomic_cmpswap_i64_1d
legalized: true legalized: true
regBankSelected: true regBankSelected: true
tracksRegLiveness: true tracksRegLiveness: true
body: | body: |
bb.0: bb.0:
liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4 liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4
; GFX6-LABEL: name: atomic_cmpswap_i64_1d ; GFX6-LABEL: name: atomic_cmpswap_i64_1d
; GFX6: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4 ; GFX6: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4
; GFX6-NEXT: {{ $}} ; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX6-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX6-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3 ; GFX6-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3
; GFX6-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; GFX6-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; GFX6-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_si:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_si [[COPY1]], [[COPY2]], [[COPY]], 15, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 7) ; GFX6-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_si:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_si [[COPY1]], [[COPY2]], [[COPY]], 15, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 8)
; GFX6-NEXT: [[COPY3:%[0-9]+]]:vreg_64 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V2_V1_si]].sub0_sub1 ; GFX6-NEXT: [[COPY3:%[0-9]+]]:vreg_64 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V2_V1_si]].sub0_sub1
; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[COPY3]] ; GFX6-NEXT: $vgpr0_vgpr1 = COPY [[COPY3]]
; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0_vgpr1 ; GFX6-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0_vgpr1
; GFX8-LABEL: name: atomic_cmpswap_i64_1d ; GFX8-LABEL: name: atomic_cmpswap_i64_1d
; GFX8: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4 ; GFX8: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4
; GFX8-NEXT: {{ $}} ; GFX8-NEXT: {{ $}}
; GFX8-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX8-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX8-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3 ; GFX8-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3
; GFX8-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; GFX8-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; GFX8-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_vi:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_vi [[COPY1]], [[COPY2]], [[COPY]], 15, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 7) ; GFX8-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_vi:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_vi [[COPY1]], [[COPY2]], [[COPY]], 15, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 8)
; GFX8-NEXT: [[COPY3:%[0-9]+]]:vreg_64 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V2_V1_vi]].sub0_sub1 ; GFX8-NEXT: [[COPY3:%[0-9]+]]:vreg_64 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V2_V1_vi]].sub0_sub1
; GFX8-NEXT: $vgpr0_vgpr1 = COPY [[COPY3]] ; GFX8-NEXT: $vgpr0_vgpr1 = COPY [[COPY3]]
; GFX8-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0_vgpr1 ; GFX8-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0_vgpr1
; GFX10-LABEL: name: atomic_cmpswap_i64_1d ; GFX10-LABEL: name: atomic_cmpswap_i64_1d
; GFX10: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4 ; GFX10: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4
; GFX10-NEXT: {{ $}} ; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX10-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX10-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3 ; GFX10-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3
; GFX10-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; GFX10-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; GFX10-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx10_:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx10 [[COPY1]], [[COPY2]], [[COPY]], 15, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 7) ; GFX10-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx10_:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx10 [[COPY1]], [[COPY2]], [[COPY]], 15, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 8)
; GFX10-NEXT: [[COPY3:%[0-9]+]]:vreg_64 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx10_]].sub0_sub1 ; GFX10-NEXT: [[COPY3:%[0-9]+]]:vreg_64 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx10_]].sub0_sub1
; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[COPY3]] ; GFX10-NEXT: $vgpr0_vgpr1 = COPY [[COPY3]]
; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0_vgpr1 ; GFX10-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0_vgpr1
; GFX11-LABEL: name: atomic_cmpswap_i64_1d ; GFX11-LABEL: name: atomic_cmpswap_i64_1d
; GFX11: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4 ; GFX11: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX11-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3
; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; GFX11-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx11_:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx11 [[COPY1]], [[COPY2]], [[COPY]], 15, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 7) ; GFX11-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx11_:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx11 [[COPY1]], [[COPY2]], [[COPY]], 15, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 8)
; GFX11-NEXT: [[COPY3:%[0-9]+]]:vreg_64 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx11_]].sub0_sub1 ; GFX11-NEXT: [[COPY3:%[0-9]+]]:vreg_64 = COPY killed [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx11_]].sub0_sub1
; GFX11-NEXT: $vgpr0_vgpr1 = COPY [[COPY3]] ; GFX11-NEXT: $vgpr0_vgpr1 = COPY [[COPY3]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0_vgpr1 ; GFX11-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0_vgpr1
%0:sgpr(<8 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 %0:sgpr(<8 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
%1:vgpr(<2 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3 %1:vgpr(<2 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
%2:vgpr(s32) = COPY $vgpr4 %2:vgpr(s32) = COPY $vgpr4
%3:vgpr(s64) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), %1(<2 x s64>), $noreg, %2(s32), %0(<8 x s32>), 0, 0, 0 :: (volatile dereferenceable load store (s64), addrspace 7) %3:vgpr(s64) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), %1(<2 x s64>), $noreg, %2(s32), %0(<8 x s32>), 0, 0, 0 :: (volatile dereferenceable load store (s64), addrspace 8)
$vgpr0_vgpr1 = COPY %3(s64) $vgpr0_vgpr1 = COPY %3(s64)
SI_RETURN_TO_EPILOG implicit $vgpr0_vgpr1 SI_RETURN_TO_EPILOG implicit $vgpr0_vgpr1
... ...
--- ---
name: atomic_cmpswap_i64_1d_no_return name: atomic_cmpswap_i64_1d_no_return
legalized: true legalized: true
regBankSelected: true regBankSelected: true
tracksRegLiveness: true tracksRegLiveness: true
body: | body: |
bb.0: bb.0:
liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4 liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4
; GFX6-LABEL: name: atomic_cmpswap_i64_1d_no_return ; GFX6-LABEL: name: atomic_cmpswap_i64_1d_no_return
; GFX6: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4 ; GFX6: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4
; GFX6-NEXT: {{ $}} ; GFX6-NEXT: {{ $}}
; GFX6-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX6-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX6-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3 ; GFX6-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3
; GFX6-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; GFX6-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; GFX6-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_si:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_si [[COPY1]], [[COPY2]], [[COPY]], 15, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 7) ; GFX6-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_si:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_si [[COPY1]], [[COPY2]], [[COPY]], 15, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 8)
; GFX6-NEXT: S_ENDPGM 0 ; GFX6-NEXT: S_ENDPGM 0
; GFX8-LABEL: name: atomic_cmpswap_i64_1d_no_return ; GFX8-LABEL: name: atomic_cmpswap_i64_1d_no_return
; GFX8: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4 ; GFX8: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4
; GFX8-NEXT: {{ $}} ; GFX8-NEXT: {{ $}}
; GFX8-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX8-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX8-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3 ; GFX8-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3
; GFX8-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; GFX8-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; GFX8-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_vi:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_vi [[COPY1]], [[COPY2]], [[COPY]], 15, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 7) ; GFX8-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_vi:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_vi [[COPY1]], [[COPY2]], [[COPY]], 15, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 8)
; GFX8-NEXT: S_ENDPGM 0 ; GFX8-NEXT: S_ENDPGM 0
; GFX10-LABEL: name: atomic_cmpswap_i64_1d_no_return ; GFX10-LABEL: name: atomic_cmpswap_i64_1d_no_return
; GFX10: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4 ; GFX10: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4
; GFX10-NEXT: {{ $}} ; GFX10-NEXT: {{ $}}
; GFX10-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX10-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX10-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3 ; GFX10-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3
; GFX10-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; GFX10-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; GFX10-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx10_:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx10 [[COPY1]], [[COPY2]], [[COPY]], 15, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 7) ; GFX10-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx10_:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx10 [[COPY1]], [[COPY2]], [[COPY]], 15, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 8)
; GFX10-NEXT: S_ENDPGM 0 ; GFX10-NEXT: S_ENDPGM 0
; GFX11-LABEL: name: atomic_cmpswap_i64_1d_no_return ; GFX11-LABEL: name: atomic_cmpswap_i64_1d_no_return
; GFX11: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4 ; GFX11: liveins: $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 ; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_256 = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
; GFX11-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:vreg_128 = COPY $vgpr0_vgpr1_vgpr2_vgpr3
; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; GFX11-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx11_:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx11 [[COPY1]], [[COPY2]], [[COPY]], 15, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 7) ; GFX11-NEXT: [[IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx11_:%[0-9]+]]:vreg_128 = IMAGE_ATOMIC_CMPSWAP_V2_V1_gfx11 [[COPY1]], [[COPY2]], [[COPY]], 15, 0, 1, 1, 0, 0, 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), addrspace 8)
; GFX11-NEXT: S_ENDPGM 0 ; GFX11-NEXT: S_ENDPGM 0
%0:sgpr(<8 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7 %0:sgpr(<8 x s32>) = COPY $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7
%1:vgpr(<2 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3 %1:vgpr(<2 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
%2:vgpr(s32) = COPY $vgpr4 %2:vgpr(s32) = COPY $vgpr4
%3:vgpr(s64) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), %1(<2 x s64>), $noreg, %2(s32), %0(<8 x s32>), 0, 0, 0 :: (volatile dereferenceable load store (s64), addrspace 7) %3:vgpr(s64) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.atomic.cmpswap.1d), %1(<2 x s64>), $noreg, %2(s32), %0(<8 x s32>), 0, 0, 0 :: (volatile dereferenceable load store (s64), addrspace 8)
S_ENDPGM 0 S_ENDPGM 0
... ...

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.buffer.atomic.add.ll

Show All 9 Linesdefine amdgpu_ps float @raw_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_OFFEN_RTN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_OFFEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.raw.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.raw.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
define amdgpu_ps float @raw_buffer_atomic_add_i32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_atomic_add_i32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_atomic_add_i32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_atomic_add_i32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_OFFEN_RTN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_OFFEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.raw.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.raw.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
define amdgpu_ps <2 x float> @raw_buffer_atomic_add_i64__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i64 %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x float> @raw_buffer_atomic_add_i64__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i64 %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_atomic_add_i64__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_atomic_add_i64__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_X2_OFFEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_ADD_X2_OFFEN_RTN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_X2_OFFEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_ADD_X2_OFFEN_RTN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_X2_OFFEN_RTN]].sub0 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_X2_OFFEN_RTN]].sub0
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_X2_OFFEN_RTN]].sub1 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_X2_OFFEN_RTN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY9]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%ret = call i64 @llvm.amdgcn.raw.buffer.atomic.add.i64(i64 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call i64 @llvm.amdgcn.raw.buffer.atomic.add.i64(i64 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i64 %ret to <2 x float> %cast = bitcast i64 %ret to <2 x float>
ret <2 x float> %cast ret <2 x float> %cast
Show All 9 Linesdefine amdgpu_ps void @raw_buffer_atomic_add_i64_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i64 %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_ATOMIC_ADD_X2_OFFEN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_ATOMIC_ADD_X2_OFFEN [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%ret = call i64 @llvm.amdgcn.raw.buffer.atomic.add.i64(i64 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call i64 @llvm.amdgcn.raw.buffer.atomic.add.i64(i64 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; All operands need regbank legalization ; All operands need regbank legalization
define amdgpu_ps float @raw_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps float @raw_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; CHECK-LABEL: name: raw_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset ; CHECK-LABEL: name: raw_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset
Show All 31 Linesdefine amdgpu_ps float @raw_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_OFFEN_RTN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_OFFEN_RTN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
▲ Show 20 Lines • Show All 42 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @raw_buffer_atomic_add_i32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: BUFFER_ATOMIC_ADD_OFFEN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_ATOMIC_ADD_OFFEN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 11 Linesdefine amdgpu_ps float @raw_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add4095(i32 %val, <4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_OFFEN_RTN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_OFFEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
%ret = call i32 @llvm.amdgcn.raw.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.raw.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
; Natural mapping + slc ; Natural mapping + slc
define amdgpu_ps float @raw_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(i32 %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(i32 %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; CHECK-LABEL: name: raw_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 3, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_OFFEN_RTN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 3, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_OFFEN_RTN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_OFFEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.raw.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 2) %ret = call i32 @llvm.amdgcn.raw.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 2)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
declare i32 @llvm.amdgcn.raw.buffer.atomic.add.i32(i32, <4 x i32>, i32, i32, i32 immarg) #0 declare i32 @llvm.amdgcn.raw.buffer.atomic.add.i32(i32, <4 x i32>, i32, i32, i32 immarg) #0
declare i64 @llvm.amdgcn.raw.buffer.atomic.add.i64(i64, <4 x i32>, i32, i32, i32 immarg) #0 declare i64 @llvm.amdgcn.raw.buffer.atomic.add.i64(i64, <4 x i32>, i32, i32, i32 immarg) #0
attributes #0 = { nounwind } attributes #0 = { nounwind }

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.buffer.atomic.cmpswap.ll

Show All 11 Linesdefine amdgpu_ps float @raw_buffer_atomic_cmpswap_i32__vgpr_val__vgpr_cmp__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, i32 %cmp, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN]].sub0 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN]].sub0
; CHECK-NEXT: $vgpr0 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY8]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.raw.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.raw.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
; Natural mapping ; Natural mapping
define amdgpu_ps void @raw_buffer_atomic_cmpswap_i32_noret__vgpr_val__vgpr_cmp__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, i32 %cmp, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_atomic_cmpswap_i32_noret__vgpr_val__vgpr_cmp__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, i32 %cmp, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_atomic_cmpswap_i32_noret__vgpr_val__vgpr_cmp__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_atomic_cmpswap_i32_noret__vgpr_val__vgpr_cmp__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%ret = call i32 @llvm.amdgcn.raw.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.raw.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; All operands need regbank legalization ; All operands need regbank legalization
define amdgpu_ps float @raw_buffer_atomic_cmpswap_i32__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, i32 inreg %cmp, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps float @raw_buffer_atomic_cmpswap_i32__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, i32 inreg %cmp, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; CHECK-LABEL: name: raw_buffer_atomic_cmpswap_i32__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset ; CHECK-LABEL: name: raw_buffer_atomic_cmpswap_i32__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset
Show All 34 Linesdefine amdgpu_ps float @raw_buffer_atomic_cmpswap_i32__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, i32 inreg %cmp, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY9]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY9]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN [[REG_SEQUENCE2]], [[COPY10]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN [[REG_SEQUENCE2]], [[COPY10]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: [[COPY15:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN]].sub0 ; CHECK-NEXT: [[COPY15:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN]].sub0
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
▲ Show 20 Lines • Show All 46 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @raw_buffer_atomic_cmpswap_i32_noret__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, i32 inreg %cmp, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY9]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY9]], %subreg.sub1
; CHECK-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE2]], [[COPY10]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE2]], [[COPY10]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 13 Linesdefine amdgpu_ps float @raw_buffer_atomic_cmpswap_i32__vgpr_val__vgpr_cmp__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4095(i32 %val, i32 %cmp, <4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 4095, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 4095, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN]].sub0 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_OFFEN_RTN]].sub0
; CHECK-NEXT: $vgpr0 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY8]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
%ret = call i32 @llvm.amdgcn.raw.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.raw.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
Show All 19 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.buffer.atomic.fadd-with-ret.ll

; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX90A %s ; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX90A %s
; RUN: not --crash llc -global-isel < %s -march=amdgcn -mcpu=gfx908 -verify-machineinstrs 2>&1 | FileCheck %s -check-prefix=GFX908 ; RUN: not --crash llc -global-isel < %s -march=amdgcn -mcpu=gfx908 -verify-machineinstrs 2>&1 | FileCheck %s -check-prefix=GFX908
declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg) declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg)
declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg) declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg)
; GFX908: LLVM ERROR: cannot select: %{{[0-9]+}}:vgpr_32(s32) = G_AMDGPU_BUFFER_ATOMIC_FADD %{{[0-9]+}}:vgpr, %{{[0-9]+}}:sgpr(<4 x s32>), %{{[0-9]+}}:vgpr(s32), %{{[0-9]+}}:vgpr, %{{[0-9]+}}:sgpr, 0, 0, 0 :: (volatile dereferenceable load store (s32), align 1, addrspace 7) (in function: buffer_atomic_add_f32_rtn) ; GFX908: LLVM ERROR: cannot select: %{{[0-9]+}}:vgpr_32(s32) = G_AMDGPU_BUFFER_ATOMIC_FADD %{{[0-9]+}}:vgpr, %{{[0-9]+}}:sgpr(<4 x s32>), %{{[0-9]+}}:vgpr(s32), %{{[0-9]+}}:vgpr, %{{[0-9]+}}:sgpr, 0, 0, 0 :: (volatile dereferenceable load store (s32), align 1, addrspace 8) (in function: buffer_atomic_add_f32_rtn)
; GFX90A-LABEL: {{^}}buffer_atomic_add_f32_rtn: ; GFX90A-LABEL: {{^}}buffer_atomic_add_f32_rtn:
; GFX90A: buffer_atomic_add_f32 v{{[0-9]+}}, v{{[0-9]+}}, s[{{[0-9:]+}}], s{{[0-9]+}} offen glc ; GFX90A: buffer_atomic_add_f32 v{{[0-9]+}}, v{{[0-9]+}}, s[{{[0-9:]+}}], s{{[0-9]+}} offen glc
define amdgpu_kernel void @buffer_atomic_add_f32_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 %soffset) { define amdgpu_kernel void @buffer_atomic_add_f32_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 %soffset) {
main_body: main_body:
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
store float %ret, ptr undef store float %ret, ptr undef
ret void ret void
Show All 10 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.buffer.atomic.fadd.ll

Show All 10 Linesdefine amdgpu_ps void @raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX908-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset ; GFX908-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset ; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 4095 %voffset.add = add i32 %voffset, 4095
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_4095__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_4095__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX908-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_4095__sgpr_soffset ; GFX908-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_4095__sgpr_soffset
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_4095__sgpr_soffset ; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_4095__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0)
ret void ret void
} }
; Natural mapping, no voffset ; Natural mapping, no voffset
define amdgpu_ps void @raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX908-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset ; GFX908-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset ; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
; All operands need regbank legalization ; All operands need regbank legalization
define amdgpu_ps void @raw_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps void @raw_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; GFX908-LABEL: name: raw_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset ; GFX908-LABEL: name: raw_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset
Show All 31 Linesdefine amdgpu_ps void @raw_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; GFX908-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; GFX908-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; GFX908-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; GFX908-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; GFX908-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; GFX908-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; GFX908-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; GFX908-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: bb.3: ; GFX908-NEXT: bb.3:
; GFX908-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GFX908-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GFX908-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GFX908-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GFX908-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: bb.4: ; GFX908-NEXT: bb.4:
; GFX908-NEXT: successors: %bb.5(0x80000000) ; GFX908-NEXT: successors: %bb.5(0x80000000)
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GFX908-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
Show All 34 Linesdefine amdgpu_ps void @raw_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; GFX90A-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; GFX90A-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; GFX90A-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; GFX90A-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; GFX90A-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; GFX90A-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; GFX90A-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; GFX90A-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: bb.3: ; GFX90A-NEXT: bb.3:
; GFX90A-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GFX90A-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GFX90A-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GFX90A-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GFX90A-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: bb.4: ; GFX90A-NEXT: bb.4:
; GFX90A-NEXT: successors: %bb.5(0x80000000) ; GFX90A-NEXT: successors: %bb.5(0x80000000)
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GFX90A-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
Show All 38 Linesdefine amdgpu_ps void @raw_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__0_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 %soffset) {
; GFX908-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec ; GFX908-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec
; GFX908-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec ; GFX908-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec
; GFX908-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; GFX908-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; GFX908-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; GFX908-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: bb.3: ; GFX908-NEXT: bb.3:
; GFX908-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GFX908-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GFX908-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GFX908-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GFX908-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: bb.4: ; GFX908-NEXT: bb.4:
; GFX908-NEXT: successors: %bb.5(0x80000000) ; GFX908-NEXT: successors: %bb.5(0x80000000)
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GFX908-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
Show All 32 Linesdefine amdgpu_ps void @raw_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__0_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 %soffset) {
; GFX90A-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec ; GFX90A-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec
; GFX90A-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec ; GFX90A-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec
; GFX90A-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; GFX90A-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; GFX90A-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; GFX90A-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: bb.3: ; GFX90A-NEXT: bb.3:
; GFX90A-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GFX90A-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GFX90A-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GFX90A-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GFX90A-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: bb.4: ; GFX90A-NEXT: bb.4:
; GFX90A-NEXT: successors: %bb.5(0x80000000) ; GFX90A-NEXT: successors: %bb.5(0x80000000)
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GFX90A-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
Show All 11 Linesdefine amdgpu_ps void @raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add4095(float %val, <4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add4095 ; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add4095
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; Natural mapping + slc ; Natural mapping + slc
define amdgpu_ps void @raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX908-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; GFX908-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; GFX90A-LABEL: name: raw_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 2) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 2)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX908-LABEL: name: raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; GFX908-LABEL: name: raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; GFX90A-LABEL: name: raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX908-LABEL: name: raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset ; GFX908-LABEL: name: raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset ; GFX90A-LABEL: name: raw_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg) #0 declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg) #0
declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg) #0 declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg) #0
attributes #0 = { nounwind } attributes #0 = { nounwind }

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.buffer.load.format.f16.ll

Show All 9 Linesdefine amdgpu_ps half @raw_buffer_load_format_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_OFFEN]] ; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_OFFEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; UNPACKED-LABEL: name: raw_buffer_load_format_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: raw_buffer_load_format_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.raw.buffer.load.format.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call half @llvm.amdgcn.raw.buffer.load.format.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret half %val ret half %val
} }
define amdgpu_ps <2 x half> @raw_buffer_load_format_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x half> @raw_buffer_load_format_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; PACKED-LABEL: name: raw_buffer_load_format_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: raw_buffer_load_format_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XY_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_XY_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XY_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_XY_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_OFFEN]] ; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_OFFEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; UNPACKED-LABEL: name: raw_buffer_load_format_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: raw_buffer_load_format_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN]].sub0 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN]].sub0
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN]].sub1 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN]].sub1
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY6]], [[COPY8]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY6]], [[COPY8]], implicit $exec
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_1:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY9]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_1:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY9]], implicit $exec
; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16
Show All 19 Linesdefine amdgpu_ps <4 x half> @raw_buffer_load_format_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub0 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub0
; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub1 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub1
; PACKED-NEXT: $vgpr0 = COPY [[COPY6]] ; PACKED-NEXT: $vgpr0 = COPY [[COPY6]]
; PACKED-NEXT: $vgpr1 = COPY [[COPY7]] ; PACKED-NEXT: $vgpr1 = COPY [[COPY7]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
; UNPACKED-LABEL: name: raw_buffer_load_format_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: raw_buffer_load_format_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub0 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub0
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub1 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub1
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub2 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub2
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub3 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub3
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY6]], [[COPY10]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY6]], [[COPY10]], implicit $exec
; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
▲ Show 20 Lines • Show All 51 Lines • ▼ Show 20 Linesdefine amdgpu_ps half @raw_buffer_load_format_f16__vgpr_rsrc__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec ; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec
; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 33 Linesdefine amdgpu_ps half @raw_buffer_load_format_f16__vgpr_rsrc__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec ; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec
; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 11 Linesdefine amdgpu_ps <4 x half> @raw_buffer_load_format_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add_4095(<4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub0 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub0
; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub1 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub1
; PACKED-NEXT: $vgpr0 = COPY [[COPY6]] ; PACKED-NEXT: $vgpr0 = COPY [[COPY6]]
; PACKED-NEXT: $vgpr1 = COPY [[COPY7]] ; PACKED-NEXT: $vgpr1 = COPY [[COPY7]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
; UNPACKED-LABEL: name: raw_buffer_load_format_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add_4095 ; UNPACKED-LABEL: name: raw_buffer_load_format_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add_4095
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub0 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub0
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub1 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub1
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub2 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub2
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub3 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub3
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY6]], [[COPY10]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY6]], [[COPY10]], implicit $exec
; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
Show All 26 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.buffer.load.format.ll

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck %s ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck %s
; Natural mapping ; Natural mapping
define amdgpu_ps float @raw_buffer_load_format_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_format_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_format_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load_format_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_X_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_X_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.format.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.format.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps <2 x float> @raw_buffer_load_format_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x float> @raw_buffer_load_format_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_format_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load_format_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XY_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_XY_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XY_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_XY_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XY_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XY_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XY_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XY_OFFEN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%val = call <2 x float> @llvm.amdgcn.raw.buffer.load.format.v2f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call <2 x float> @llvm.amdgcn.raw.buffer.load.format.v2f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret <2 x float> %val ret <2 x float> %val
} }
define amdgpu_ps <3 x float> @raw_buffer_load_format_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <3 x float> @raw_buffer_load_format_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_format_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load_format_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZ_OFFEN:%[0-9]+]]:vreg_96 = BUFFER_LOAD_FORMAT_XYZ_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZ_OFFEN:%[0-9]+]]:vreg_96 = BUFFER_LOAD_FORMAT_XYZ_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub1
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub2 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub2
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY8]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%val = call <3 x float> @llvm.amdgcn.raw.buffer.load.format.v3f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call <3 x float> @llvm.amdgcn.raw.buffer.load.format.v3f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret <3 x float> %val ret <3 x float> %val
} }
define amdgpu_ps <4 x float> @raw_buffer_load_format_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <4 x float> @raw_buffer_load_format_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_format_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load_format_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub1
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub2 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub2
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub3 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub3
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr3 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr3 = COPY [[COPY9]]
Show All 37 Linesdefine amdgpu_ps float @raw_buffer_load_format_f32__vgpr_rsrc__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 11 Linesdefine amdgpu_ps <4 x float> @raw_buffer_load_format_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add_4095(<4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub1
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub2 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub2
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub3 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub3
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr3 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr3 = COPY [[COPY9]]
Show All 12 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.buffer.load.ll

Show All 9 Linesdefine amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
; Copies for VGPR arguments ; Copies for VGPR arguments
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__sgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 inreg %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__sgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 inreg %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__sgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__sgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr7 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr7
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[COPY4]] ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[COPY4]]
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY6]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY6]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
; Waterfall for rsrc ; Waterfall for rsrc
define amdgpu_ps float @raw_buffer_load_f32__vgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__vgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
Show All 26 Linesdefine amdgpu_ps float @raw_buffer_load_f32__vgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY8]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY8]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE1]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE1]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 38 Linesdefine amdgpu_ps float @raw_buffer_load_f32__vgpr_rsrc__vgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 12 Linesdefine amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 1, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 1, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 1) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 1)
ret float %val ret float %val
} }
; Natural mapping + slc ; Natural mapping + slc
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 2, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 2, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 2) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 2)
ret float %val ret float %val
} }
; Natural mapping + dlc ; Natural mapping + dlc
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 4, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 4, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 4) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 4)
ret float %val ret float %val
} }
; Natural mapping + slc + dlc ; Natural mapping + slc + dlc
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_dlc ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_dlc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 6, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 6, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 6) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 6)
ret float %val ret float %val
} }
; Natural mapping + glc + dlc ; Natural mapping + glc + dlc
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc_dlc ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc_dlc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 5, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 5, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 5) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 5)
ret float %val ret float %val
} }
; Natural mapping + glc + slc + dlc ; Natural mapping + glc + slc + dlc
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc_slc_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc_slc_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc_slc_dlc ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc_slc_dlc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 7, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 7, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 7) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 7)
ret float %val ret float %val
} }
; Natural mapping ; Natural mapping
define amdgpu_ps <2 x float> @raw_buffer_load_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x float> @raw_buffer_load_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORDX2_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORDX2_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_OFFEN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%val = call <2 x float> @llvm.amdgcn.raw.buffer.load.v2f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call <2 x float> @llvm.amdgcn.raw.buffer.load.v2f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret <2 x float> %val ret <2 x float> %val
} }
define amdgpu_ps <3 x float> @raw_buffer_load_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <3 x float> @raw_buffer_load_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORDX3_OFFEN:%[0-9]+]]:vreg_96 = BUFFER_LOAD_DWORDX3_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORDX3_OFFEN:%[0-9]+]]:vreg_96 = BUFFER_LOAD_DWORDX3_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_OFFEN]].sub1
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_OFFEN]].sub2 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_OFFEN]].sub2
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY8]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%val = call <3 x float> @llvm.amdgcn.raw.buffer.load.v3f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call <3 x float> @llvm.amdgcn.raw.buffer.load.v3f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret <3 x float> %val ret <3 x float> %val
} }
define amdgpu_ps <4 x float> @raw_buffer_load_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <4 x float> @raw_buffer_load_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORDX4_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORDX4_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub1
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub2 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub2
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub3 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub3
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr3 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr3 = COPY [[COPY9]]
Show All 9 Linesdefine amdgpu_ps half @raw_buffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_USHORT_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_USHORT_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_USHORT_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_USHORT_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_USHORT_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_USHORT_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.raw.buffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call half @llvm.amdgcn.raw.buffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret half %val ret half %val
} }
define amdgpu_ps <2 x half> @raw_buffer_load_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x half> @raw_buffer_load_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call <2 x half> @llvm.amdgcn.raw.buffer.load.v2f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call <2 x half> @llvm.amdgcn.raw.buffer.load.v2f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret <2 x half> %val ret <2 x half> %val
} }
; FIXME: Crashes ; FIXME: Crashes
; define amdgpu_ps <3 x half> @raw_buffer_load_v3f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { ; define amdgpu_ps <3 x half> @raw_buffer_load_v3f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; %val = call <3 x half> @llvm.amdgcn.raw.buffer.load.v3f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) ; %val = call <3 x half> @llvm.amdgcn.raw.buffer.load.v3f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
; ret <3 x half> %val ; ret <3 x half> %val
; } ; }
define amdgpu_ps <4 x half> @raw_buffer_load_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <4 x half> @raw_buffer_load_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORDX2_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORDX2_OFFEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_OFFEN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%val = call <4 x half> @llvm.amdgcn.raw.buffer.load.v4f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call <4 x half> @llvm.amdgcn.raw.buffer.load.v4f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret <4 x half> %val ret <4 x half> %val
} }
define amdgpu_ps float @raw_buffer_load_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset_zext(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset_zext(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset_zext ; CHECK-LABEL: name: raw_buffer_load_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset_zext
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_UBYTE_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_UBYTE_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s8), addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_UBYTE_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_UBYTE_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s8), addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_UBYTE_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_UBYTE_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call i8 @llvm.amdgcn.raw.buffer.load.i8(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call i8 @llvm.amdgcn.raw.buffer.load.i8(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
%zext = zext i8 %val to i32 %zext = zext i8 %val to i32
%cast = bitcast i32 %zext to float %cast = bitcast i32 %zext to float
ret float %cast ret float %cast
} }
define amdgpu_ps float @raw_buffer_load_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset_sext(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset_sext(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset_sext ; CHECK-LABEL: name: raw_buffer_load_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset_sext
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_SBYTE_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_SBYTE_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s8), addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_SBYTE_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_SBYTE_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s8), addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_SBYTE_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_SBYTE_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call i8 @llvm.amdgcn.raw.buffer.load.i8(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call i8 @llvm.amdgcn.raw.buffer.load.i8(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
%zext = sext i8 %val to i32 %zext = sext i8 %val to i32
%cast = bitcast i32 %zext to float %cast = bitcast i32 %zext to float
ret float %cast ret float %cast
} }
define amdgpu_ps float @raw_buffer_load_i16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_zext(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_i16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_zext(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_i16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_zext ; CHECK-LABEL: name: raw_buffer_load_i16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_zext
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_USHORT_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_USHORT_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_USHORT_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_USHORT_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_USHORT_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_USHORT_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call i16 @llvm.amdgcn.raw.buffer.load.i16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call i16 @llvm.amdgcn.raw.buffer.load.i16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
%zext = zext i16 %val to i32 %zext = zext i16 %val to i32
%cast = bitcast i32 %zext to float %cast = bitcast i32 %zext to float
ret float %cast ret float %cast
} }
define amdgpu_ps float @raw_buffer_load_i16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_sext(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_i16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_sext(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_i16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_sext ; CHECK-LABEL: name: raw_buffer_load_i16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_sext
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_SSHORT_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_SSHORT_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_SSHORT_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_SSHORT_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_SSHORT_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_SSHORT_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call i16 @llvm.amdgcn.raw.buffer.load.i16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call i16 @llvm.amdgcn.raw.buffer.load.i16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
%sext = sext i16 %val to i32 %sext = sext i16 %val to i32
%cast = bitcast i32 %sext to float %cast = bitcast i32 %sext to float
ret float %cast ret float %cast
} }
Show All 28 Linesdefine amdgpu_ps half @raw_buffer_load_f16__vgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY8]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY8]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[BUFFER_LOAD_USHORT_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_USHORT_OFFEN [[COPY4]], [[REG_SEQUENCE1]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_USHORT_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_USHORT_OFFEN [[COPY4]], [[REG_SEQUENCE1]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 35 Linesdefine amdgpu_ps float @raw_buffer_load_i8__vgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY8]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY8]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[BUFFER_LOAD_UBYTE_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_UBYTE_OFFEN [[COPY4]], [[REG_SEQUENCE1]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s8), addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_UBYTE_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_UBYTE_OFFEN [[COPY4]], [[REG_SEQUENCE1]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s8), addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 12 Linesdefine amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vdpr_voffset__sgpr_soffset__voffset0(<4 x i32> inreg %rsrc, i32 inreg %soffset) {
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET [[REG_SEQUENCE]], [[COPY4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET [[REG_SEQUENCE]], [[COPY4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFSET]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFSET]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4095(<4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4095(<4 x i32> inreg %rsrc, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4095 ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET [[REG_SEQUENCE]], [[COPY4]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET [[REG_SEQUENCE]], [[COPY4]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFSET]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFSET]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4096(<4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4096(<4 x i32> inreg %rsrc, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4096 ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY5]], [[REG_SEQUENCE]], [[COPY4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY5]], [[REG_SEQUENCE]], [[COPY4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 4096, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 4096, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add16(<4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add16(<4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add16 ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 16, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 16, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 16 %voffset = add i32 %voffset.base, 16
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4095(<4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4095(<4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4095 ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4096(<4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4096(<4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4096 ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY4]], [[COPY6]], 0, implicit $exec ; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY4]], [[COPY6]], 0, implicit $exec
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 4096 %voffset = add i32 %voffset.base, 4096
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4095(<4 x i32> inreg %rsrc, i32 %voffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4095(<4 x i32> inreg %rsrc, i32 %voffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4095 ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 4095, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 4095, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4096(<4 x i32> inreg %rsrc, i32 %voffset) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4096(<4 x i32> inreg %rsrc, i32 %voffset) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4096 ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 4096, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 4096, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add16(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add16(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add16 ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY5]], [[S_MOV_B32_]], implicit-def $scc ; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY5]], [[S_MOV_B32_]], implicit-def $scc
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%soffset = add i32 %soffset.base, 16 %soffset = add i32 %soffset.base, 16
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4095(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4095(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4095 ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095
; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY5]], [[S_MOV_B32_]], implicit-def $scc ; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY5]], [[S_MOV_B32_]], implicit-def $scc
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%soffset = add i32 %soffset.base, 4095 %soffset = add i32 %soffset.base, 4095
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4096(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) { define amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4096(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) {
; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4096 ; CHECK-LABEL: name: raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY5]], [[S_MOV_B32_]], implicit-def $scc ; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY5]], [[S_MOV_B32_]], implicit-def $scc
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%soffset = add i32 %soffset.base, 4096 %soffset = add i32 %soffset.base, 4096
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
; An add of the offset is necessary, with a waterfall loop. Make sure the add is done outside of the waterfall loop. ; An add of the offset is necessary, with a waterfall loop. Make sure the add is done outside of the waterfall loop.
Show All 29 Linesdefine amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add5000(<4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset.base) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY8]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY8]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE1]], [[S_ADD_I32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[COPY4]], [[REG_SEQUENCE1]], [[S_ADD_I32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 39 Linesdefine amdgpu_ps float @raw_buffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add5000(<4 x i32> %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE1]], [[COPY5]], 904, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFEN [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE1]], [[COPY5]], 904, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 20 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.buffer.store.format.f16.ll

Show All 9 Linesdefine amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f16(<4 x i32> inreg %rsrc, half %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f16 ; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f16
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__voffset_4095__sgpr_soffset_f16(<4 x i32> inreg %rsrc, half %val, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__voffset_4095__sgpr_soffset_f16(<4 x i32> inreg %rsrc, half %val, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__voffset_4095__sgpr_soffset_f16 ; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__voffset_4095__sgpr_soffset_f16
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_gfx80_OFFSET_exact [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_gfx80_OFFSET_exact [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__voffset_4095__sgpr_soffset_f16 ; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__voffset_4095__sgpr_soffset_f16
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_OFFSET_exact [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_OFFSET_exact [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.f16(half %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.f16(half %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16 ; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY7]], [[COPY4]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY7]], [[COPY4]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16 ; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> inreg %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> inreg %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16 ; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
Show All 9 Linesdefine amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> inreg %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY8]], [[COPY4]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY8]], [[COPY4]], implicit $exec
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_1:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY9]], [[COPY5]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_1:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY9]], [[COPY5]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1, [[COPY5]], %subreg.sub2, [[V_LSHRREV_B32_e64_1]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1, [[COPY5]], %subreg.sub2, [[V_LSHRREV_B32_e64_1]], %subreg.sub3
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16 ; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; PACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; Make sure unpack code is emitted outside of loop ; Make sure unpack code is emitted outside of loop
define amdgpu_ps void @raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16 ; UNPACKED-LABEL: name: raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16
Show All 33 Linesdefine amdgpu_ps void @raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY12]], [[COPY10]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY12]], [[COPY10]], implicit $exec
; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY13]], [[COPY11]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY13]], [[COPY11]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE2]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE2]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 31 Linesdefine amdgpu_ps void @raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; PACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY11]], [[COPY9]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY11]], [[COPY9]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE2]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE2]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 15 Linesdefine amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4095(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095
; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]] ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY6]], [[COPY4]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY6]], [[COPY4]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4095 ; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4095
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095 ; PACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 4095, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 4095, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4096(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4096(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4096 ; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4096
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]] ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY6]], [[COPY4]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY6]], [[COPY4]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4096 ; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4096
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; PACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 4096, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 4096, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_16(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_16(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_16 ; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_16
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY7]], [[COPY4]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY7]], [[COPY4]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 16, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 16, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_16 ; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_16
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 16, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 16, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 16 %voffset.add = add i32 %voffset, 16
call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4095(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4095(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4095 ; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4095
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY7]], [[COPY4]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY7]], [[COPY4]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4095 ; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4095
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 4095 %voffset.add = add i32 %voffset, 4095
call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4096(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4096(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4096 ; UNPACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4096
Show All 10 Linesdefine amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4096(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY5]], [[COPY7]], 0, implicit $exec ; UNPACKED-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY5]], [[COPY7]], 0, implicit $exec
; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]] ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY8]], [[COPY4]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY8]], [[COPY4]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4096 ; PACKED-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_voffset_add_4096
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; PACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; PACKED-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY5]], [[COPY7]], 0, implicit $exec ; PACKED-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY5]], [[COPY7]], 0, implicit $exec
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY4]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 4096 %voffset.add = add i32 %voffset, 4096
call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
; Check what happens with offset add inside a waterfall loop ; Check what happens with offset add inside a waterfall loop
Show All 38 Linesdefine amdgpu_ps void @raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16_add_4096(<4 x i32> %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY13]], [[COPY11]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY13]], [[COPY11]], implicit $exec
; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY14]], [[COPY12]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY14]], [[COPY12]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE2]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE2]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 34 Linesdefine amdgpu_ps void @raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16_add_4096(<4 x i32> %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY11]], [[COPY9]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY11]], [[COPY9]], implicit $exec
; PACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY12]], [[COPY10]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY12]], [[COPY10]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE2]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE2]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 10 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.buffer.store.format.f32.ll

Show All 9 Linesdefine amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__voffset_4095__sgpr_soffset_f32(<4 x i32> inreg %rsrc, float %val, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__voffset_4095__sgpr_soffset_f32(<4 x i32> inreg %rsrc, float %val, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__voffset_4095__sgpr_soffset_f32 ; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__voffset_4095__sgpr_soffset_f32
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_FORMAT_X_OFFSET_exact [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_X_OFFSET_exact [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.f32(float %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.f32(float %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32 ; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v3f32(<4 x i32> inreg %rsrc, <3 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v3f32(<4 x i32> inreg %rsrc, <3 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v3f32 ; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v3f32
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZ_OFFEN_exact [[REG_SEQUENCE1]], [[COPY7]], [[REG_SEQUENCE]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZ_OFFEN_exact [[REG_SEQUENCE1]], [[COPY7]], [[REG_SEQUENCE]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32(<4 x i32> inreg %rsrc, <4 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32(<4 x i32> inreg %rsrc, <4 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32 ; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[COPY8]], [[REG_SEQUENCE]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[COPY8]], [[REG_SEQUENCE]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32(<4 x i32> %rsrc, <4 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32(<4 x i32> %rsrc, <4 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32 ; CHECK-LABEL: name: raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
Show All 29 Linesdefine amdgpu_ps void @raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32(<4 x i32> %rsrc, <4 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY12]], [[COPY10]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY12]], [[COPY10]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY13]], [[COPY11]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY13]], [[COPY11]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[COPY8]], [[REG_SEQUENCE2]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[COPY8]], [[REG_SEQUENCE2]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 13 Linesdefine amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_soffset4095(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095
; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 4095, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 4095, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_soffset4096(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_soffset4096(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_soffset4096 ; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_soffset4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 4096, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 4096, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_16(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_16(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_16 ; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 16, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 16, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 16 %voffset.add = add i32 %voffset, 16
call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_4095(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_4095(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_4095 ; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 4095, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 4095, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 4095 %voffset.add = add i32 %voffset, 4095
call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_4096(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_4096(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_4096 ; CHECK-LABEL: name: raw_buffer_store_format__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32_voffset_add_4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY6]], [[COPY8]], 0, implicit $exec ; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY6]], [[COPY8]], 0, implicit $exec
; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE1]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 4096 %voffset.add = add i32 %voffset, 4096
call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
; Check what happens with offset add inside a waterfall loop ; Check what happens with offset add inside a waterfall loop
Show All 35 Linesdefine amdgpu_ps void @raw_buffer_store_format__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32_add_4096(<4 x i32> %rsrc, <4 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY13]], [[COPY11]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY13]], [[COPY11]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY14]], [[COPY12]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY14]], [[COPY12]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE2]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFEN_exact [[REG_SEQUENCE1]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE2]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 11 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.buffer.store.ll

Show All 10 Linesdefine amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; Copies for VGPR arguments ; Copies for VGPR arguments
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__sgpr_val__sgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, float inreg %val, i32 inreg %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__sgpr_val__sgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, float inreg %val, i32 inreg %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__sgpr_val__sgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__sgpr_val__sgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr7 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr7
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr8 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr8
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[COPY4]] ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[COPY4]]
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[COPY5]] ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[COPY5]]
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY7]], [[COPY8]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY7]], [[COPY8]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; Waterfall for rsrc ; Waterfall for rsrc
define amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset
Show All 26 Linesdefine amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 26 Linesdefine amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__vgpr_soffset(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64 = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64 = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_64_xexec = COPY [[V_CMP_EQ_U32_e64_]] ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_64_xexec = COPY [[V_CMP_EQ_U32_e64_]]
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[COPY7]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[COPY7]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[V_READFIRSTLANE_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[V_READFIRSTLANE_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 38 Linesdefine amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, float %val, i32 %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 11 Linesdefine amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 1) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 1)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_slc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_slc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_slc ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_slc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 2, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 2, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 2) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 2)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_slc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_slc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_slc ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_slc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 3, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 3, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 3) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 3)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_dlc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_dlc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_dlc ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_dlc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 4, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 4, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 4) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 4)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_slc_dlc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_slc_dlc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_slc_dlc ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_slc_dlc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 6, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 6, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 6) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 6)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_dlc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_dlc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_dlc ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_dlc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 5, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 5, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 5) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 5)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_slc_dlc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_slc_dlc(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_slc_dlc ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_glc_slc_dlc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 7, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 7, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 7) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 7)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32(<4 x i32> inreg %rsrc, <2 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f32
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORDX2_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORDX2_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v3f32(<4 x i32> inreg %rsrc, <3 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v3f32(<4 x i32> inreg %rsrc, <3 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v3f32 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v3f32
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORDX3_OFFEN_exact [[REG_SEQUENCE1]], [[COPY7]], [[REG_SEQUENCE]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORDX3_OFFEN_exact [[REG_SEQUENCE1]], [[COPY7]], [[REG_SEQUENCE]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32(<4 x i32> inreg %rsrc, <4 x float> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32(<4 x i32> inreg %rsrc, <4 x float> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f32
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORDX4_OFFEN_exact [[REG_SEQUENCE1]], [[COPY8]], [[REG_SEQUENCE]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORDX4_OFFEN_exact [[REG_SEQUENCE1]], [[COPY8]], [[REG_SEQUENCE]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_i8(<4 x i32> inreg %rsrc, i32 %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_i8(<4 x i32> inreg %rsrc, i32 %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_i8 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_i8
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_BYTE_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 7) ; CHECK-NEXT: BUFFER_STORE_BYTE_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%val.trunc = trunc i32 %val to i8 %val.trunc = trunc i32 %val to i8
call void @llvm.amdgcn.raw.buffer.store.i8(i8 %val.trunc, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.i8(i8 %val.trunc, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_i16(<4 x i32> inreg %rsrc, i32 %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_i16(<4 x i32> inreg %rsrc, i32 %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_i16 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_i16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_SHORT_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_SHORT_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%val.trunc = trunc i32 %val to i16 %val.trunc = trunc i32 %val to i16
call void @llvm.amdgcn.raw.buffer.store.i16(i16 %val.trunc, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.i16(i16 %val.trunc, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f16(<4 x i32> inreg %rsrc, half %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f16(<4 x i32> inreg %rsrc, half %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f16 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_SHORT_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_SHORT_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> inreg %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> inreg %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORDX2_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORDX2_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16 ; CHECK-LABEL: name: raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
Show All 27 Linesdefine amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v4f16(<4 x i32> %rsrc, <4 x half> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY11]], [[COPY9]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY11]], [[COPY9]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: BUFFER_STORE_DWORDX2_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE2]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORDX2_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE2]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 10 Linesdefine amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__sgpr_soffset_f32_voffset4095(<4 x i32> inreg %rsrc, float %val, i32 inreg %soffset) {
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFSET_exact [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFSET_exact [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 4095, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__sgpr_soffset_f32_voffset4096(<4 x i32> inreg %rsrc, float %val, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__sgpr_soffset_f32_voffset4096(<4 x i32> inreg %rsrc, float %val, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__sgpr_soffset_f32_voffset4096 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__sgpr_soffset_f32_voffset4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY6]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY6]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 4096, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 4096, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_16(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_16(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_16 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 16, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 16, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 16 %voffset.add = add i32 %voffset, 16
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_4095(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_4095(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_4095 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 4095 %voffset.add = add i32 %voffset, 4095
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_4096(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_4096(<4 x i32> inreg %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_4096 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_f32_voffset_add_4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY5]], [[COPY7]], 0, implicit $exec ; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY5]], [[COPY7]], 0, implicit $exec
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 4096 %voffset.add = add i32 %voffset, 4096
call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4095(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4095(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4095 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 4095, i32 0) call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 4095, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4096(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4096(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4096 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 4096, i32 0) call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 4096, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_16(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_16(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_16 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 16, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 16, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 16 %voffset.add = add i32 %voffset, 16
call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_4095(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_4095(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_4095 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 4095 %voffset.add = add i32 %voffset, 4095
call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_4096(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_4096(<4 x i32> inreg %rsrc, <2 x half> %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_4096 ; CHECK-LABEL: name: raw_buffer_store__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_v2f16_soffset_add_4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY5]], [[COPY7]], 0, implicit $exec ; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64_xexec = V_ADD_CO_U32_e64 [[COPY5]], [[COPY7]], 0, implicit $exec
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 4096 %voffset.add = add i32 %voffset, 4096
call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0) call void @llvm.amdgcn.raw.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
; An add of the offset is necessary, with a waterfall loop. Make sure the add is done outside of the waterfall loop. ; An add of the offset is necessary, with a waterfall loop. Make sure the add is done outside of the waterfall loop.
define amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_offset_add_5000(<4 x i32> %rsrc, float %val, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_offset_add_5000(<4 x i32> %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
Show All 30 Linesdefine amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset_offset_add_5000(<4 x i32> %rsrc, float %val, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY11]], [[COPY9]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY11]], [[COPY9]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE1]], [[COPY6]], 904, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[V_ADD_CO_U32_e64_]], [[REG_SEQUENCE1]], [[COPY6]], 904, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 37 Linesdefine amdgpu_ps void @raw_buffer_store__vgpr_rsrc__vgpr_val__5000_voffset__sgpr_soffset_offset(<4 x i32> %rsrc, float %val, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY5]], 904, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_OFFEN_exact [[COPY4]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY5]], 904, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 17 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.tbuffer.load.f16.ll

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefix=UNPACKED %s ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefix=UNPACKED %s
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx810 -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefix=PACKED %s ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx810 -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck -check-prefix=PACKED %s
define amdgpu_ps half @raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps half @raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN]] ; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret half %val ret half %val
} }
define amdgpu_ps <2 x half> @raw_tbuffer_load_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x half> @raw_tbuffer_load_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_load_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: raw_tbuffer_load_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN]].sub0 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN]].sub0
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN]].sub1 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_OFFEN]].sub1
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY6]], [[COPY8]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY6]], [[COPY8]], implicit $exec
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_1:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY9]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_1:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY9]], implicit $exec
; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]] ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_1]]
; UNPACKED-NEXT: [[V_LSHLREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHLREV_B32_e64 [[COPY10]], [[V_AND_B32_e64_1]], implicit $exec ; UNPACKED-NEXT: [[V_LSHLREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHLREV_B32_e64 [[COPY10]], [[V_AND_B32_e64_1]], implicit $exec
; UNPACKED-NEXT: [[V_OR_B32_e64_:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 [[V_AND_B32_e64_]], [[V_LSHLREV_B32_e64_]], implicit $exec ; UNPACKED-NEXT: [[V_OR_B32_e64_:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 [[V_AND_B32_e64_]], [[V_LSHLREV_B32_e64_]], implicit $exec
; UNPACKED-NEXT: $vgpr0 = COPY [[V_OR_B32_e64_]] ; UNPACKED-NEXT: $vgpr0 = COPY [[V_OR_B32_e64_]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: raw_tbuffer_load_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: raw_tbuffer_load_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XY_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_XY_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XY_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_XY_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_OFFEN]] ; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_OFFEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call <2 x half> @llvm.amdgcn.raw.tbuffer.load.v2f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call <2 x half> @llvm.amdgcn.raw.tbuffer.load.v2f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret <2 x half> %val ret <2 x half> %val
} }
; FIXME: Crashes ; FIXME: Crashes
; define amdgpu_ps <3 x half> @raw_tbuffer_load_v3f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { ; define amdgpu_ps <3 x half> @raw_tbuffer_load_v3f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; %val = call <3 x half> @llvm.amdgcn.raw.tbuffer.load.v3f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) ; %val = call <3 x half> @llvm.amdgcn.raw.tbuffer.load.v3f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
; ret <3 x half> %val ; ret <3 x half> %val
; } ; }
define amdgpu_ps <4 x half> @raw_tbuffer_load_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <4 x half> @raw_tbuffer_load_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_load_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: raw_tbuffer_load_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub0 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub0
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub1 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub1
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub2 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub2
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub3 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_OFFEN]].sub3
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY6]], [[COPY10]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY6]], [[COPY10]], implicit $exec
; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
Show All 18 Linesdefine amdgpu_ps <4 x half> @raw_tbuffer_load_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_OFFEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_D16_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_OFFEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_D16_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub0 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub0
; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub1 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_OFFEN]].sub1
; PACKED-NEXT: $vgpr0 = COPY [[COPY6]] ; PACKED-NEXT: $vgpr0 = COPY [[COPY6]]
; PACKED-NEXT: $vgpr1 = COPY [[COPY7]] ; PACKED-NEXT: $vgpr1 = COPY [[COPY7]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%val = call <4 x half> @llvm.amdgcn.raw.tbuffer.load.v4f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call <4 x half> @llvm.amdgcn.raw.tbuffer.load.v4f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret <4 x half> %val ret <4 x half> %val
} }
Show All 32 Linesdefine amdgpu_ps half @raw_tbuffer_load_f16__vgpr_rsrc__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec ; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec
; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 33 Linesdefine amdgpu_ps half @raw_tbuffer_load_f16__vgpr_rsrc__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec ; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec
; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 11 Linesdefine amdgpu_ps half @raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 1, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 1, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc ; PACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 1, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 1, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN]] ; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 1) %val = call half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 1)
ret half %val ret half %val
} }
define amdgpu_ps half @raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps half @raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; UNPACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 2, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 2, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; PACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 2, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 2, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN]] ; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 2) %val = call half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 2)
ret half %val ret half %val
} }
define amdgpu_ps half @raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps half @raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc ; UNPACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 3, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 3, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc ; PACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 3, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 3, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN]] ; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 3) %val = call half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 3)
ret half %val ret half %val
} }
define amdgpu_ps half @raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps half @raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc ; UNPACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 4, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 4, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_OFFEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc ; PACKED-LABEL: name: raw_tbuffer_load_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 4, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 4, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN]] ; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_OFFEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 4) %val = call half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 4)
ret half %val ret half %val
} }
declare half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0 declare half @llvm.amdgcn.raw.tbuffer.load.f16(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0
declare <2 x half> @llvm.amdgcn.raw.tbuffer.load.v2f16(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0 declare <2 x half> @llvm.amdgcn.raw.tbuffer.load.v2f16(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0
declare <3 x half> @llvm.amdgcn.raw.tbuffer.load.v3f16(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0 declare <3 x half> @llvm.amdgcn.raw.tbuffer.load.v3f16(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0
declare <4 x half> @llvm.amdgcn.raw.tbuffer.load.v4f16(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0 declare <4 x half> @llvm.amdgcn.raw.tbuffer.load.v4f16(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0
attributes #0 = { nounwind readonly } attributes #0 = { nounwind readonly }

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.tbuffer.load.ll

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck %s ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck %s
; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck %s ; RUN: llc -global-isel -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -stop-after=instruction-select -verify-machineinstrs -o - %s | FileCheck %s
define amdgpu_ps float @raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps <2 x float> @raw_tbuffer_load_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x float> @raw_tbuffer_load_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_load_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_tbuffer_load_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XY_OFFEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_XY_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XY_OFFEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_XY_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XY_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XY_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XY_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XY_OFFEN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%val = call <2 x float> @llvm.amdgcn.raw.tbuffer.load.v2f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call <2 x float> @llvm.amdgcn.raw.tbuffer.load.v2f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret <2 x float> %val ret <2 x float> %val
} }
define amdgpu_ps <3 x float> @raw_tbuffer_load_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <3 x float> @raw_tbuffer_load_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_load_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_tbuffer_load_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XYZ_OFFEN:%[0-9]+]]:vreg_96 = TBUFFER_LOAD_FORMAT_XYZ_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XYZ_OFFEN:%[0-9]+]]:vreg_96 = TBUFFER_LOAD_FORMAT_XYZ_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub1
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub2 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_OFFEN]].sub2
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY8]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%val = call <3 x float> @llvm.amdgcn.raw.tbuffer.load.v3f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call <3 x float> @llvm.amdgcn.raw.tbuffer.load.v3f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret <3 x float> %val ret <3 x float> %val
} }
define amdgpu_ps <4 x float> @raw_tbuffer_load_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <4 x float> @raw_tbuffer_load_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_load_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_tbuffer_load_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XYZW_OFFEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XYZW_OFFEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub0 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub0
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub1 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub1
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub2 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub2
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub3 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_OFFEN]].sub3
; CHECK-NEXT: $vgpr0 = COPY [[COPY6]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY6]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr3 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr3 = COPY [[COPY9]]
Show All 36 Linesdefine amdgpu_ps float @raw_tbuffer_load_f32__vgpr_rsrc__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY5]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY5]], implicit $exec
; CHECK-NEXT: [[S_AND_B32_1:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[S_AND_B32_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B32_1:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[S_AND_B32_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY6]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc ; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_]] ; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 11 Linesdefine amdgpu_ps float @raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 1, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 1, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 1) %val = call float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 1)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; CHECK-LABEL: name: raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 2, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 2, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 2) %val = call float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 2)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc ; CHECK-LABEL: name: raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 3, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 3, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 3) %val = call float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 3)
ret float %val ret float %val
} }
define amdgpu_ps float @raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc ; CHECK-LABEL: name: raw_tbuffer_load_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 4, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_OFFEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_OFFEN [[COPY4]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 4, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_OFFEN]] ; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_OFFEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 4) %val = call float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 4)
ret float %val ret float %val
} }
declare float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0 declare float @llvm.amdgcn.raw.tbuffer.load.f32(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0
declare <2 x float> @llvm.amdgcn.raw.tbuffer.load.v2f32(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0 declare <2 x float> @llvm.amdgcn.raw.tbuffer.load.v2f32(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0
declare <3 x float> @llvm.amdgcn.raw.tbuffer.load.v3f32(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0 declare <3 x float> @llvm.amdgcn.raw.tbuffer.load.v3f32(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0
declare <4 x float> @llvm.amdgcn.raw.tbuffer.load.v4f32(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0 declare <4 x float> @llvm.amdgcn.raw.tbuffer.load.v4f32(<4 x i32>, i32, i32, i32 immarg, i32 immarg) #0
attributes #0 = { nounwind readonly } attributes #0 = { nounwind readonly }

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.tbuffer.store.f16.ll

Show All 9 Linesdefine amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_store_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: raw_tbuffer_store_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY7]], [[COPY]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY7]], [[COPY]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_XY_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_tbuffer_store_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: raw_tbuffer_store_v2f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_XY_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret void ret void
} }
; FIXME: Crashes ; FIXME: Crashes
; define amdgpu_ps void @raw_tbuffer_store_v3f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<3 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { ; define amdgpu_ps void @raw_tbuffer_store_v3f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<3 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; call void @llvm.amdgcn.raw.tbuffer.store.v3f16(<3 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) ; call void @llvm.amdgcn.raw.tbuffer.store.v3f16(<3 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
Show All 15 Linesdefine amdgpu_ps void @raw_tbuffer_store_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY8]], [[COPY]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY8]], [[COPY]], implicit $exec
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_1:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY9]], [[COPY1]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_1:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY9]], [[COPY1]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1, [[COPY1]], %subreg.sub2, [[V_LSHRREV_B32_e64_1]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1, [[COPY1]], %subreg.sub2, [[V_LSHRREV_B32_e64_1]], %subreg.sub3
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_XYZW_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_XYZW_gfx80_OFFEN_exact [[REG_SEQUENCE1]], [[COPY6]], [[REG_SEQUENCE]], [[COPY7]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_tbuffer_store_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: raw_tbuffer_store_v4f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; PACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_XYZW_OFFEN_exact [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_XYZW_OFFEN_exact [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret void ret void
} }
; Waterfall for rsrc ; Waterfall for rsrc
define amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__sgpr_soffset(half %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__sgpr_soffset(half %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__sgpr_soffset
Show All 26 Linesdefine amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__sgpr_soffset(half %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 29 Linesdefine amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__sgpr_soffset(half %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; PACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 38 Linesdefine amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__vgpr_soffset(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset) {
; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 32 Linesdefine amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__vgpr_voffset__vgpr_soffset(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset) {
; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 39 Linesdefine amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__sgpr_voffset__vgpr_soffset(half %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 33 Linesdefine amdgpu_ps void @raw_tbuffer_store_f16__vgpr_rsrc__sgpr_voffset__vgpr_soffset(half %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 11 Linesdefine amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 1, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 1, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc ; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 1, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 1, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 1) call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 1)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 2, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 2, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 2, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 2, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 2) call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 2)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 3, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 3, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc ; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 3, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 3, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 3) call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 3)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(half %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc ; UNPACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 4, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_gfx80_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 4, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc ; PACKED-LABEL: name: raw_tbuffer_store_f16__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 4, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_D16_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 4, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 4) call void @llvm.amdgcn.raw.tbuffer.store.f16(half %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 4)
ret void ret void
} }
declare void @llvm.amdgcn.raw.tbuffer.store.f16(half, <4 x i32>, i32, i32, i32 immarg, i32 immarg) declare void @llvm.amdgcn.raw.tbuffer.store.f16(half, <4 x i32>, i32, i32, i32 immarg, i32 immarg)
declare void @llvm.amdgcn.raw.tbuffer.store.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg, i32 immarg) declare void @llvm.amdgcn.raw.tbuffer.store.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg, i32 immarg)
declare void @llvm.amdgcn.raw.tbuffer.store.v3f16(<3 x half>, <4 x i32>, i32, i32, i32 immarg, i32 immarg) declare void @llvm.amdgcn.raw.tbuffer.store.v3f16(<3 x half>, <4 x i32>, i32, i32, i32 immarg, i32 immarg)
declare void @llvm.amdgcn.raw.tbuffer.store.v4f16(<4 x half>, <4 x i32>, i32, i32, i32 immarg, i32 immarg) declare void @llvm.amdgcn.raw.tbuffer.store.v4f16(<4 x half>, <4 x i32>, i32, i32, i32 immarg, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.tbuffer.store.i8.ll

Show All 9 Linesdefine amdgpu_ps void @raw_tbuffer_store_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i8 %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: raw_tbuffer_store_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: raw_tbuffer_store_i8__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.i8(i8 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.i8(i8 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
; Waterfall for rsrc ; Waterfall for rsrc
define amdgpu_ps void @raw_tbuffer_store_i8__vgpr_rsrc__vgpr_voffset__sgpr_soffset(i8 %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_i8__vgpr_rsrc__vgpr_voffset__sgpr_soffset(i8 %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: raw_tbuffer_store_i8__vgpr_rsrc__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: raw_tbuffer_store_i8__vgpr_rsrc__vgpr_voffset__sgpr_soffset
Show All 26 Linesdefine amdgpu_ps void @raw_tbuffer_store_i8__vgpr_rsrc__vgpr_voffset__sgpr_soffset(i8 %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 29 Linesdefine amdgpu_ps void @raw_tbuffer_store_i8__vgpr_rsrc__vgpr_voffset__sgpr_soffset(i8 %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; PACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 38 Linesdefine amdgpu_ps void @raw_tbuffer_store_i8__vgpr_rsrc__vgpr_voffset__vgpr_soffset(i8 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset) {
; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 32 Linesdefine amdgpu_ps void @raw_tbuffer_store_i8__vgpr_rsrc__vgpr_voffset__vgpr_soffset(i8 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset) {
; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 39 Linesdefine amdgpu_ps void @raw_tbuffer_store_i8__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i8 %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; UNPACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 7) ; UNPACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 33 Linesdefine amdgpu_ps void @raw_tbuffer_store_i8__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i8 %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; PACKED-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 7) ; PACKED-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.5: ; PACKED-NEXT: bb.5:
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.i8(i8 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.i8(i8 %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
declare void @llvm.amdgcn.raw.tbuffer.store.i8(i8, <4 x i32>, i32, i32, i32 immarg, i32 immarg) declare void @llvm.amdgcn.raw.tbuffer.store.i8(i8, <4 x i32>, i32, i32, i32 immarg, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.raw.tbuffer.store.ll

Show All 10 Linesdefine amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret void ret void
} }
; Natural mapping ; Natural mapping
define amdgpu_ps void @raw_tbuffer_store_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<2 x float> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<2 x float> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_tbuffer_store_v2f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_XY_OFFEN_exact [[REG_SEQUENCE]], [[COPY6]], [[REG_SEQUENCE1]], [[COPY7]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret void ret void
} }
; Natural mapping ; Natural mapping
define amdgpu_ps void @raw_tbuffer_store_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<3 x float> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<3 x float> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_tbuffer_store_v3f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY3]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY5]], %subreg.sub2, [[COPY6]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY3]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY5]], %subreg.sub2, [[COPY6]], %subreg.sub3
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_XYZ_OFFEN_exact [[REG_SEQUENCE]], [[COPY7]], [[REG_SEQUENCE1]], [[COPY8]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_XYZ_OFFEN_exact [[REG_SEQUENCE]], [[COPY7]], [[REG_SEQUENCE1]], [[COPY8]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret void ret void
} }
; Natural mapping ; Natural mapping
define amdgpu_ps void @raw_tbuffer_store_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x float> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<4 x float> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_tbuffer_store_v4f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_XYZW_OFFEN_exact [[REG_SEQUENCE]], [[COPY8]], [[REG_SEQUENCE1]], [[COPY9]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_XYZW_OFFEN_exact [[REG_SEQUENCE]], [[COPY8]], [[REG_SEQUENCE1]], [[COPY9]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret void ret void
} }
; Copies for VGPR arguments ; Copies for VGPR arguments
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__sgpr_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 inreg %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__sgpr_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 inreg %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__sgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__sgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr7 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr7
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[COPY5]] ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[COPY5]]
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
; Waterfall for rsrc ; Waterfall for rsrc
define amdgpu_ps void @raw_tbuffer_store_f32__vgpr_rsrc__vgpr_voffset__sgpr_soffset(float %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__vgpr_rsrc__vgpr_voffset__sgpr_soffset(float %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__vgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_tbuffer_store_f32__vgpr_rsrc__vgpr_voffset__sgpr_soffset
Show All 26 Linesdefine amdgpu_ps void @raw_tbuffer_store_f32__vgpr_rsrc__vgpr_voffset__sgpr_soffset(float %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; CHECK-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 94, 1, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[COPY6]], 0, 94, 1, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc ; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_]] ; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 38 Linesdefine amdgpu_ps void @raw_tbuffer_store_f32__vgpr_rsrc__vgpr_voffset__vgpr_soffset(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[S_AND_B32_1:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[S_AND_B32_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B32_1:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[S_AND_B32_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc ; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_]] ; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 39 Linesdefine amdgpu_ps void @raw_tbuffer_store_f32__vgpr_rsrc__sgpr_voffset__vgpr_soffset(float %val, <4 x i32> %rsrc, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[S_AND_B32_1:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[S_AND_B32_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B32_1:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[S_AND_B32_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY7]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc ; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_]] ; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 12 Linesdefine amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_glc(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 1, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 1, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 1) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 1)
ret void ret void
} }
; Natural mapping + slc ; Natural mapping + slc
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 2, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 2, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 2) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 2)
ret void ret void
} }
; Natural mapping + glc + slc ; Natural mapping + glc + slc
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc_glc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 3, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 3, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 3) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 3)
ret void ret void
} }
; Natural mapping + dlc ; Natural mapping + dlc
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_dlc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 4, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 4, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 4) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 78, i32 4)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vdpr_voffset__sgpr_soffset__voffset0(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vdpr_voffset__sgpr_soffset__voffset0(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vdpr_voffset__sgpr_soffset__voffset0 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vdpr_voffset__sgpr_soffset__voffset0
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFSET_exact [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFSET_exact [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4095(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4095(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4095 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFSET_exact [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFSET_exact [[COPY]], [[REG_SEQUENCE]], [[COPY5]], 4095, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4096(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4096(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4096 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY6]], [[REG_SEQUENCE]], [[COPY5]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY6]], [[REG_SEQUENCE]], [[COPY5]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 4096, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 4096, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add16(float %val, <4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add16(float %val, <4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add16 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 16, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 16, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset = add i32 %voffset.base, 16 %voffset = add i32 %voffset.base, 16
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4095(float %val, <4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4095(float %val, <4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4095 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4096(float %val, <4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4096(float %val, <4 x i32> inreg %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4096 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset__voffset_add4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 [[COPY5]], [[COPY7]], 0, implicit $exec ; CHECK-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 [[COPY5]], [[COPY7]], 0, implicit $exec
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[V_ADD_U32_e64_]], [[REG_SEQUENCE]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[V_ADD_U32_e64_]], [[REG_SEQUENCE]], [[COPY6]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%voffset = add i32 %voffset.base, 4096 %voffset = add i32 %voffset.base, 4096
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4095(float %val, <4 x i32> inreg %rsrc, i32 %voffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4095(float %val, <4 x i32> inreg %rsrc, i32 %voffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4095 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 4095, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 4095, i32 94, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4096(float %val, <4 x i32> inreg %rsrc, i32 %voffset) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4096(float %val, <4 x i32> inreg %rsrc, i32 %voffset) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4096 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 4096, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 4096, i32 94, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add16(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add16(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add16 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add16
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY6]], [[S_MOV_B32_]], implicit-def $scc ; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY6]], [[S_MOV_B32_]], implicit-def $scc
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%soffset = add i32 %soffset.base, 16 %soffset = add i32 %soffset.base, 16
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4095(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4095(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4095 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4095
; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY6]], [[S_MOV_B32_]], implicit-def $scc ; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY6]], [[S_MOV_B32_]], implicit-def $scc
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%soffset = add i32 %soffset.base, 4095 %soffset = add i32 %soffset.base, 4095
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4096(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4096(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset.base) {
; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4096 ; CHECK-LABEL: name: raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add4096
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 4096
; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY6]], [[S_MOV_B32_]], implicit-def $scc ; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY6]], [[S_MOV_B32_]], implicit-def $scc
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[S_ADD_I32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%soffset = add i32 %soffset.base, 4096 %soffset = add i32 %soffset.base, 4096
call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0) call void @llvm.amdgcn.raw.tbuffer.store.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 94, i32 0)
ret void ret void
} }
; An add of the offset is necessary, with a waterfall loop. Make sure the add is done outside of the waterfall loop. ; An add of the offset is necessary, with a waterfall loop. Make sure the add is done outside of the waterfall loop.
define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add5000(float %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset.base) { define amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add5000(float %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset.base) {
Show All 29 Linesdefine amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_soffset_add5000(float %val, <4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset.base) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY9]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; CHECK-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[S_ADD_I32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[COPY5]], [[REG_SEQUENCE1]], [[S_ADD_I32_]], 0, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc ; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_1]] ; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_1]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 39 Linesdefine amdgpu_ps void @raw_tbuffer_store_f32__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add5000(float %val, <4 x i32> %rsrc, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY10]], [[COPY8]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY11]], [[COPY9]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U64_e64_1:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 [[COPY11]], [[COPY9]], implicit $exec
; CHECK-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 [[V_CMP_EQ_U64_e64_]], [[V_CMP_EQ_U64_e64_1]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B32_:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[V_ADD_U32_e64_]], [[REG_SEQUENCE1]], [[COPY6]], 904, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: TBUFFER_STORE_FORMAT_X_OFFEN_exact [[COPY]], [[V_ADD_U32_e64_]], [[REG_SEQUENCE1]], [[COPY6]], 904, 94, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc ; CHECK-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, [[S_AND_SAVEEXEC_B32_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_1]] ; CHECK-NEXT: $exec_lo = S_MOV_B32_term [[S_MOV_B32_1]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 11 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.buffer.atomic.add.ll

Show All 11 Linesdefine amdgpu_ps float @struct_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_BOTHEN_RTN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_BOTHEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.struct.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.struct.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
define amdgpu_ps float @struct_buffer_atomic_add_i32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @struct_buffer_atomic_add_i32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_atomic_add_i32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_atomic_add_i32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_BOTHEN_RTN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_BOTHEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.struct.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.struct.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
define amdgpu_ps <2 x float> @struct_buffer_atomic_add_i64__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i64 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x float> @struct_buffer_atomic_add_i64__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i64 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_atomic_add_i64__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_atomic_add_i64__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN]].sub0 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN]].sub0
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN]].sub1 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_X2_BOTHEN_RTN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY9]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY10]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY10]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%ret = call i64 @llvm.amdgcn.struct.buffer.atomic.add.i64(i64 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call i64 @llvm.amdgcn.struct.buffer.atomic.add.i64(i64 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i64 %ret to <2 x float> %cast = bitcast i64 %ret to <2 x float>
ret <2 x float> %cast ret <2 x float> %cast
Show All 11 Linesdefine amdgpu_ps void @struct_buffer_atomic_add_i64_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i64 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: BUFFER_ATOMIC_ADD_X2_BOTHEN [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_ATOMIC_ADD_X2_BOTHEN [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%ret = call i64 @llvm.amdgcn.struct.buffer.atomic.add.i64(i64 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call i64 @llvm.amdgcn.struct.buffer.atomic.add.i64(i64 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; All register operands need legalization ; All register operands need legalization
define amdgpu_ps float @struct_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps float @struct_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-LABEL: name: struct_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset ; CHECK-LABEL: name: struct_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset
Show All 34 Linesdefine amdgpu_ps float @struct_buffer_atomic_add_i32__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
▲ Show 20 Lines • Show All 45 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @struct_buffer_atomic_add_i32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; CHECK-NEXT: BUFFER_ATOMIC_ADD_BOTHEN [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_ATOMIC_ADD_BOTHEN [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 14 Linesdefine amdgpu_ps float @struct_buffer_atomic_add_i32__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(i32 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 3, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_ADD_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_BOTHEN_RTN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 3, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_BOTHEN_RTN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_BOTHEN_RTN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.struct.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2) %ret = call i32 @llvm.amdgcn.struct.buffer.atomic.add.i32(i32 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
declare i32 @llvm.amdgcn.struct.buffer.atomic.add.i32(i32, <4 x i32>, i32, i32, i32, i32 immarg) #0 declare i32 @llvm.amdgcn.struct.buffer.atomic.add.i32(i32, <4 x i32>, i32, i32, i32, i32 immarg) #0
declare i64 @llvm.amdgcn.struct.buffer.atomic.add.i64(i64, <4 x i32>, i32, i32, i32, i32 immarg) #0 declare i64 @llvm.amdgcn.struct.buffer.atomic.add.i64(i64, <4 x i32>, i32, i32, i32, i32 immarg) #0
attributes #0 = { nounwind } attributes #0 = { nounwind }

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.buffer.atomic.cmpswap.ll

Show All 14 Linesdefine amdgpu_ps float @struct_buffer_atomic_cmpswap_i32__vgpr_val__vgpr_cmp__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, i32 %cmp, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY8]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN]].sub0 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN]].sub0
; CHECK-NEXT: $vgpr0 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY9]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%ret = call i32 @llvm.amdgcn.struct.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.struct.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
Show All 10 Linesdefine amdgpu_ps void @struct_buffer_atomic_cmpswap_noret_i32__vgpr_val__vgpr_cmp__sgpr_rsrc__vgpr_voffset__sgpr_soffset(i32 %val, i32 %cmp, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: BUFFER_ATOMIC_CMPSWAP_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY8]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_ATOMIC_CMPSWAP_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY8]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%ret = call i32 @llvm.amdgcn.struct.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.struct.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; All operands need legalization ; All operands need legalization
define amdgpu_ps float @struct_buffer_atomic_cmpswap_i32__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, i32 inreg %cmp, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps float @struct_buffer_atomic_cmpswap_i32__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, i32 inreg %cmp, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-LABEL: name: struct_buffer_atomic_cmpswap_i32__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset ; CHECK-LABEL: name: struct_buffer_atomic_cmpswap_i32__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset
Show All 37 Linesdefine amdgpu_ps float @struct_buffer_atomic_cmpswap_i32__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, i32 inreg %cmp, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY11]], %subreg.sub0, [[COPY12]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY11]], %subreg.sub0, [[COPY12]], %subreg.sub1
; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN [[REG_SEQUENCE3]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN [[REG_SEQUENCE3]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: [[COPY17:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN]].sub0 ; CHECK-NEXT: [[COPY17:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN]].sub0
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
▲ Show 20 Lines • Show All 49 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @struct_buffer_atomic_cmpswap_i32_noret__sgpr_val__sgpr_cmp__vgpr_rsrc__sgpr_voffset__vgpr_soffset(i32 inreg %val, i32 inreg %cmp, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY11]], %subreg.sub0, [[COPY12]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY11]], %subreg.sub0, [[COPY12]], %subreg.sub1
; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; CHECK-NEXT: BUFFER_ATOMIC_CMPSWAP_BOTHEN [[REG_SEQUENCE3]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_ATOMIC_CMPSWAP_BOTHEN [[REG_SEQUENCE3]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 15 Linesdefine amdgpu_ps float @struct_buffer_atomic_cmpswap_i32__vgpr_val__vgpr_cmp__sgpr_rsrc__vgpr_voffset__sgpr_soffset_voffset_add4095(i32 %val, i32 %cmp, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY8]], 4095, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN:%[0-9]+]]:vreg_64 = BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY8]], 4095, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN]].sub0 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_CMPSWAP_BOTHEN_RTN]].sub0
; CHECK-NEXT: $vgpr0 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY9]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
%ret = call i32 @llvm.amdgcn.struct.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call i32 @llvm.amdgcn.struct.buffer.atomic.cmpswap.i32(i32 %val, i32 %cmp, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%cast = bitcast i32 %ret to float %cast = bitcast i32 %ret to float
ret float %cast ret float %cast
} }
declare i32 @llvm.amdgcn.struct.buffer.atomic.cmpswap.i32(i32, i32, <4 x i32>, i32, i32, i32, i32 immarg) #0 declare i32 @llvm.amdgcn.struct.buffer.atomic.cmpswap.i32(i32, i32, <4 x i32>, i32, i32, i32, i32 immarg) #0
declare i64 @llvm.amdgcn.struct.buffer.atomic.cmpswap.i64(i64, i64, <4 x i32>, i32, i32, i32, i32 immarg) #0 declare i64 @llvm.amdgcn.struct.buffer.atomic.cmpswap.i64(i64, i64, <4 x i32>, i32, i32, i32, i32 immarg) #0
attributes #0 = { nounwind } attributes #0 = { nounwind }

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.buffer.atomic.fadd-with-ret.ll

; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX90A %s ; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GFX90A %s
; RUN: not --crash llc -global-isel < %s -march=amdgcn -mcpu=gfx908 -verify-machineinstrs 2>&1 | FileCheck %s -check-prefix=GFX908 ; RUN: not --crash llc -global-isel < %s -march=amdgcn -mcpu=gfx908 -verify-machineinstrs 2>&1 | FileCheck %s -check-prefix=GFX908
; GFX908: LLVM ERROR: cannot select: %{{[0-9]+}}:vgpr_32(s32) = G_AMDGPU_BUFFER_ATOMIC_FADD %{{[0-9]+}}:vgpr, %{{[0-9]+}}:sgpr(<4 x s32>), %{{[0-9]+}}:vgpr(s32), %{{[0-9]+}}:vgpr, %{{[0-9]+}}:sgpr, 0, 0, -1 :: (volatile dereferenceable load store (s32), align 1, addrspace 7) (in function: buffer_atomic_add_f32_rtn) ; GFX908: LLVM ERROR: cannot select: %{{[0-9]+}}:vgpr_32(s32) = G_AMDGPU_BUFFER_ATOMIC_FADD %{{[0-9]+}}:vgpr, %{{[0-9]+}}:sgpr(<4 x s32>), %{{[0-9]+}}:vgpr(s32), %{{[0-9]+}}:vgpr, %{{[0-9]+}}:sgpr, 0, 0, -1 :: (volatile dereferenceable load store (s32), align 1, addrspace 8) (in function: buffer_atomic_add_f32_rtn)
declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg) declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg)
declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg) declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg)
; GFX90A-LABEL: {{^}}buffer_atomic_add_f32_rtn: ; GFX90A-LABEL: {{^}}buffer_atomic_add_f32_rtn:
; GFX90A: buffer_atomic_add_f32 v{{[0-9]+}}, v[{{[0-9:]+}}], s[{{[0-9:]+}}], s{{[0-9]+}} idxen offen glc ; GFX90A: buffer_atomic_add_f32 v{{[0-9]+}}, v[{{[0-9:]+}}], s[{{[0-9:]+}}], s{{[0-9]+}} idxen offen glc
define amdgpu_kernel void @buffer_atomic_add_f32_rtn(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset) { define amdgpu_kernel void @buffer_atomic_add_f32_rtn(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset) {
Show All 14 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.buffer.atomic.fadd.ll

Show All 12 Linesdefine amdgpu_ps void @struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX908-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX908-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset ; GFX908-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX908-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset ; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset_plus4095__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%voffset.add = add i32 %voffset, 4095 %voffset.add = add i32 %voffset, 4095
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset.add, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset.add, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__4095_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__4095_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX908-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__4095_voffset__sgpr_soffset ; GFX908-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__4095_voffset__sgpr_soffset
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__4095_voffset__sgpr_soffset ; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__4095_voffset__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 4095, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 4095, i32 %soffset, i32 0)
ret void ret void
} }
; Natural mapping, no voffset ; Natural mapping, no voffset
define amdgpu_ps void @struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX908-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset ; GFX908-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset ; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
; All register operands need legalization ; All register operands need legalization
define amdgpu_ps void @struct_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps void @struct_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; GFX908-LABEL: name: struct_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset ; GFX908-LABEL: name: struct_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset
Show All 34 Linesdefine amdgpu_ps void @struct_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; GFX908-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec ; GFX908-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; GFX908-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; GFX908-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; GFX908-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; GFX908-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: bb.3: ; GFX908-NEXT: bb.3:
; GFX908-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GFX908-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1 ; GFX908-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GFX908-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GFX908-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GFX908-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: bb.4: ; GFX908-NEXT: bb.4:
; GFX908-NEXT: successors: %bb.5(0x80000000) ; GFX908-NEXT: successors: %bb.5(0x80000000)
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GFX908-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
Show All 37 Linesdefine amdgpu_ps void @struct_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; GFX90A-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec ; GFX90A-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; GFX90A-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; GFX90A-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; GFX90A-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; GFX90A-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: bb.3: ; GFX90A-NEXT: bb.3:
; GFX90A-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GFX90A-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1 ; GFX90A-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GFX90A-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GFX90A-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GFX90A-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: bb.4: ; GFX90A-NEXT: bb.4:
; GFX90A-NEXT: successors: %bb.5(0x80000000) ; GFX90A-NEXT: successors: %bb.5(0x80000000)
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GFX90A-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
Show All 40 Linesdefine amdgpu_ps void @struct_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__0_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 %soffset) {
; GFX908-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; GFX908-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; GFX908-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; GFX908-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; GFX908-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; GFX908-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; GFX908-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; GFX908-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: bb.3: ; GFX908-NEXT: bb.3:
; GFX908-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GFX908-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GFX908-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GFX908-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GFX908-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: bb.4: ; GFX908-NEXT: bb.4:
; GFX908-NEXT: successors: %bb.5(0x80000000) ; GFX908-NEXT: successors: %bb.5(0x80000000)
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GFX908-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
Show All 34 Linesdefine amdgpu_ps void @struct_buffer_atomic_add_f32_noret__sgpr_val__vgpr_rsrc__0_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 %soffset) {
; GFX90A-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec ; GFX90A-NEXT: [[V_READFIRSTLANE_B32_4:%[0-9]+]]:sreg_32 = V_READFIRSTLANE_B32 [[COPY6]], implicit $exec
; GFX90A-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; GFX90A-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; GFX90A-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; GFX90A-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; GFX90A-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; GFX90A-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: bb.3: ; GFX90A-NEXT: bb.3:
; GFX90A-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GFX90A-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY7]], [[COPY8]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GFX90A-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GFX90A-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GFX90A-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: bb.4: ; GFX90A-NEXT: bb.4:
; GFX90A-NEXT: successors: %bb.5(0x80000000) ; GFX90A-NEXT: successors: %bb.5(0x80000000)
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GFX90A-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
Show All 14 Linesdefine amdgpu_ps void @struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX908-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc ; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset_slc
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset_slc(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset_slc(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX908-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset_slc ; GFX908-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset_slc
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset_slc ; GFX90A-LABEL: name: struct_buffer_atomic_add_f32_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset_slc
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 2) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 2)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX908-LABEL: name: struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; GFX908-LABEL: name: struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX908-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset ; GFX90A-LABEL: name: struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__vgpr_voffset__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX908-LABEL: name: struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset ; GFX908-LABEL: name: struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset
; GFX908: bb.1 (%ir-block.0): ; GFX908: bb.1 (%ir-block.0):
; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX908-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A-LABEL: name: struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset ; GFX90A-LABEL: name: struct_buffer_atomic_add_v2f16_noret__vgpr_val__sgpr_rsrc__0_voffset__sgpr_soffset
; GFX90A: bb.1 (%ir-block.0): ; GFX90A: bb.1 (%ir-block.0):
; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; GFX90A-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; GFX90A-NEXT: {{ $}} ; GFX90A-NEXT: {{ $}}
; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; GFX90A-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; GFX90A-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; GFX90A-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; GFX90A-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; GFX90A-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; GFX90A-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; GFX90A-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 7) ; GFX90A-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY]], [[COPY5]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (<2 x s16>), align 1, addrspace 8)
; GFX90A-NEXT: S_ENDPGM 0 ; GFX90A-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg) #0 declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg) #0
declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg) #0 declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg) #0
attributes #0 = { nounwind } attributes #0 = { nounwind }

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.buffer.load.format.f16.ll

Show All 10 Linesdefine amdgpu_ps half @struct_buffer_load_format_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: struct_buffer_load_format_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: struct_buffer_load_format_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN]] ; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.struct.buffer.load.format.f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call half @llvm.amdgcn.struct.buffer.load.format.f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret half %val ret half %val
} }
define amdgpu_ps <2 x half> @struct_buffer_load_format_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x half> @struct_buffer_load_format_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: struct_buffer_load_format_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: struct_buffer_load_format_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN]].sub0 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN]].sub0
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN]].sub1 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN]].sub1
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY9]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY9]], implicit $exec
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_1:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY8]], [[COPY10]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_1:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY8]], [[COPY10]], implicit $exec
; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16
Show All 10 Linesdefine amdgpu_ps <2 x half> @struct_buffer_load_format_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XY_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_XY_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XY_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_XY_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_BOTHEN]] ; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_XY_BOTHEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call <2 x half> @llvm.amdgcn.struct.buffer.load.format.v2f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call <2 x half> @llvm.amdgcn.struct.buffer.load.format.v2f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret <2 x half> %val ret <2 x half> %val
} }
; FIXME: Crashes ; FIXME: Crashes
; define amdgpu_ps <3 x half> @struct_buffer_load_format_v3f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { ; define amdgpu_ps <3 x half> @struct_buffer_load_format_v3f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
Show All 10 Linesdefine amdgpu_ps <4 x half> @struct_buffer_load_format_v4f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub0 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub0
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub1 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub1
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub2 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub2
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub3 ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub3
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535
; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY11]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY11]], implicit $exec
; UNPACKED-NEXT: [[COPY12:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY12:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
Show All 20 Linesdefine amdgpu_ps <4 x half> @struct_buffer_load_format_v4f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN]].sub0 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN]].sub0
; PACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN]].sub1 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN]].sub1
; PACKED-NEXT: $vgpr0 = COPY [[COPY7]] ; PACKED-NEXT: $vgpr0 = COPY [[COPY7]]
; PACKED-NEXT: $vgpr1 = COPY [[COPY8]] ; PACKED-NEXT: $vgpr1 = COPY [[COPY8]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%val = call <4 x half> @llvm.amdgcn.struct.buffer.load.format.v4f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call <4 x half> @llvm.amdgcn.struct.buffer.load.format.v4f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret <4 x half> %val ret <4 x half> %val
} }
Show All 36 Linesdefine amdgpu_ps <4 x half> @struct_buffer_load_format_v4f16__vpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
▲ Show 20 Lines • Show All 57 Lines • ▼ Show 20 Linesdefine amdgpu_ps <4 x half> @struct_buffer_load_format_v4f16__vpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 16 Linesdefine amdgpu_ps half @struct_buffer_load_format_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffsset_add_4095(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset.base, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: struct_buffer_load_format_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffsset_add_4095 ; PACKED-LABEL: name: struct_buffer_load_format_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffsset_add_4095
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN]] ; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
%val = call half @llvm.amdgcn.struct.buffer.load.format.f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call half @llvm.amdgcn.struct.buffer.load.format.f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret half %val ret half %val
} }
define amdgpu_ps half @struct_buffer_load_format_i16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps half @struct_buffer_load_format_i16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: struct_buffer_load_format_i16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: struct_buffer_load_format_i16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; PACKED-LABEL: name: struct_buffer_load_format_i16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: struct_buffer_load_format_i16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN]] ; PACKED-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_D16_X_BOTHEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call i16 @llvm.amdgcn.struct.buffer.load.format.i16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call i16 @llvm.amdgcn.struct.buffer.load.format.i16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%fval = bitcast i16 %val to half %fval = bitcast i16 %val to half
ret half %fval ret half %fval
} }
declare half @llvm.amdgcn.struct.buffer.load.format.f16(<4 x i32>, i32, i32, i32, i32 immarg) #0 declare half @llvm.amdgcn.struct.buffer.load.format.f16(<4 x i32>, i32, i32, i32, i32 immarg) #0
declare <2 x half> @llvm.amdgcn.struct.buffer.load.format.v2f16(<4 x i32>, i32, i32, i32, i32 immarg) #0 declare <2 x half> @llvm.amdgcn.struct.buffer.load.format.v2f16(<4 x i32>, i32, i32, i32, i32 immarg) #0
declare <3 x half> @llvm.amdgcn.struct.buffer.load.format.v3f16(<4 x i32>, i32, i32, i32, i32 immarg) #0 declare <3 x half> @llvm.amdgcn.struct.buffer.load.format.v3f16(<4 x i32>, i32, i32, i32, i32 immarg) #0
declare <4 x half> @llvm.amdgcn.struct.buffer.load.format.v4f16(<4 x i32>, i32, i32, i32, i32 immarg) #0 declare <4 x half> @llvm.amdgcn.struct.buffer.load.format.v4f16(<4 x i32>, i32, i32, i32, i32 immarg) #0
declare i16 @llvm.amdgcn.struct.buffer.load.format.i16(<4 x i32>, i32, i32, i32, i32 immarg) #0 declare i16 @llvm.amdgcn.struct.buffer.load.format.i16(<4 x i32>, i32, i32, i32, i32 immarg) #0
attributes #0 = { nounwind readonly } attributes #0 = { nounwind readonly }

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.buffer.load.format.ll

Show All 9 Linesdefine amdgpu_ps float @struct_buffer_load_format_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_X_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_X_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.struct.buffer.load.format.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.struct.buffer.load.format.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps <2 x float> @struct_buffer_load_format_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x float> @struct_buffer_load_format_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_format_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load_format_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XY_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_XY_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XY_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_XY_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XY_BOTHEN]].sub0 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XY_BOTHEN]].sub0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XY_BOTHEN]].sub1 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XY_BOTHEN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY8]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%val = call <2 x float> @llvm.amdgcn.struct.buffer.load.format.v2f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call <2 x float> @llvm.amdgcn.struct.buffer.load.format.v2f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret <2 x float> %val ret <2 x float> %val
} }
define amdgpu_ps <3 x float> @struct_buffer_load_format_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <3 x float> @struct_buffer_load_format_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_format_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load_format_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZ_BOTHEN:%[0-9]+]]:vreg_96 = BUFFER_LOAD_FORMAT_XYZ_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZ_BOTHEN:%[0-9]+]]:vreg_96 = BUFFER_LOAD_FORMAT_XYZ_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub0 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub1 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub1
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub2 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub2
; CHECK-NEXT: $vgpr0 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY9]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%val = call <3 x float> @llvm.amdgcn.struct.buffer.load.format.v3f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call <3 x float> @llvm.amdgcn.struct.buffer.load.format.v3f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
Show All 9 Linesdefine amdgpu_ps <4 x float> @struct_buffer_load_format_v4f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub0 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub1 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub1
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub2 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub2
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub3 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub3
; CHECK-NEXT: $vgpr0 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY9]]
; CHECK-NEXT: $vgpr3 = COPY [[COPY10]] ; CHECK-NEXT: $vgpr3 = COPY [[COPY10]]
Show All 40 Linesdefine amdgpu_ps <4 x float> @struct_buffer_load_format_v4f32__vpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 20 Linesdefine amdgpu_ps float @struct_buffer_load_format_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffsset_add_4095(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_X_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_X_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
%val = call float @llvm.amdgcn.struct.buffer.load.format.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.struct.buffer.load.format.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @struct_buffer_load_format_i32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @struct_buffer_load_format_i32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_format_i32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load_format_i32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_X_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_FORMAT_X_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call i32 @llvm.amdgcn.struct.buffer.load.format.i32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call i32 @llvm.amdgcn.struct.buffer.load.format.i32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%fval = bitcast i32 %val to float %fval = bitcast i32 %val to float
ret float %fval ret float %fval
} }
define amdgpu_cs void @struct_buffer_load_format_v4i32_tfe(<4 x i32> inreg %rsrc, ptr addrspace(1) %value, ptr addrspace(1) %status) { define amdgpu_cs void @struct_buffer_load_format_v4i32_tfe(<4 x i32> inreg %rsrc, ptr addrspace(1) %value, ptr addrspace(1) %status) {
Show All 9 Linesdefine amdgpu_cs void @struct_buffer_load_format_v4i32_tfe(<4 x i32> inreg %rsrc, ptr addrspace(1) %value, ptr addrspace(1) %status) {
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN:%[0-9]+]]:vreg_160 = BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN [[COPY8]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN:%[0-9]+]]:vreg_160 = BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN [[COPY8]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN]].sub0 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN]].sub0
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN]].sub1 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN]].sub1
; CHECK-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN]].sub2 ; CHECK-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN]].sub2
; CHECK-NEXT: [[COPY12:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN]].sub3 ; CHECK-NEXT: [[COPY12:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN]].sub3
; CHECK-NEXT: [[COPY13:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN]].sub4 ; CHECK-NEXT: [[COPY13:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZW_TFE_IDXEN]].sub4
; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1, [[COPY11]], %subreg.sub2, [[COPY12]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1, [[COPY11]], %subreg.sub2, [[COPY12]], %subreg.sub3
; CHECK-NEXT: FLAT_STORE_DWORDX4 [[REG_SEQUENCE1]], [[REG_SEQUENCE3]], 0, 0, implicit $exec, implicit $flat_scr :: (store (<4 x s32>) into %ir.value, addrspace 1) ; CHECK-NEXT: FLAT_STORE_DWORDX4 [[REG_SEQUENCE1]], [[REG_SEQUENCE3]], 0, 0, implicit $exec, implicit $flat_scr :: (store (<4 x s32>) into %ir.value, addrspace 1)
; CHECK-NEXT: FLAT_STORE_DWORD [[REG_SEQUENCE2]], [[COPY13]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %ir.status, addrspace 1) ; CHECK-NEXT: FLAT_STORE_DWORD [[REG_SEQUENCE2]], [[COPY13]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %ir.status, addrspace 1)
Show All 22 Linesdefine amdgpu_cs void @struct_buffer_load_format_v3i32_tfe(<4 x i32> inreg %rsrc, ptr addrspace(1) %value, ptr addrspace(1) %status) {
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN [[COPY8]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN [[COPY8]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN]].sub0 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN]].sub0
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN]].sub1 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN]].sub1
; CHECK-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN]].sub2 ; CHECK-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN]].sub2
; CHECK-NEXT: [[COPY12:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN]].sub3 ; CHECK-NEXT: [[COPY12:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_XYZ_TFE_IDXEN]].sub3
; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1, [[COPY11]], %subreg.sub2 ; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1, [[COPY11]], %subreg.sub2
; CHECK-NEXT: FLAT_STORE_DWORDX3 [[REG_SEQUENCE1]], [[REG_SEQUENCE3]], 0, 0, implicit $exec, implicit $flat_scr :: (store (<3 x s32>) into %ir.value, align 16, addrspace 1) ; CHECK-NEXT: FLAT_STORE_DWORDX3 [[REG_SEQUENCE1]], [[REG_SEQUENCE3]], 0, 0, implicit $exec, implicit $flat_scr :: (store (<3 x s32>) into %ir.value, align 16, addrspace 1)
; CHECK-NEXT: FLAT_STORE_DWORD [[REG_SEQUENCE2]], [[COPY12]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %ir.status, addrspace 1) ; CHECK-NEXT: FLAT_STORE_DWORD [[REG_SEQUENCE2]], [[COPY12]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %ir.status, addrspace 1)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
Show All 21 Linesdefine amdgpu_cs void @struct_buffer_load_format_i32_tfe(<4 x i32> inreg %rsrc, ptr addrspace(1) %value, ptr addrspace(1) %status) {
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_TFE_IDXEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_X_TFE_IDXEN [[COPY8]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_TFE_IDXEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_FORMAT_X_TFE_IDXEN [[COPY8]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_X_TFE_IDXEN]].sub0 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_X_TFE_IDXEN]].sub0
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_X_TFE_IDXEN]].sub1 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_FORMAT_X_TFE_IDXEN]].sub1
; CHECK-NEXT: FLAT_STORE_DWORD [[REG_SEQUENCE1]], [[COPY9]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %ir.value, addrspace 1) ; CHECK-NEXT: FLAT_STORE_DWORD [[REG_SEQUENCE1]], [[COPY9]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %ir.value, addrspace 1)
; CHECK-NEXT: FLAT_STORE_DWORD [[REG_SEQUENCE2]], [[COPY10]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %ir.status, addrspace 1) ; CHECK-NEXT: FLAT_STORE_DWORD [[REG_SEQUENCE2]], [[COPY10]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into %ir.status, addrspace 1)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%load = call { i32, i32 } @llvm.amdgcn.struct.buffer.load.format.sl_i32i32s(<4 x i32> %rsrc, i32 0, i32 0, i32 0, i32 0) %load = call { i32, i32 } @llvm.amdgcn.struct.buffer.load.format.sl_i32i32s(<4 x i32> %rsrc, i32 0, i32 0, i32 0, i32 0)
%v = extractvalue { i32, i32 } %load, 0 %v = extractvalue { i32, i32 } %load, 0
Show All 18 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.buffer.load.ll

Show All 10 Linesdefine amdgpu_ps float @struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
; Natural mapping ; Natural mapping
define amdgpu_ps <2 x float> @struct_buffer_load_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x float> @struct_buffer_load_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORDX2_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORDX2_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_BOTHEN]].sub0 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_BOTHEN]].sub0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_BOTHEN]].sub1 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_BOTHEN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY8]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%val = call <2 x float> @llvm.amdgcn.struct.buffer.load.v2f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call <2 x float> @llvm.amdgcn.struct.buffer.load.v2f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret <2 x float> %val ret <2 x float> %val
} }
; Natural mapping ; Natural mapping
define amdgpu_ps <3 x float> @struct_buffer_load_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <3 x float> @struct_buffer_load_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORDX3_BOTHEN:%[0-9]+]]:vreg_96 = BUFFER_LOAD_DWORDX3_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORDX3_BOTHEN:%[0-9]+]]:vreg_96 = BUFFER_LOAD_DWORDX3_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_BOTHEN]].sub0 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_BOTHEN]].sub0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_BOTHEN]].sub1 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_BOTHEN]].sub1
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_BOTHEN]].sub2 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX3_BOTHEN]].sub2
; CHECK-NEXT: $vgpr0 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY9]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%val = call <3 x float> @llvm.amdgcn.struct.buffer.load.v3f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call <3 x float> @llvm.amdgcn.struct.buffer.load.v3f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
Show All 10 Linesdefine amdgpu_ps <4 x float> @struct_buffer_load_v4f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORDX4_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORDX4_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_BOTHEN]].sub0 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_BOTHEN]].sub0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_BOTHEN]].sub1 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_BOTHEN]].sub1
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_BOTHEN]].sub2 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_BOTHEN]].sub2
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_BOTHEN]].sub3 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_BOTHEN]].sub3
; CHECK-NEXT: $vgpr0 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY9]]
; CHECK-NEXT: $vgpr3 = COPY [[COPY10]] ; CHECK-NEXT: $vgpr3 = COPY [[COPY10]]
Show All 13 Linesdefine amdgpu_ps float @struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_vindex0(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY4]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY4]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY5]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 0, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 0, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
; Natural mapping ; Natural mapping
define amdgpu_ps float @struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffset_add4095(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset.base, i32 inreg %soffset) { define amdgpu_ps float @struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffset_add4095(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffset_add4095 ; CHECK-LABEL: name: struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffset_add4095
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
%val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps float @struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_soffset_64(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset) { define amdgpu_ps float @struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_soffset_64(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset) {
; CHECK-LABEL: name: struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_soffset_64 ; CHECK-LABEL: name: struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_soffset_64
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 64 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 64
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 64, i32 0) %val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 64, i32 0)
ret float %val ret float %val
} }
; Need to legalize all reg operands ; Need to legalize all reg operands
define amdgpu_ps float @struct_buffer_load_f32__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps float @struct_buffer_load_f32__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
Show All 33 Linesdefine amdgpu_ps float @struct_buffer_load_f32__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 13 Linesdefine amdgpu_ps float @struct_buffer_load_i8_zext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_UBYTE_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_UBYTE_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s8), addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_UBYTE_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_UBYTE_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s8), addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_UBYTE_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_UBYTE_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call i8 @llvm.amdgcn.struct.buffer.load.i8(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call i8 @llvm.amdgcn.struct.buffer.load.i8(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%ext = zext i8 %val to i32 %ext = zext i8 %val to i32
%cast = bitcast i32 %ext to float %cast = bitcast i32 %ext to float
ret float %cast ret float %cast
} }
define amdgpu_ps float @struct_buffer_load_i8_sext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @struct_buffer_load_i8_sext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_i8_sext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load_i8_sext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_SBYTE_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_SBYTE_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s8), addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_SBYTE_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_SBYTE_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s8), addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_SBYTE_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_SBYTE_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call i8 @llvm.amdgcn.struct.buffer.load.i8(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call i8 @llvm.amdgcn.struct.buffer.load.i8(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%ext = sext i8 %val to i32 %ext = sext i8 %val to i32
%cast = bitcast i32 %ext to float %cast = bitcast i32 %ext to float
ret float %cast ret float %cast
} }
define amdgpu_ps float @struct_buffer_load_i16_zext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @struct_buffer_load_i16_zext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_i16_zext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load_i16_zext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_USHORT_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_USHORT_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_USHORT_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_USHORT_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_USHORT_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_USHORT_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call i16 @llvm.amdgcn.struct.buffer.load.i16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call i16 @llvm.amdgcn.struct.buffer.load.i16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%ext = zext i16 %val to i32 %ext = zext i16 %val to i32
%cast = bitcast i32 %ext to float %cast = bitcast i32 %ext to float
ret float %cast ret float %cast
} }
define amdgpu_ps float @struct_buffer_load_i16_sext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @struct_buffer_load_i16_sext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_i16_sext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load_i16_sext__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_SSHORT_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_SSHORT_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_SSHORT_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_SSHORT_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_SSHORT_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_SSHORT_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call i16 @llvm.amdgcn.struct.buffer.load.i16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call i16 @llvm.amdgcn.struct.buffer.load.i16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
%ext = sext i16 %val to i32 %ext = sext i16 %val to i32
%cast = bitcast i32 %ext to float %cast = bitcast i32 %ext to float
ret float %cast ret float %cast
} }
; Natural mapping ; Natural mapping
define amdgpu_ps half @struct_buffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps half @struct_buffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_USHORT_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_USHORT_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_USHORT_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_USHORT_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_USHORT_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_USHORT_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.struct.buffer.load.f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call half @llvm.amdgcn.struct.buffer.load.f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret half %val ret half %val
} }
; Natural mapping ; Natural mapping
define amdgpu_ps <2 x half> @struct_buffer_load_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x half> @struct_buffer_load_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call <2 x half> @llvm.amdgcn.struct.buffer.load.v2f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call <2 x half> @llvm.amdgcn.struct.buffer.load.v2f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret <2 x half> %val ret <2 x half> %val
} }
; FIXME: Crashes ; FIXME: Crashes
; define amdgpu_ps <3 x half> @struct_buffer_load_v3f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { ; define amdgpu_ps <3 x half> @struct_buffer_load_v3f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
Show All 11 Linesdefine amdgpu_ps <4 x half> @struct_buffer_load_v4f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORDX2_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORDX2_BOTHEN:%[0-9]+]]:vreg_64 = BUFFER_LOAD_DWORDX2_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_BOTHEN]].sub0 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_BOTHEN]].sub0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_BOTHEN]].sub1 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_LOAD_DWORDX2_BOTHEN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY8]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%val = call <4 x half> @llvm.amdgcn.struct.buffer.load.v4f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call <4 x half> @llvm.amdgcn.struct.buffer.load.v4f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret <4 x half> %val ret <4 x half> %val
} }
; Natural mapping + glc ; Natural mapping + glc
define amdgpu_ps float @struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_glc(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_glc(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_glc ; CHECK-LABEL: name: struct_buffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_glc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_BOTHEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 1, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[BUFFER_LOAD_DWORD_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 1) %val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 1)
ret float %val ret float %val
} }
declare i8 @llvm.amdgcn.struct.buffer.load.i8(<4 x i32>, i32, i32, i32, i32 immarg) #0 declare i8 @llvm.amdgcn.struct.buffer.load.i8(<4 x i32>, i32, i32, i32, i32 immarg) #0
declare i16 @llvm.amdgcn.struct.buffer.load.i16(<4 x i32>, i32, i32, i32, i32 immarg) #0 declare i16 @llvm.amdgcn.struct.buffer.load.i16(<4 x i32>, i32, i32, i32, i32 immarg) #0
Show All 10 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.buffer.store.format.f16.ll

Show All 11 Linesdefine amdgpu_ps void @struct_buffer_store_format_f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(half %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_gfx80_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_gfx80_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: struct_buffer_store_format_f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: struct_buffer_store_format_f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; PACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.format.f16(half %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.format.f16(half %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_format_v2f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store_format_v2f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-LABEL: name: struct_buffer_store_format_v2f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: struct_buffer_store_format_v2f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY8]], [[COPY]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY8]], [[COPY]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1
; UNPACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_BOTHEN_exact [[REG_SEQUENCE1]], [[REG_SEQUENCE2]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_gfx80_BOTHEN_exact [[REG_SEQUENCE1]], [[REG_SEQUENCE2]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: struct_buffer_store_format_v2f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: struct_buffer_store_format_v2f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; PACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XY_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.format.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; FIXME: ; FIXME:
; define amdgpu_ps void @struct_buffer_store_format_v3f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<3 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { ; define amdgpu_ps void @struct_buffer_store_format_v3f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<3 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; call void @llvm.amdgcn.struct.buffer.store.format.v3f16(<3 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) ; call void @llvm.amdgcn.struct.buffer.store.format.v3f16(<3 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
Show All 17 Linesdefine amdgpu_ps void @struct_buffer_store_format_v4f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 16
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY9]], [[COPY]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY9]], [[COPY]], implicit $exec
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_1:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY10]], [[COPY1]], implicit $exec ; UNPACKED-NEXT: [[V_LSHRREV_B32_e64_1:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 [[COPY10]], [[COPY1]], implicit $exec
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1, [[COPY1]], %subreg.sub2, [[V_LSHRREV_B32_e64_1]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[V_LSHRREV_B32_e64_]], %subreg.sub1, [[COPY1]], %subreg.sub2, [[V_LSHRREV_B32_e64_1]], %subreg.sub3
; UNPACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_gfx80_BOTHEN_exact [[REG_SEQUENCE1]], [[REG_SEQUENCE2]], [[REG_SEQUENCE]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_gfx80_BOTHEN_exact [[REG_SEQUENCE1]], [[REG_SEQUENCE2]], [[REG_SEQUENCE]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: struct_buffer_store_format_v4f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: struct_buffer_store_format_v4f16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; PACKED-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_XYZW_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.format.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.format.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_format_f16__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(half inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps void @struct_buffer_store_format_f16__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(half inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; UNPACKED-LABEL: name: struct_buffer_store_format_f16__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset ; UNPACKED-LABEL: name: struct_buffer_store_format_f16__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
Show All 33 Linesdefine amdgpu_ps void @struct_buffer_store_format_f16__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(half inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_gfx80_BOTHEN_exact [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_gfx80_BOTHEN_exact [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 37 Linesdefine amdgpu_ps void @struct_buffer_store_format_f16__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(half inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_BOTHEN_exact [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_BOTHEN_exact [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 13 Linesdefine amdgpu_ps void @struct_buffer_store_format_i16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(i16 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_gfx80_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; UNPACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_gfx80_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; UNPACKED-NEXT: S_ENDPGM 0 ; UNPACKED-NEXT: S_ENDPGM 0
; PACKED-LABEL: name: struct_buffer_store_format_i16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: struct_buffer_store_format_i16__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; PACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; PACKED-NEXT: BUFFER_STORE_FORMAT_D16_X_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; PACKED-NEXT: S_ENDPGM 0 ; PACKED-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.format.i16(i16 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.format.i16(i16 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
declare void @llvm.amdgcn.struct.buffer.store.format.f16(half, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.format.f16(half, <4 x i32>, i32, i32, i32, i32 immarg)
declare void @llvm.amdgcn.struct.buffer.store.format.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.format.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg)
declare void @llvm.amdgcn.struct.buffer.store.format.v3f16(<3 x half>, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.format.v3f16(<3 x half>, <4 x i32>, i32, i32, i32, i32 immarg)
declare void @llvm.amdgcn.struct.buffer.store.format.v4f16(<4 x half>, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.format.v4f16(<4 x half>, <4 x i32>, i32, i32, i32, i32 immarg)
declare void @llvm.amdgcn.struct.buffer.store.format.i16(i16, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.format.i16(i16, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.buffer.store.format.f32.ll

Show All 10 Linesdefine amdgpu_ps void @struct_buffer_store_format_f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_FORMAT_X_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_X_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.format.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.format.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_format_v2f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<2 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store_format_v2f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<2 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store_format_v2f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_store_format_v2f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XY_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.format.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_format_v3f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<3 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store_format_v3f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<3 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store_format_v3f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_store_format_v3f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY3]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY5]], %subreg.sub2, [[COPY6]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY3]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY5]], %subreg.sub2, [[COPY6]], %subreg.sub3
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZ_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZ_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.format.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.format.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_format_v4f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store_format_v4f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store_format_v4f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_store_format_v4f32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY $vgpr5 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY $vgpr5
; CHECK-NEXT: [[COPY10:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY9]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY9]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZW_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY10]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_XYZW_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY10]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.format.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.format.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_format_f32__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps void @struct_buffer_store_format_f32__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-LABEL: name: struct_buffer_store_format_f32__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset ; CHECK-LABEL: name: struct_buffer_store_format_f32__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
Show All 33 Linesdefine amdgpu_ps void @struct_buffer_store_format_f32__sgpr_val__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(float inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY7]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY9]], %subreg.sub0, [[COPY10]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_FORMAT_X_BOTHEN_exact [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_X_BOTHEN_exact [[COPY8]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 13 Linesdefine amdgpu_ps void @struct_buffer_store_format_i32__vgpr_val__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_FORMAT_X_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_FORMAT_X_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.format.i32(i32 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.format.i32(i32 %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
declare void @llvm.amdgcn.struct.buffer.store.format.f32(float, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.format.f32(float, <4 x i32>, i32, i32, i32, i32 immarg)
declare void @llvm.amdgcn.struct.buffer.store.format.v2f32(<2 x float>, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.format.v2f32(<2 x float>, <4 x i32>, i32, i32, i32, i32 immarg)
declare void @llvm.amdgcn.struct.buffer.store.format.v3f32(<3 x float>, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.format.v3f32(<3 x float>, <4 x i32>, i32, i32, i32, i32 immarg)
declare void @llvm.amdgcn.struct.buffer.store.format.v4f32(<4 x float>, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.format.v4f32(<4 x float>, <4 x i32>, i32, i32, i32, i32 immarg)
declare void @llvm.amdgcn.struct.buffer.store.format.i32(i32, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.format.i32(i32, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.buffer.store.ll

Show All 11 Linesdefine amdgpu_ps void @struct_buffer_store_f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_DWORD_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_v2f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<2 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store_v2f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<2 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store_v2f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_store_v2f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_DWORDX2_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORDX2_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.v2f32(<2 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_v3f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<3 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store_v3f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<3 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store_v3f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_store_v3f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_96 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY3]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY5]], %subreg.sub2, [[COPY6]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY3]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY5]], %subreg.sub2, [[COPY6]], %subreg.sub3
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_DWORDX3_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORDX3_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY9]], 0, 0, 0, implicit $exec :: (dereferenceable store (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.v3f32(<3 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_v4f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store_v4f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x float> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store_v4f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_store_v4f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY6]], %subreg.sub2, [[COPY7]], %subreg.sub3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr4
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY $vgpr5 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY $vgpr5
; CHECK-NEXT: [[COPY10:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY9]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY9]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_DWORDX4_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY10]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORDX4_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY10]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.v4f32(<4 x float> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_v4f32_vgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x float> inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps void @struct_buffer_store_v4f32_vgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x float> inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-LABEL: name: struct_buffer_store_v4f32_vgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__vgpr_soffset ; CHECK-LABEL: name: struct_buffer_store_v4f32_vgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__vgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
Show All 37 Linesdefine amdgpu_ps void @struct_buffer_store_v4f32_vgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x float> inreg %val, <4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY10]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY10]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY12]], %subreg.sub0, [[COPY13]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE3:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY12]], %subreg.sub0, [[COPY13]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_DWORDX4_BOTHEN_exact [[COPY11]], [[REG_SEQUENCE3]], [[REG_SEQUENCE2]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORDX4_BOTHEN_exact [[COPY11]], [[REG_SEQUENCE3]], [[REG_SEQUENCE2]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 13 Linesdefine amdgpu_ps void @struct_buffer_store_i8_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_BYTE_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 7) ; CHECK-NEXT: BUFFER_STORE_BYTE_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s8), addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%val.trunc = trunc i32 %val to i8 %val.trunc = trunc i32 %val to i8
call void @llvm.amdgcn.struct.buffer.store.i8(i8 %val.trunc, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.i8(i8 %val.trunc, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_i16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store_i16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(i32 %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store_i16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_store_i16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_SHORT_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_SHORT_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (s16), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
%val.trunc = trunc i32 %val to i16 %val.trunc = trunc i32 %val to i16
call void @llvm.amdgcn.struct.buffer.store.i16(i16 %val.trunc, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.i16(i16 %val.trunc, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset_glc(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store_f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset_glc(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store_f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset_glc ; CHECK-LABEL: name: struct_buffer_store_f32_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset_glc
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_DWORD_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 1, 0, implicit $exec :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 1) call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 1)
ret void ret void
} }
define amdgpu_ps void @struct_buffer_store_v2f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store_v2f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store_v2f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_store_v2f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1, $vgpr2
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY2]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY4]], %subreg.sub3
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY6]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_DWORD_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORD_BOTHEN_exact [[COPY]], [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY7]], 0, 0, 0, implicit $exec :: (dereferenceable store (<2 x s16>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; FIXME: ; FIXME:
; define amdgpu_ps void @struct_buffer_store_v3f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<3 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { ; define amdgpu_ps void @struct_buffer_store_v3f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<3 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; call void @llvm.amdgcn.struct.buffer.store.v3f16(<3 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) ; call void @llvm.amdgcn.struct.buffer.store.v3f16(<3 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
Show All 12 Linesdefine amdgpu_ps void @struct_buffer_store_v4f16_sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY5]], %subreg.sub3
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY7]], %subreg.sub1
; CHECK-NEXT: BUFFER_STORE_DWORDX2_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 7) ; CHECK-NEXT: BUFFER_STORE_DWORDX2_BOTHEN_exact [[REG_SEQUENCE]], [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[COPY8]], 0, 0, 0, implicit $exec :: (dereferenceable store (<4 x s16>), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.v4f16(<4 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
declare void @llvm.amdgcn.struct.buffer.store.i8(i8, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.i8(i8, <4 x i32>, i32, i32, i32, i32 immarg)
declare void @llvm.amdgcn.struct.buffer.store.i16(i16, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.i16(i16, <4 x i32>, i32, i32, i32, i32 immarg)
Show All 9 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.tbuffer.load.f16.ll

Show All 12 Linesdefine amdgpu_ps half @struct_tbuffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN]] ; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; UNPACKED-LABEL: name: struct_tbuffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: struct_tbuffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.struct.tbuffer.load.f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call half @llvm.amdgcn.struct.tbuffer.load.f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret half %val ret half %val
} }
define amdgpu_ps <2 x half> @struct_tbuffer_load_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x half> @struct_tbuffer_load_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; PACKED-LABEL: name: struct_tbuffer_load_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; PACKED-LABEL: name: struct_tbuffer_load_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; PACKED: bb.1 (%ir-block.0): ; PACKED: bb.1 (%ir-block.0):
; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; PACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; PACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XY_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_XY_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XY_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_XY_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_BOTHEN]] ; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_BOTHEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; UNPACKED-LABEL: name: struct_tbuffer_load_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: struct_tbuffer_load_v2f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN]].sub0 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN]].sub0
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN]].sub1 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XY_gfx80_BOTHEN]].sub1
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY9]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY9]], implicit $exec
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_1:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY8]], [[COPY10]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_1:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY8]], [[COPY10]], implicit $exec
; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16 ; UNPACKED-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 16
Show All 21 Linesdefine amdgpu_ps <4 x half> @struct_tbuffer_load_v4f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN]].sub0 ; PACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN]].sub0
; PACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN]].sub1 ; PACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN]].sub1
; PACKED-NEXT: $vgpr0 = COPY [[COPY7]] ; PACKED-NEXT: $vgpr0 = COPY [[COPY7]]
; PACKED-NEXT: $vgpr1 = COPY [[COPY8]] ; PACKED-NEXT: $vgpr1 = COPY [[COPY8]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
; UNPACKED-LABEL: name: struct_tbuffer_load_v4f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; UNPACKED-LABEL: name: struct_tbuffer_load_v4f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub0 ; UNPACKED-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub0
; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub1 ; UNPACKED-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub1
; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub2 ; UNPACKED-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub2
; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub3 ; UNPACKED-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN]].sub3
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 65535
; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY11]], implicit $exec ; UNPACKED-NEXT: [[V_AND_B32_e64_:%[0-9]+]]:vgpr_32 = V_AND_B32_e64 [[COPY7]], [[COPY11]], implicit $exec
; UNPACKED-NEXT: [[COPY12:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY12:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
Show All 26 Linesdefine amdgpu_ps half @struct_tbuffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_vindex0(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 ; PACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; PACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY4]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY4]], %subreg.sub1
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN]] ; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; UNPACKED-LABEL: name: struct_tbuffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_vindex0 ; UNPACKED-LABEL: name: struct_tbuffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_vindex0
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 ; UNPACKED-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY4]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY4]], %subreg.sub1
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call half @llvm.amdgcn.struct.tbuffer.load.f16(<4 x i32> %rsrc, i32 0, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call half @llvm.amdgcn.struct.tbuffer.load.f16(<4 x i32> %rsrc, i32 0, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret half %val ret half %val
} }
define amdgpu_ps <4 x half> @struct_tbuffer_load_v4f16__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps <4 x half> @struct_tbuffer_load_v4f16__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; PACKED-LABEL: name: struct_tbuffer_load_v4f16__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset ; PACKED-LABEL: name: struct_tbuffer_load_v4f16__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset
Show All 32 Linesdefine amdgpu_ps <4 x half> @struct_tbuffer_load_v4f16__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; PACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; PACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; PACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.3: ; PACKED-NEXT: bb.3:
; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; PACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_D16_XYZW_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; PACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; PACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: bb.4: ; PACKED-NEXT: bb.4:
; PACKED-NEXT: successors: %bb.5(0x80000000) ; PACKED-NEXT: successors: %bb.5(0x80000000)
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; PACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; PACKED-NEXT: {{ $}} ; PACKED-NEXT: {{ $}}
Show All 39 Linesdefine amdgpu_ps <4 x half> @struct_tbuffer_load_v4f16__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; UNPACKED-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; UNPACKED-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; UNPACKED-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.3: ; UNPACKED-NEXT: bb.3:
; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; UNPACKED-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_D16_XYZW_gfx80_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s16>), align 1, addrspace 8)
; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; UNPACKED-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; UNPACKED-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: bb.4: ; UNPACKED-NEXT: bb.4:
; UNPACKED-NEXT: successors: %bb.5(0x80000000) ; UNPACKED-NEXT: successors: %bb.5(0x80000000)
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; UNPACKED-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
Show All 34 Linesdefine amdgpu_ps half @struct_tbuffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffset_add4095(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset.base, i32 inreg %soffset) {
; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; PACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; PACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; PACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; PACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; PACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; PACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; PACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; PACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; PACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN]] ; PACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_BOTHEN]]
; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; PACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
; UNPACKED-LABEL: name: struct_tbuffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffset_add4095 ; UNPACKED-LABEL: name: struct_tbuffer_load_f16__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffset_add4095
; UNPACKED: bb.1 (%ir-block.0): ; UNPACKED: bb.1 (%ir-block.0):
; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; UNPACKED-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; UNPACKED-NEXT: {{ $}} ; UNPACKED-NEXT: {{ $}}
; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; UNPACKED-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; UNPACKED-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; UNPACKED-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; UNPACKED-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; UNPACKED-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; UNPACKED-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; UNPACKED-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; UNPACKED-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; UNPACKED-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 7) ; UNPACKED-NEXT: [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 78, 0, 0, implicit $exec :: (dereferenceable load (s16), align 1, addrspace 8)
; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]] ; UNPACKED-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_D16_X_gfx80_BOTHEN]]
; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; UNPACKED-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
%val = call half @llvm.amdgcn.struct.tbuffer.load.f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call half @llvm.amdgcn.struct.tbuffer.load.f16(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret half %val ret half %val
} }
declare half @llvm.amdgcn.struct.tbuffer.load.f16(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0 declare half @llvm.amdgcn.struct.tbuffer.load.f16(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0
declare <2 x half> @llvm.amdgcn.struct.tbuffer.load.v2f16(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0 declare <2 x half> @llvm.amdgcn.struct.tbuffer.load.v2f16(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0
declare <3 x half> @llvm.amdgcn.struct.tbuffer.load.v3f16(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0 declare <3 x half> @llvm.amdgcn.struct.tbuffer.load.v3f16(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0
declare <4 x half> @llvm.amdgcn.struct.tbuffer.load.v4f16(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0 declare <4 x half> @llvm.amdgcn.struct.tbuffer.load.v4f16(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0
attributes #0 = { nounwind readonly } attributes #0 = { nounwind readonly }

llvm/test/CodeGen/AMDGPU/GlobalISel/llvm.amdgcn.struct.tbuffer.load.ll

Show All 11 Linesdefine amdgpu_ps float @struct_tbuffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.struct.tbuffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call float @llvm.amdgcn.struct.tbuffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps <2 x float> @struct_tbuffer_load_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x float> @struct_tbuffer_load_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_tbuffer_load_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_tbuffer_load_v2f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XY_BOTHEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_XY_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XY_BOTHEN:%[0-9]+]]:vreg_64 = TBUFFER_LOAD_FORMAT_XY_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<2 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XY_BOTHEN]].sub0 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XY_BOTHEN]].sub0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XY_BOTHEN]].sub1 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XY_BOTHEN]].sub1
; CHECK-NEXT: $vgpr0 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY8]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1
%val = call <2 x float> @llvm.amdgcn.struct.tbuffer.load.v2f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call <2 x float> @llvm.amdgcn.struct.tbuffer.load.v2f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret <2 x float> %val ret <2 x float> %val
} }
define amdgpu_ps <3 x float> @struct_tbuffer_load_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <3 x float> @struct_tbuffer_load_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_tbuffer_load_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_tbuffer_load_v3f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $vgpr0, $vgpr1
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XYZ_BOTHEN:%[0-9]+]]:vreg_96 = TBUFFER_LOAD_FORMAT_XYZ_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XYZ_BOTHEN:%[0-9]+]]:vreg_96 = TBUFFER_LOAD_FORMAT_XYZ_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<3 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub0 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub1 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub1
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub2 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZ_BOTHEN]].sub2
; CHECK-NEXT: $vgpr0 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY9]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0, implicit $vgpr1, implicit $vgpr2
%val = call <3 x float> @llvm.amdgcn.struct.tbuffer.load.v3f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call <3 x float> @llvm.amdgcn.struct.tbuffer.load.v3f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0)
Show All 9 Linesdefine amdgpu_ps <4 x float> @struct_tbuffer_load_v4f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub0 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub0
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub1 ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub1
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub2 ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub2
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub3 ; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN]].sub3
; CHECK-NEXT: $vgpr0 = COPY [[COPY7]] ; CHECK-NEXT: $vgpr0 = COPY [[COPY7]]
; CHECK-NEXT: $vgpr1 = COPY [[COPY8]] ; CHECK-NEXT: $vgpr1 = COPY [[COPY8]]
; CHECK-NEXT: $vgpr2 = COPY [[COPY9]] ; CHECK-NEXT: $vgpr2 = COPY [[COPY9]]
; CHECK-NEXT: $vgpr3 = COPY [[COPY10]] ; CHECK-NEXT: $vgpr3 = COPY [[COPY10]]
Show All 12 Linesdefine amdgpu_ps float @struct_tbuffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_vindex0(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 ; CHECK-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY4]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY4]], %subreg.sub1
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY5]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.struct.tbuffer.load.f32(<4 x i32> %rsrc, i32 0, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call float @llvm.amdgcn.struct.tbuffer.load.f32(<4 x i32> %rsrc, i32 0, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret float %val ret float %val
} }
define amdgpu_ps <4 x float> @struct_tbuffer_load_v4f32__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) { define amdgpu_ps <4 x float> @struct_tbuffer_load_v4f32__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-LABEL: name: struct_tbuffer_load_v4f32__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset ; CHECK-LABEL: name: struct_tbuffer_load_v4f32__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset
Show All 32 Linesdefine amdgpu_ps <4 x float> @struct_tbuffer_load_v4f32__vgpr_rsrc__sgpr_vindex__sgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_64_xexec = V_CMP_EQ_U32_e64 [[V_READFIRSTLANE_B32_4]], [[COPY6]], implicit $exec
; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B64_1:%[0-9]+]]:sreg_64_xexec = S_AND_B64 [[S_AND_B64_]], [[V_CMP_EQ_U32_e64_]], implicit-def dead $scc
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[S_AND_B64_1]], implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY8]], %subreg.sub1
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = TBUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE2]], [[REG_SEQUENCE1]], [[V_READFIRSTLANE_B32_4]], 0, 78, 0, 0, implicit $exec :: (dereferenceable load (<4 x s32>), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 20 Linesdefine amdgpu_ps float @struct_tbuffer_load_f32__sgpr_rsrc__vgpr_vindex__vgpr_voffset__sgpr_soffset_voffset_add4095(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset.base, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY $sgpr5
; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; CHECK-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY1]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY3]], %subreg.sub3
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY $sgpr6
; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; CHECK-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY5]], %subreg.sub1
; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 78, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[TBUFFER_LOAD_FORMAT_X_BOTHEN:%[0-9]+]]:vgpr_32 = TBUFFER_LOAD_FORMAT_X_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE]], [[COPY6]], 4095, 78, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_BOTHEN]] ; CHECK-NEXT: $vgpr0 = COPY [[TBUFFER_LOAD_FORMAT_X_BOTHEN]]
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%voffset = add i32 %voffset.base, 4095 %voffset = add i32 %voffset.base, 4095
%val = call float @llvm.amdgcn.struct.tbuffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0) %val = call float @llvm.amdgcn.struct.tbuffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 78, i32 0)
ret float %val ret float %val
} }
declare float @llvm.amdgcn.struct.tbuffer.load.f32(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0 declare float @llvm.amdgcn.struct.tbuffer.load.f32(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0
declare <2 x float> @llvm.amdgcn.struct.tbuffer.load.v2f32(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0 declare <2 x float> @llvm.amdgcn.struct.tbuffer.load.v2f32(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0
declare <3 x float> @llvm.amdgcn.struct.tbuffer.load.v3f32(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0 declare <3 x float> @llvm.amdgcn.struct.tbuffer.load.v3f32(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0
declare <4 x float> @llvm.amdgcn.struct.tbuffer.load.v4f32(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0 declare <4 x float> @llvm.amdgcn.struct.tbuffer.load.v4f32(<4 x i32>, i32, i32, i32, i32 immarg, i32 immarg) #0
attributes #0 = { nounwind readonly } attributes #0 = { nounwind readonly }

llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-amdgcn.image.load.1d.ll

Show All 13 Linesdefine amdgpu_ps void @load_1d_vgpr_vaddr__sgpr_srsrc(<8 x i32> inreg %rsrc, i32 %s) {
; FAST-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; FAST-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; FAST-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; FAST-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; FAST-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7 ; FAST-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7
; FAST-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8 ; FAST-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8
; FAST-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9 ; FAST-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9
; FAST-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; FAST-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; FAST-NEXT: [[COPY8:%[0-9]+]]:vgpr(s32) = COPY $vgpr0 ; FAST-NEXT: [[COPY8:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
; FAST-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF ; FAST-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF
; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY8]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY8]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; FAST-NEXT: [[COPY9:%[0-9]+]]:vgpr(p1) = COPY [[DEF]](p1) ; FAST-NEXT: [[COPY9:%[0-9]+]]:vgpr(p1) = COPY [[DEF]](p1)
; FAST-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[COPY9]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1) ; FAST-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[COPY9]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1)
; FAST-NEXT: S_ENDPGM 0 ; FAST-NEXT: S_ENDPGM 0
; GREEDY-LABEL: name: load_1d_vgpr_vaddr__sgpr_srsrc ; GREEDY-LABEL: name: load_1d_vgpr_vaddr__sgpr_srsrc
; GREEDY: bb.1 (%ir-block.0): ; GREEDY: bb.1 (%ir-block.0):
; GREEDY-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0 ; GREEDY-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $vgpr0
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2 ; GREEDY-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; GREEDY-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3 ; GREEDY-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; GREEDY-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4 ; GREEDY-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
; GREEDY-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; GREEDY-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; GREEDY-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; GREEDY-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; GREEDY-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7 ; GREEDY-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7
; GREEDY-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8 ; GREEDY-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8
; GREEDY-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9 ; GREEDY-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9
; GREEDY-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GREEDY-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GREEDY-NEXT: [[COPY8:%[0-9]+]]:vgpr(s32) = COPY $vgpr0 ; GREEDY-NEXT: [[COPY8:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
; GREEDY-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF ; GREEDY-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF
; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY8]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY8]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GREEDY-NEXT: [[COPY9:%[0-9]+]]:vgpr(p1) = COPY [[DEF]](p1) ; GREEDY-NEXT: [[COPY9:%[0-9]+]]:vgpr(p1) = COPY [[DEF]](p1)
; GREEDY-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[COPY9]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1) ; GREEDY-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[COPY9]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1)
; GREEDY-NEXT: S_ENDPGM 0 ; GREEDY-NEXT: S_ENDPGM 0
%v = call <4 x float> @llvm.amdgcn.image.load.1d.v4f32.i32(i32 15, i32 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.load.1d.v4f32.i32(i32 15, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
store <4 x float> %v, ptr addrspace(1) undef store <4 x float> %v, ptr addrspace(1) undef
ret void ret void
} }
Show All 10 Linesdefine amdgpu_ps void @load_1d_sgpr_vaddr__sgpr_srsrc(<8 x i32> inreg %rsrc, i32 inreg %s) {
; FAST-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; FAST-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; FAST-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7 ; FAST-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7
; FAST-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8 ; FAST-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8
; FAST-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9 ; FAST-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9
; FAST-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; FAST-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; FAST-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10 ; FAST-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10
; FAST-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF ; FAST-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF
; FAST-NEXT: [[COPY9:%[0-9]+]]:vgpr(s32) = COPY [[COPY8]](s32) ; FAST-NEXT: [[COPY9:%[0-9]+]]:vgpr(s32) = COPY [[COPY8]](s32)
; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY9]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY9]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; FAST-NEXT: [[COPY10:%[0-9]+]]:vgpr(p1) = COPY [[DEF]](p1) ; FAST-NEXT: [[COPY10:%[0-9]+]]:vgpr(p1) = COPY [[DEF]](p1)
; FAST-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[COPY10]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1) ; FAST-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[COPY10]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1)
; FAST-NEXT: S_ENDPGM 0 ; FAST-NEXT: S_ENDPGM 0
; GREEDY-LABEL: name: load_1d_sgpr_vaddr__sgpr_srsrc ; GREEDY-LABEL: name: load_1d_sgpr_vaddr__sgpr_srsrc
; GREEDY: bb.1 (%ir-block.0): ; GREEDY: bb.1 (%ir-block.0):
; GREEDY-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10 ; GREEDY-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2 ; GREEDY-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; GREEDY-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3 ; GREEDY-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; GREEDY-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4 ; GREEDY-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
; GREEDY-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; GREEDY-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; GREEDY-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; GREEDY-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; GREEDY-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7 ; GREEDY-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7
; GREEDY-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8 ; GREEDY-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8
; GREEDY-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9 ; GREEDY-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9
; GREEDY-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GREEDY-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GREEDY-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10 ; GREEDY-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10
; GREEDY-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF ; GREEDY-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF
; GREEDY-NEXT: [[COPY9:%[0-9]+]]:vgpr(s32) = COPY [[COPY8]](s32) ; GREEDY-NEXT: [[COPY9:%[0-9]+]]:vgpr(s32) = COPY [[COPY8]](s32)
; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY9]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY9]](s32), [[BUILD_VECTOR]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GREEDY-NEXT: [[COPY10:%[0-9]+]]:vgpr(p1) = COPY [[DEF]](p1) ; GREEDY-NEXT: [[COPY10:%[0-9]+]]:vgpr(p1) = COPY [[DEF]](p1)
; GREEDY-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[COPY10]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1) ; GREEDY-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[COPY10]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1)
; GREEDY-NEXT: S_ENDPGM 0 ; GREEDY-NEXT: S_ENDPGM 0
%v = call <4 x float> @llvm.amdgcn.image.load.1d.v4f32.i32(i32 15, i32 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.load.1d.v4f32.i32(i32 15, i32 %s, <8 x i32> %rsrc, i32 0, i32 0)
store <4 x float> %v, ptr addrspace(1) undef store <4 x float> %v, ptr addrspace(1) undef
ret void ret void
} }
▲ Show 20 Lines • Show All 42 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @load_1d_vgpr_vaddr__vgpr_srsrc(<8 x i32> %rsrc, i32 %s) {
; FAST-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]] ; FAST-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]]
; FAST-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]] ; FAST-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]]
; FAST-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1) ; FAST-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1)
; FAST-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; FAST-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: bb.3: ; FAST-NEXT: bb.3:
; FAST-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; FAST-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY8]](s32), [[BUILD_VECTOR1]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY8]](s32), [[BUILD_VECTOR1]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; FAST-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; FAST-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; FAST-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; FAST-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: bb.4: ; FAST-NEXT: bb.4:
; FAST-NEXT: successors: %bb.5(0x80000000) ; FAST-NEXT: successors: %bb.5(0x80000000)
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; FAST-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @load_1d_vgpr_vaddr__vgpr_srsrc(<8 x i32> %rsrc, i32 %s) {
; GREEDY-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]] ; GREEDY-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]]
; GREEDY-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]] ; GREEDY-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]]
; GREEDY-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1) ; GREEDY-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1)
; GREEDY-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; GREEDY-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: bb.3: ; GREEDY-NEXT: bb.3:
; GREEDY-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GREEDY-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY8]](s32), [[BUILD_VECTOR1]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY8]](s32), [[BUILD_VECTOR1]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GREEDY-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GREEDY-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GREEDY-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GREEDY-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: bb.4: ; GREEDY-NEXT: bb.4:
; GREEDY-NEXT: successors: %bb.5(0x80000000) ; GREEDY-NEXT: successors: %bb.5(0x80000000)
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GREEDY-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @load_1d_sgpr_vaddr__vgpr_srsrc(<8 x i32> %rsrc, i32 inreg %s) {
; FAST-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]] ; FAST-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]]
; FAST-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]] ; FAST-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]]
; FAST-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1) ; FAST-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1)
; FAST-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; FAST-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: bb.3: ; FAST-NEXT: bb.3:
; FAST-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; FAST-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY9]](s32), [[BUILD_VECTOR1]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY9]](s32), [[BUILD_VECTOR1]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; FAST-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; FAST-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; FAST-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; FAST-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: bb.4: ; FAST-NEXT: bb.4:
; FAST-NEXT: successors: %bb.5(0x80000000) ; FAST-NEXT: successors: %bb.5(0x80000000)
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; FAST-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
▲ Show 20 Lines • Show All 45 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @load_1d_sgpr_vaddr__vgpr_srsrc(<8 x i32> %rsrc, i32 inreg %s) {
; GREEDY-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]] ; GREEDY-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]]
; GREEDY-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]] ; GREEDY-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]]
; GREEDY-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1) ; GREEDY-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1)
; GREEDY-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; GREEDY-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: bb.3: ; GREEDY-NEXT: bb.3:
; GREEDY-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GREEDY-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY9]](s32), [[BUILD_VECTOR1]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, [[COPY9]](s32), [[BUILD_VECTOR1]](<8 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GREEDY-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GREEDY-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GREEDY-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GREEDY-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: bb.4: ; GREEDY-NEXT: bb.4:
; GREEDY-NEXT: successors: %bb.5(0x80000000) ; GREEDY-NEXT: successors: %bb.5(0x80000000)
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GREEDY-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
Show All 12 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-amdgcn.image.sample.1d.ll

Show All 18 Linesdefine amdgpu_ps void @sample_1d_vgpr_vaddr__sgpr_rsrc__sgpr_samp(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float %s) {
; FAST-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; FAST-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; FAST-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10 ; FAST-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10
; FAST-NEXT: [[COPY9:%[0-9]+]]:sgpr(s32) = COPY $sgpr11 ; FAST-NEXT: [[COPY9:%[0-9]+]]:sgpr(s32) = COPY $sgpr11
; FAST-NEXT: [[COPY10:%[0-9]+]]:sgpr(s32) = COPY $sgpr12 ; FAST-NEXT: [[COPY10:%[0-9]+]]:sgpr(s32) = COPY $sgpr12
; FAST-NEXT: [[COPY11:%[0-9]+]]:sgpr(s32) = COPY $sgpr13 ; FAST-NEXT: [[COPY11:%[0-9]+]]:sgpr(s32) = COPY $sgpr13
; FAST-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; FAST-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; FAST-NEXT: [[COPY12:%[0-9]+]]:vgpr(s32) = COPY $vgpr0 ; FAST-NEXT: [[COPY12:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
; FAST-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF ; FAST-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF
; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; FAST-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[DEF]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1) ; FAST-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[DEF]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1)
; FAST-NEXT: S_ENDPGM 0 ; FAST-NEXT: S_ENDPGM 0
; GREEDY-LABEL: name: sample_1d_vgpr_vaddr__sgpr_rsrc__sgpr_samp ; GREEDY-LABEL: name: sample_1d_vgpr_vaddr__sgpr_rsrc__sgpr_samp
; GREEDY: bb.1 (%ir-block.0): ; GREEDY: bb.1 (%ir-block.0):
; GREEDY-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0 ; GREEDY-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $vgpr0
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2 ; GREEDY-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; GREEDY-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3 ; GREEDY-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; GREEDY-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4 ; GREEDY-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
; GREEDY-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; GREEDY-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; GREEDY-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; GREEDY-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; GREEDY-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7 ; GREEDY-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7
; GREEDY-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8 ; GREEDY-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8
; GREEDY-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9 ; GREEDY-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9
; GREEDY-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GREEDY-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GREEDY-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10 ; GREEDY-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10
; GREEDY-NEXT: [[COPY9:%[0-9]+]]:sgpr(s32) = COPY $sgpr11 ; GREEDY-NEXT: [[COPY9:%[0-9]+]]:sgpr(s32) = COPY $sgpr11
; GREEDY-NEXT: [[COPY10:%[0-9]+]]:sgpr(s32) = COPY $sgpr12 ; GREEDY-NEXT: [[COPY10:%[0-9]+]]:sgpr(s32) = COPY $sgpr12
; GREEDY-NEXT: [[COPY11:%[0-9]+]]:sgpr(s32) = COPY $sgpr13 ; GREEDY-NEXT: [[COPY11:%[0-9]+]]:sgpr(s32) = COPY $sgpr13
; GREEDY-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GREEDY-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GREEDY-NEXT: [[COPY12:%[0-9]+]]:vgpr(s32) = COPY $vgpr0 ; GREEDY-NEXT: [[COPY12:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
; GREEDY-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF ; GREEDY-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF
; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GREEDY-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[DEF]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1) ; GREEDY-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[DEF]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1)
; GREEDY-NEXT: S_ENDPGM 0 ; GREEDY-NEXT: S_ENDPGM 0
%v = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 false, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 false, i32 0, i32 0)
store <4 x float> %v, ptr addrspace(1) undef store <4 x float> %v, ptr addrspace(1) undef
ret void ret void
} }
; Copy required for VGPR input ; Copy required for VGPR input
Show All 14 Linesdefine amdgpu_ps void @sample_1d_sgpr_vaddr__sgpr_rsrc__sgpr_samp(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, float inreg %s) {
; FAST-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10 ; FAST-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10
; FAST-NEXT: [[COPY9:%[0-9]+]]:sgpr(s32) = COPY $sgpr11 ; FAST-NEXT: [[COPY9:%[0-9]+]]:sgpr(s32) = COPY $sgpr11
; FAST-NEXT: [[COPY10:%[0-9]+]]:sgpr(s32) = COPY $sgpr12 ; FAST-NEXT: [[COPY10:%[0-9]+]]:sgpr(s32) = COPY $sgpr12
; FAST-NEXT: [[COPY11:%[0-9]+]]:sgpr(s32) = COPY $sgpr13 ; FAST-NEXT: [[COPY11:%[0-9]+]]:sgpr(s32) = COPY $sgpr13
; FAST-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; FAST-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; FAST-NEXT: [[COPY12:%[0-9]+]]:sgpr(s32) = COPY $sgpr14 ; FAST-NEXT: [[COPY12:%[0-9]+]]:sgpr(s32) = COPY $sgpr14
; FAST-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF ; FAST-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF
; FAST-NEXT: [[COPY13:%[0-9]+]]:vgpr(s32) = COPY [[COPY12]](s32) ; FAST-NEXT: [[COPY13:%[0-9]+]]:vgpr(s32) = COPY [[COPY12]](s32)
; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY13]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY13]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; FAST-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[DEF]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1) ; FAST-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[DEF]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1)
; FAST-NEXT: S_ENDPGM 0 ; FAST-NEXT: S_ENDPGM 0
; GREEDY-LABEL: name: sample_1d_sgpr_vaddr__sgpr_rsrc__sgpr_samp ; GREEDY-LABEL: name: sample_1d_sgpr_vaddr__sgpr_rsrc__sgpr_samp
; GREEDY: bb.1 (%ir-block.0): ; GREEDY: bb.1 (%ir-block.0):
; GREEDY-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $sgpr14 ; GREEDY-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9, $sgpr10, $sgpr11, $sgpr12, $sgpr13, $sgpr14
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2 ; GREEDY-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; GREEDY-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3 ; GREEDY-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; GREEDY-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4 ; GREEDY-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
; GREEDY-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; GREEDY-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; GREEDY-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; GREEDY-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; GREEDY-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7 ; GREEDY-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7
; GREEDY-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8 ; GREEDY-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8
; GREEDY-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9 ; GREEDY-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9
; GREEDY-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32) ; GREEDY-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<8 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32), [[COPY4]](s32), [[COPY5]](s32), [[COPY6]](s32), [[COPY7]](s32)
; GREEDY-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10 ; GREEDY-NEXT: [[COPY8:%[0-9]+]]:sgpr(s32) = COPY $sgpr10
; GREEDY-NEXT: [[COPY9:%[0-9]+]]:sgpr(s32) = COPY $sgpr11 ; GREEDY-NEXT: [[COPY9:%[0-9]+]]:sgpr(s32) = COPY $sgpr11
; GREEDY-NEXT: [[COPY10:%[0-9]+]]:sgpr(s32) = COPY $sgpr12 ; GREEDY-NEXT: [[COPY10:%[0-9]+]]:sgpr(s32) = COPY $sgpr12
; GREEDY-NEXT: [[COPY11:%[0-9]+]]:sgpr(s32) = COPY $sgpr13 ; GREEDY-NEXT: [[COPY11:%[0-9]+]]:sgpr(s32) = COPY $sgpr13
; GREEDY-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32) ; GREEDY-NEXT: [[BUILD_VECTOR1:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY8]](s32), [[COPY9]](s32), [[COPY10]](s32), [[COPY11]](s32)
; GREEDY-NEXT: [[COPY12:%[0-9]+]]:sgpr(s32) = COPY $sgpr14 ; GREEDY-NEXT: [[COPY12:%[0-9]+]]:sgpr(s32) = COPY $sgpr14
; GREEDY-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF ; GREEDY-NEXT: [[DEF:%[0-9]+]]:sgpr(p1) = G_IMPLICIT_DEF
; GREEDY-NEXT: [[COPY13:%[0-9]+]]:vgpr(s32) = COPY [[COPY12]](s32) ; GREEDY-NEXT: [[COPY13:%[0-9]+]]:vgpr(s32) = COPY [[COPY12]](s32)
; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY13]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY13]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GREEDY-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[DEF]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1) ; GREEDY-NEXT: G_STORE [[AMDGPU_INTRIN_IMAGE_LOAD]](<4 x s32>), [[DEF]](p1) :: (store (<4 x s32>) into `ptr addrspace(1) undef`, addrspace 1)
; GREEDY-NEXT: S_ENDPGM 0 ; GREEDY-NEXT: S_ENDPGM 0
%v = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 false, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.1d.v4f32.f32(i32 15, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 false, i32 0, i32 0)
store <4 x float> %v, ptr addrspace(1) undef store <4 x float> %v, ptr addrspace(1) undef
ret void ret void
} }
; Waterfall loop for rsrc ; Waterfall loop for rsrc
▲ Show 20 Lines • Show All 46 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @sample_1d_vgpr_vaddr__vgpr_rsrc__sgpr_samp(<8 x i32> %rsrc, <4 x i32> inreg %samp, float %s) {
; FAST-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]] ; FAST-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]]
; FAST-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]] ; FAST-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]]
; FAST-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1) ; FAST-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1)
; FAST-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; FAST-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: bb.3: ; FAST-NEXT: bb.3:
; FAST-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; FAST-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR2]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR2]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; FAST-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; FAST-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; FAST-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; FAST-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: bb.4: ; FAST-NEXT: bb.4:
; FAST-NEXT: successors: %bb.5(0x80000000) ; FAST-NEXT: successors: %bb.5(0x80000000)
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; FAST-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @sample_1d_vgpr_vaddr__vgpr_rsrc__sgpr_samp(<8 x i32> %rsrc, <4 x i32> inreg %samp, float %s) {
; GREEDY-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]] ; GREEDY-NEXT: [[ICMP3:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV15]](s64), [[UV11]]
; GREEDY-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]] ; GREEDY-NEXT: [[AND2:%[0-9]+]]:vcc(s1) = G_AND [[AND1]], [[ICMP3]]
; GREEDY-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1) ; GREEDY-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND2]](s1)
; GREEDY-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; GREEDY-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: bb.3: ; GREEDY-NEXT: bb.3:
; GREEDY-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GREEDY-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR2]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR2]](<8 x s32>), [[BUILD_VECTOR1]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GREEDY-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GREEDY-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GREEDY-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GREEDY-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: bb.4: ; GREEDY-NEXT: bb.4:
; GREEDY-NEXT: successors: %bb.5(0x80000000) ; GREEDY-NEXT: successors: %bb.5(0x80000000)
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GREEDY-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
▲ Show 20 Lines • Show All 47 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @sample_1d_vgpr_vaddr__sgpr_rsrc__vgpr_samp(<8 x i32> inreg %rsrc, <4 x i32> %samp, float %s) {
; FAST-NEXT: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV7]](s64), [[UV5]] ; FAST-NEXT: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV7]](s64), [[UV5]]
; FAST-NEXT: [[AND:%[0-9]+]]:vcc(s1) = G_AND [[ICMP]], [[ICMP1]] ; FAST-NEXT: [[AND:%[0-9]+]]:vcc(s1) = G_AND [[ICMP]], [[ICMP1]]
; FAST-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND]](s1) ; FAST-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND]](s1)
; FAST-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; FAST-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: bb.3: ; FAST-NEXT: bb.3:
; FAST-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; FAST-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR2]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR2]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; FAST-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; FAST-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; FAST-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; FAST-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: bb.4: ; FAST-NEXT: bb.4:
; FAST-NEXT: successors: %bb.5(0x80000000) ; FAST-NEXT: successors: %bb.5(0x80000000)
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; FAST-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
Show All 40 Linesdefine amdgpu_ps void @sample_1d_vgpr_vaddr__sgpr_rsrc__vgpr_samp(<8 x i32> inreg %rsrc, <4 x i32> %samp, float %s) {
; GREEDY-NEXT: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV7]](s64), [[UV5]] ; GREEDY-NEXT: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV7]](s64), [[UV5]]
; GREEDY-NEXT: [[AND:%[0-9]+]]:vcc(s1) = G_AND [[ICMP]], [[ICMP1]] ; GREEDY-NEXT: [[AND:%[0-9]+]]:vcc(s1) = G_AND [[ICMP]], [[ICMP1]]
; GREEDY-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND]](s1) ; GREEDY-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND]](s1)
; GREEDY-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; GREEDY-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: bb.3: ; GREEDY-NEXT: bb.3:
; GREEDY-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GREEDY-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR2]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR]](<8 x s32>), [[BUILD_VECTOR2]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GREEDY-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GREEDY-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GREEDY-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GREEDY-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: bb.4: ; GREEDY-NEXT: bb.4:
; GREEDY-NEXT: successors: %bb.5(0x80000000) ; GREEDY-NEXT: successors: %bb.5(0x80000000)
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GREEDY-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
▲ Show 20 Lines • Show All 67 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @sample_1d_vgpr_vaddr__vgpr_rsrc__vgpr_samp(<8 x i32> %rsrc, <4 x i32> %samp, float %s) {
; FAST-NEXT: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV23]](s64), [[UV21]] ; FAST-NEXT: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV23]](s64), [[UV21]]
; FAST-NEXT: [[AND4:%[0-9]+]]:vcc(s1) = G_AND [[AND3]], [[ICMP5]] ; FAST-NEXT: [[AND4:%[0-9]+]]:vcc(s1) = G_AND [[AND3]], [[ICMP5]]
; FAST-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND4]](s1) ; FAST-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND4]](s1)
; FAST-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; FAST-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: bb.3: ; FAST-NEXT: bb.3:
; FAST-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; FAST-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR2]](<8 x s32>), [[BUILD_VECTOR3]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; FAST-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR2]](<8 x s32>), [[BUILD_VECTOR3]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; FAST-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; FAST-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; FAST-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; FAST-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: bb.4: ; FAST-NEXT: bb.4:
; FAST-NEXT: successors: %bb.5(0x80000000) ; FAST-NEXT: successors: %bb.5(0x80000000)
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
; FAST-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; FAST-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; FAST-NEXT: {{ $}} ; FAST-NEXT: {{ $}}
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @sample_1d_vgpr_vaddr__vgpr_rsrc__vgpr_samp(<8 x i32> %rsrc, <4 x i32> %samp, float %s) {
; GREEDY-NEXT: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV23]](s64), [[UV21]] ; GREEDY-NEXT: [[ICMP5:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV23]](s64), [[UV21]]
; GREEDY-NEXT: [[AND4:%[0-9]+]]:vcc(s1) = G_AND [[AND3]], [[ICMP5]] ; GREEDY-NEXT: [[AND4:%[0-9]+]]:vcc(s1) = G_AND [[AND3]], [[ICMP5]]
; GREEDY-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND4]](s1) ; GREEDY-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND4]](s1)
; GREEDY-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; GREEDY-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: bb.3: ; GREEDY-NEXT: bb.3:
; GREEDY-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; GREEDY-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR2]](<8 x s32>), [[BUILD_VECTOR3]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) ; GREEDY-NEXT: [[AMDGPU_INTRIN_IMAGE_LOAD:%[0-9]+]]:vgpr(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.1d), 15, [[COPY12]](s32), [[BUILD_VECTOR2]](<8 x s32>), [[BUILD_VECTOR3]](<4 x s32>), 0, 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8)
; GREEDY-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; GREEDY-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; GREEDY-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; GREEDY-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: bb.4: ; GREEDY-NEXT: bb.4:
; GREEDY-NEXT: successors: %bb.5(0x80000000) ; GREEDY-NEXT: successors: %bb.5(0x80000000)
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
; GREEDY-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; GREEDY-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; GREEDY-NEXT: {{ $}} ; GREEDY-NEXT: {{ $}}
Show All 11 Lines

llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-amdgcn.raw.buffer.load.ll

Show All 11 Linesdefine amdgpu_ps float @raw_buffer_load__sgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32) ; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32)
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr(s32) = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0 ; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr(s32) = COPY [[C]](s32) ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr(s32) = COPY [[C]](s32)
; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY6]](s32), [[COPY4]], [[COPY5]], 0, 0, 0 :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY6]](s32), [[COPY4]], [[COPY5]], 0, 0, 0 :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32) ; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32)
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
; Copies for VGPR arguments ; Copies for VGPR arguments
define amdgpu_ps float @raw_buffer_load__sgpr_rsrc__sgpr_val__sgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 inreg %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load__sgpr_rsrc__sgpr_val__sgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 inreg %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: raw_buffer_load__sgpr_rsrc__sgpr_val__sgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: raw_buffer_load__sgpr_rsrc__sgpr_val__sgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32) ; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32)
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7
; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0 ; CHECK-NEXT: [[C:%[0-9]+]]:sgpr(s32) = G_CONSTANT i32 0
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr(s32) = COPY [[C]](s32) ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr(s32) = COPY [[C]](s32)
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr(s32) = COPY [[COPY4]](s32) ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr(s32) = COPY [[COPY4]](s32)
; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY6]](s32), [[COPY7]], [[COPY5]], 0, 0, 0 :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY6]](s32), [[COPY7]], [[COPY5]], 0, 0, 0 :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32) ; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32)
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
; Waterfall for rsrc ; Waterfall for rsrc
define amdgpu_ps float @raw_buffer_load__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @raw_buffer_load__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
Show All 30 Linesdefine amdgpu_ps float @raw_buffer_load__vgpr_rsrc__vgpr_val__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV7]](s64), [[UV5]] ; CHECK-NEXT: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV7]](s64), [[UV5]]
; CHECK-NEXT: [[AND:%[0-9]+]]:vcc(s1) = G_AND [[ICMP]], [[ICMP1]] ; CHECK-NEXT: [[AND:%[0-9]+]]:vcc(s1) = G_AND [[ICMP]], [[ICMP1]]
; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND]](s1) ; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND]](s1)
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR1]](<4 x s32>), [[COPY6]](s32), [[COPY4]], [[COPY5]], 0, 0, 0 :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR1]](<4 x s32>), [[COPY6]](s32), [[COPY4]], [[COPY5]], 0, 0, 0 :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 30 Linesdefine amdgpu_ps float @raw_buffer_load__sgpr_rsrc__vgpr_val__vgpr_voffset__vgpr_soffset(<4 x i32> inreg %rsrc, i32 %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32(s32) = V_READFIRSTLANE_B32 [[COPY5]](s32), implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32(s32) = V_READFIRSTLANE_B32 [[COPY5]](s32), implicit $exec
; CHECK-NEXT: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_]](s32), [[COPY5]] ; CHECK-NEXT: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_]](s32), [[COPY5]]
; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[ICMP]](s1) ; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[ICMP]](s1)
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY6]](s32), [[COPY4]], [[V_READFIRSTLANE_B32_]], 0, 0, 0 :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY6]](s32), [[COPY4]], [[V_READFIRSTLANE_B32_]], 0, 0, 0 :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
▲ Show 20 Lines • Show All 42 Lines • ▼ Show 20 Linesdefine amdgpu_ps float @raw_buffer_load__vgpr_rsrc__vgpr_val__vgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 %voffset, i32 %soffset) {
; CHECK-NEXT: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_4]](s32), [[COPY5]] ; CHECK-NEXT: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_4]](s32), [[COPY5]]
; CHECK-NEXT: [[AND1:%[0-9]+]]:vcc(s1) = G_AND [[AND]], [[ICMP2]] ; CHECK-NEXT: [[AND1:%[0-9]+]]:vcc(s1) = G_AND [[AND]], [[ICMP2]]
; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND1]](s1) ; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND1]](s1)
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR1]](<4 x s32>), [[COPY6]](s32), [[COPY4]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0 :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR1]](<4 x s32>), [[COPY6]](s32), [[COPY4]], [[V_READFIRSTLANE_B32_4]], 0, 0, 0 :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.5: ; CHECK-NEXT: bb.5:
; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32) ; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32)
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
declare float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32>, i32, i32, i32 immarg) declare float @llvm.amdgcn.raw.buffer.load.f32(<4 x i32>, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-amdgcn.struct.buffer.load.ll

Show All 10 Linesdefine amdgpu_ps float @struct_buffer_load__sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32) ; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32)
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr(s32) = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr(s32) = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr(s32) = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY4]](s32), [[COPY5]], [[COPY6]], 0, 0, -1 :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY4]](s32), [[COPY5]], [[COPY6]], 0, 0, -1 :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32) ; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32)
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
; Copies for VGPR arguments ; Copies for VGPR arguments
define amdgpu_ps float @struct_buffer_load__sgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 inreg %soffset) { define amdgpu_ps float @struct_buffer_load__sgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, i32 inreg %vindex, i32 inreg %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_load__sgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_load__sgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32) ; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32)
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8
; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr(s32) = COPY [[COPY4]](s32) ; CHECK-NEXT: [[COPY7:%[0-9]+]]:vgpr(s32) = COPY [[COPY4]](s32)
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr(s32) = COPY [[COPY5]](s32) ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr(s32) = COPY [[COPY5]](s32)
; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY7]](s32), [[COPY8]], [[COPY6]], 0, 0, -1 :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY7]](s32), [[COPY8]], [[COPY6]], 0, 0, -1 :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32) ; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32)
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
; Waterfall for rsrc ; Waterfall for rsrc
define amdgpu_ps float @struct_buffer_load__vgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @struct_buffer_load__vgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
Show All 29 Linesdefine amdgpu_ps float @struct_buffer_load__vgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV7]](s64), [[UV5]] ; CHECK-NEXT: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV7]](s64), [[UV5]]
; CHECK-NEXT: [[AND:%[0-9]+]]:vcc(s1) = G_AND [[ICMP]], [[ICMP1]] ; CHECK-NEXT: [[AND:%[0-9]+]]:vcc(s1) = G_AND [[ICMP]], [[ICMP1]]
; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND]](s1) ; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND]](s1)
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR1]](<4 x s32>), [[COPY4]](s32), [[COPY5]], [[COPY6]], 0, 0, -1 :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR1]](<4 x s32>), [[COPY4]](s32), [[COPY5]], [[COPY6]], 0, 0, -1 :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 29 Linesdefine amdgpu_ps float @struct_buffer_load__sgpr_rsrc__vgpr_val__vgpr_vindex_vgpr_voffset__vgpr_soffset(<4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32(s32) = V_READFIRSTLANE_B32 [[COPY6]](s32), implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32(s32) = V_READFIRSTLANE_B32 [[COPY6]](s32), implicit $exec
; CHECK-NEXT: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_]](s32), [[COPY6]] ; CHECK-NEXT: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_]](s32), [[COPY6]]
; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[ICMP]](s1) ; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[ICMP]](s1)
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY4]](s32), [[COPY5]], [[V_READFIRSTLANE_B32_]], 0, 0, -1 :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR]](<4 x s32>), [[COPY4]](s32), [[COPY5]], [[V_READFIRSTLANE_B32_]], 0, 0, -1 :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
▲ Show 20 Lines • Show All 41 Lines • ▼ Show 20 Linesdefine amdgpu_ps float @struct_buffer_load__vgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset) {
; CHECK-NEXT: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_4]](s32), [[COPY6]] ; CHECK-NEXT: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_4]](s32), [[COPY6]]
; CHECK-NEXT: [[AND1:%[0-9]+]]:vcc(s1) = G_AND [[AND]], [[ICMP2]] ; CHECK-NEXT: [[AND1:%[0-9]+]]:vcc(s1) = G_AND [[AND]], [[ICMP2]]
; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND1]](s1) ; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND1]](s1)
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR1]](<4 x s32>), [[COPY4]](s32), [[COPY5]], [[V_READFIRSTLANE_B32_4]], 0, 0, -1 :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[AMDGPU_BUFFER_LOAD:%[0-9]+]]:vgpr(s32) = G_AMDGPU_BUFFER_LOAD [[BUILD_VECTOR1]](<4 x s32>), [[COPY4]](s32), [[COPY5]], [[V_READFIRSTLANE_B32_4]], 0, 0, -1 :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.5: ; CHECK-NEXT: bb.5:
; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32) ; CHECK-NEXT: $vgpr0 = COPY [[AMDGPU_BUFFER_LOAD]](s32)
; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0 ; CHECK-NEXT: SI_RETURN_TO_EPILOG implicit $vgpr0
%val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %val = call float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret float %val ret float %val
} }
declare float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32>, i32, i32, i32, i32 immarg) declare float @llvm.amdgcn.struct.buffer.load.f32(<4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/GlobalISel/regbankselect-amdgcn.struct.buffer.store.ll

Show All 11 Linesdefine amdgpu_ps void @struct_buffer_store__sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, float %val, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32) ; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32)
; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr(s32) = COPY $vgpr0 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:vgpr(s32) = COPY $vgpr0
; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr(s32) = COPY $vgpr1 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:vgpr(s32) = COPY $vgpr1
; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr(s32) = COPY $vgpr2 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:vgpr(s32) = COPY $vgpr2
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; CHECK-NEXT: G_AMDGPU_BUFFER_STORE [[COPY4]](s32), [[BUILD_VECTOR]](<4 x s32>), [[COPY5]](s32), [[COPY6]], [[COPY7]], 0, 0, -1 :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: G_AMDGPU_BUFFER_STORE [[COPY4]](s32), [[BUILD_VECTOR]](<4 x s32>), [[COPY5]](s32), [[COPY6]], [[COPY7]], 0, 0, -1 :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; Copies for VGPR arguments ; Copies for VGPR arguments
define amdgpu_ps void @struct_buffer_store__sgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, float inreg %val, i32 inreg %vindex, i32 inreg %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store__sgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__sgpr_soffset(<4 x i32> inreg %rsrc, float inreg %val, i32 inreg %vindex, i32 inreg %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store__sgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_store__sgpr_rsrc__sgpr_val__sgpr_vindex__sgpr_voffset__sgpr_soffset
; CHECK: bb.1 (%ir-block.0): ; CHECK: bb.1 (%ir-block.0):
; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9 ; CHECK-NEXT: liveins: $sgpr2, $sgpr3, $sgpr4, $sgpr5, $sgpr6, $sgpr7, $sgpr8, $sgpr9
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2 ; CHECK-NEXT: [[COPY:%[0-9]+]]:sgpr(s32) = COPY $sgpr2
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3 ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sgpr(s32) = COPY $sgpr3
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4 ; CHECK-NEXT: [[COPY2:%[0-9]+]]:sgpr(s32) = COPY $sgpr4
; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5 ; CHECK-NEXT: [[COPY3:%[0-9]+]]:sgpr(s32) = COPY $sgpr5
; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32) ; CHECK-NEXT: [[BUILD_VECTOR:%[0-9]+]]:sgpr(<4 x s32>) = G_BUILD_VECTOR [[COPY]](s32), [[COPY1]](s32), [[COPY2]](s32), [[COPY3]](s32)
; CHECK-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6 ; CHECK-NEXT: [[COPY4:%[0-9]+]]:sgpr(s32) = COPY $sgpr6
; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7 ; CHECK-NEXT: [[COPY5:%[0-9]+]]:sgpr(s32) = COPY $sgpr7
; CHECK-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8 ; CHECK-NEXT: [[COPY6:%[0-9]+]]:sgpr(s32) = COPY $sgpr8
; CHECK-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9 ; CHECK-NEXT: [[COPY7:%[0-9]+]]:sgpr(s32) = COPY $sgpr9
; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr(s32) = COPY [[COPY4]](s32) ; CHECK-NEXT: [[COPY8:%[0-9]+]]:vgpr(s32) = COPY [[COPY4]](s32)
; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr(s32) = COPY [[COPY5]](s32) ; CHECK-NEXT: [[COPY9:%[0-9]+]]:vgpr(s32) = COPY [[COPY5]](s32)
; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr(s32) = COPY [[COPY6]](s32) ; CHECK-NEXT: [[COPY10:%[0-9]+]]:vgpr(s32) = COPY [[COPY6]](s32)
; CHECK-NEXT: G_AMDGPU_BUFFER_STORE [[COPY8]](s32), [[BUILD_VECTOR]](<4 x s32>), [[COPY9]](s32), [[COPY10]], [[COPY7]], 0, 0, -1 :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: G_AMDGPU_BUFFER_STORE [[COPY8]](s32), [[BUILD_VECTOR]](<4 x s32>), [[COPY9]](s32), [[COPY10]], [[COPY7]], 0, 0, -1 :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
; Waterfall for rsrc ; Waterfall for rsrc
define amdgpu_ps void @struct_buffer_store__vgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, float %val, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @struct_buffer_store__vgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, float %val, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-LABEL: name: struct_buffer_store__vgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset ; CHECK-LABEL: name: struct_buffer_store__vgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset
Show All 29 Linesdefine amdgpu_ps void @struct_buffer_store__vgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__sgpr_soffset(<4 x i32> %rsrc, float %val, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; CHECK-NEXT: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV7]](s64), [[UV5]] ; CHECK-NEXT: [[ICMP1:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[UV7]](s64), [[UV5]]
; CHECK-NEXT: [[AND:%[0-9]+]]:vcc(s1) = G_AND [[ICMP]], [[ICMP1]] ; CHECK-NEXT: [[AND:%[0-9]+]]:vcc(s1) = G_AND [[ICMP]], [[ICMP1]]
; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND]](s1) ; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND]](s1)
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: G_AMDGPU_BUFFER_STORE [[COPY4]](s32), [[BUILD_VECTOR1]](<4 x s32>), [[COPY5]](s32), [[COPY6]], [[COPY7]], 0, 0, -1 :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: G_AMDGPU_BUFFER_STORE [[COPY4]](s32), [[BUILD_VECTOR1]](<4 x s32>), [[COPY5]](s32), [[COPY6]], [[COPY7]], 0, 0, -1 :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
Show All 29 Linesdefine amdgpu_ps void @struct_buffer_store__sgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__vgpr_soffset(<4 x i32> inreg %rsrc, float %val, i32 %vindex, i32 %voffset, i32 %soffset) {
; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32(s32) = V_READFIRSTLANE_B32 [[COPY7]](s32), implicit $exec ; CHECK-NEXT: [[V_READFIRSTLANE_B32_:%[0-9]+]]:sreg_32(s32) = V_READFIRSTLANE_B32 [[COPY7]](s32), implicit $exec
; CHECK-NEXT: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_]](s32), [[COPY7]] ; CHECK-NEXT: [[ICMP:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_]](s32), [[COPY7]]
; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[ICMP]](s1) ; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[ICMP]](s1)
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: G_AMDGPU_BUFFER_STORE [[COPY4]](s32), [[BUILD_VECTOR]](<4 x s32>), [[COPY5]](s32), [[COPY6]], [[V_READFIRSTLANE_B32_]], 0, 0, -1 :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: G_AMDGPU_BUFFER_STORE [[COPY4]](s32), [[BUILD_VECTOR]](<4 x s32>), [[COPY5]](s32), [[COPY6]], [[V_READFIRSTLANE_B32_]], 0, 0, -1 :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
▲ Show 20 Lines • Show All 41 Lines • ▼ Show 20 Linesdefine amdgpu_ps void @struct_buffer_store__vgpr_rsrc__vgpr_val__vgpr_vindex__vgpr_voffset__vgpr_soffset(<4 x i32> %rsrc, float %val, i32 %vindex, i32 %voffset, i32 %soffset) {
; CHECK-NEXT: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_4]](s32), [[COPY7]] ; CHECK-NEXT: [[ICMP2:%[0-9]+]]:vcc(s1) = G_ICMP intpred(eq), [[V_READFIRSTLANE_B32_4]](s32), [[COPY7]]
; CHECK-NEXT: [[AND1:%[0-9]+]]:vcc(s1) = G_AND [[AND]], [[ICMP2]] ; CHECK-NEXT: [[AND1:%[0-9]+]]:vcc(s1) = G_AND [[AND]], [[ICMP2]]
; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND1]](s1) ; CHECK-NEXT: [[INT:%[0-9]+]]:sreg_64_xexec(s64) = G_INTRINSIC intrinsic(@llvm.amdgcn.ballot), [[AND1]](s1)
; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec ; CHECK-NEXT: [[S_AND_SAVEEXEC_B64_:%[0-9]+]]:sreg_64_xexec = S_AND_SAVEEXEC_B64 killed [[INT]](s64), implicit-def $exec, implicit-def $scc, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3: ; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000) ; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: G_AMDGPU_BUFFER_STORE [[COPY4]](s32), [[BUILD_VECTOR1]](<4 x s32>), [[COPY5]](s32), [[COPY6]], [[V_READFIRSTLANE_B32_4]], 0, 0, -1 :: (dereferenceable store (s32), align 1, addrspace 7) ; CHECK-NEXT: G_AMDGPU_BUFFER_STORE [[COPY4]](s32), [[BUILD_VECTOR1]](<4 x s32>), [[COPY5]](s32), [[COPY6]], [[V_READFIRSTLANE_B32_4]], 0, 0, -1 :: (dereferenceable store (s32), align 1, addrspace 8)
; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc ; CHECK-NEXT: $exec = S_XOR_B64_term $exec, [[S_AND_SAVEEXEC_B64_]], implicit-def $scc
; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec ; CHECK-NEXT: SI_WATERFALL_LOOP %bb.2, implicit $exec
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4: ; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.5(0x80000000) ; CHECK-NEXT: successors: %bb.5(0x80000000)
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]] ; CHECK-NEXT: $exec = S_MOV_B64_term [[S_MOV_B64_]]
; CHECK-NEXT: {{ $}} ; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.5: ; CHECK-NEXT: bb.5:
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) call void @llvm.amdgcn.struct.buffer.store.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
declare void @llvm.amdgcn.struct.buffer.store.f32(float, <4 x i32>, i32, i32, i32, i32 immarg) declare void @llvm.amdgcn.struct.buffer.store.f32(float, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/addrspacecast-captured.ll

; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -passes=amdgpu-promote-alloca < %s | FileCheck %s ; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -passes=amdgpu-promote-alloca < %s | FileCheck %s
; Nothing should be done if the addrspacecast is captured. ; Nothing should be done if the addrspacecast is captured.
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
declare void @consume_ptr2int(i32) #0 declare void @consume_ptr2int(i32) #0
; CHECK-LABEL: @addrspacecast_captured( ; CHECK-LABEL: @addrspacecast_captured(
; CHECK: %data = alloca i32, align 4, addrspace(5) ; CHECK: %data = alloca i32, align 4, addrspace(5)
; CHECK: %cast = addrspacecast ptr addrspace(5) %data to ptr ; CHECK: %cast = addrspacecast ptr addrspace(5) %data to ptr
; CHECK: %ptr2int = ptrtoint ptr %cast to i32 ; CHECK: %ptr2int = ptrtoint ptr %cast to i32
; CHECK: store i32 %ptr2int, ptr addrspace(1) %out ; CHECK: store i32 %ptr2int, ptr addrspace(1) %out
Show All 36 Lines

llvm/test/CodeGen/AMDGPU/annotate-kernel-features-hsa.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-globals ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-globals
; RUN: opt -mtriple=amdgcn-unknown-amdhsa -S -amdgpu-annotate-kernel-features < %s | FileCheck -check-prefixes=HSA,AKF_HSA %s ; RUN: opt -mtriple=amdgcn-unknown-amdhsa -S -amdgpu-annotate-kernel-features < %s | FileCheck -check-prefixes=HSA,AKF_HSA %s
; RUN: opt -mtriple=amdgcn-unknown-amdhsa -S -amdgpu-attributor < %s | FileCheck -check-prefixes=HSA,ATTRIBUTOR_HSA %s ; RUN: opt -mtriple=amdgcn-unknown-amdhsa -S -amdgpu-attributor < %s | FileCheck -check-prefixes=HSA,ATTRIBUTOR_HSA %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
declare i32 @llvm.amdgcn.workgroup.id.x() #0 declare i32 @llvm.amdgcn.workgroup.id.x() #0
declare i32 @llvm.amdgcn.workgroup.id.y() #0 declare i32 @llvm.amdgcn.workgroup.id.y() #0
declare i32 @llvm.amdgcn.workgroup.id.z() #0 declare i32 @llvm.amdgcn.workgroup.id.z() #0
declare i32 @llvm.amdgcn.workitem.id.x() #0 declare i32 @llvm.amdgcn.workitem.id.x() #0
declare i32 @llvm.amdgcn.workitem.id.y() #0 declare i32 @llvm.amdgcn.workitem.id.y() #0
declare i32 @llvm.amdgcn.workitem.id.z() #0 declare i32 @llvm.amdgcn.workitem.id.z() #0
▲ Show 20 Lines • Show All 641 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/buffer-atomic-fadd.f32-no-rtn.ll

Show All 10 Linesdefine amdgpu_ps void @buffer_atomic_fadd_f32_offset_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX908_GFX11-NEXT: {{ $}} ; GFX908_GFX11-NEXT: {{ $}}
; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3 ; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908_GFX11-NEXT: S_ENDPGM 0 ; GFX908_GFX11-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offset_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offset_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f32_offen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f32_offen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_offen_no_rtn ; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_offen_no_rtn
; GFX908_GFX11: bb.0 (%ir-block.0): ; GFX908_GFX11: bb.0 (%ir-block.0):
; GFX908_GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX908_GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX908_GFX11-NEXT: {{ $}} ; GFX908_GFX11-NEXT: {{ $}}
; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908_GFX11-NEXT: S_ENDPGM 0 ; GFX908_GFX11-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offen_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f32_idxen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f32_idxen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_idxen_no_rtn ; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_idxen_no_rtn
; GFX908_GFX11: bb.0 (%ir-block.0): ; GFX908_GFX11: bb.0 (%ir-block.0):
; GFX908_GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX908_GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX908_GFX11-NEXT: {{ $}} ; GFX908_GFX11-NEXT: {{ $}}
; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908_GFX11-NEXT: S_ENDPGM 0 ; GFX908_GFX11-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_idxen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_idxen_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f32_bothen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f32_bothen_no_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_bothen_no_rtn ; GFX908_GFX11-LABEL: name: buffer_atomic_fadd_f32_bothen_no_rtn
; GFX908_GFX11: bb.0 (%ir-block.0): ; GFX908_GFX11: bb.0 (%ir-block.0):
; GFX908_GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4 ; GFX908_GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4
; GFX908_GFX11-NEXT: {{ $}} ; GFX908_GFX11-NEXT: {{ $}}
; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX908_GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX908_GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908_GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX908_GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX908_GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX908_GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX908_GFX11-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX908_GFX11-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908_GFX11-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; GFX908_GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX908_GFX11-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX908_GFX11-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908_GFX11-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908_GFX11-NEXT: S_ENDPGM 0 ; GFX908_GFX11-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_bothen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_bothen_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F32_BOTHEN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
ret void ret void
} }
declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg) declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg)
declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg) declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/buffer-atomic-fadd.f32-rtn.ll

Show All 9 Linesdefine amdgpu_ps float @buffer_atomic_fadd_f32_offset_rtn(float %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFSET_RTN [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFSET_RTN [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_offset_rtn ; GFX11-LABEL: name: buffer_atomic_fadd_f32_offset_rtn
; GFX11: bb.0 (%ir-block.0): ; GFX11: bb.0 (%ir-block.0):
; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4 ; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX11-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX11-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3 ; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFSET_RTN [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFSET_RTN [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN]] ; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFSET_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret float %ret ret float %ret
} }
define amdgpu_ps float @buffer_atomic_fadd_f32_offen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @buffer_atomic_fadd_f32_offen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_offen_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_offen_rtn ; GFX11-LABEL: name: buffer_atomic_fadd_f32_offen_rtn
; GFX11: bb.0 (%ir-block.0): ; GFX11: bb.0 (%ir-block.0):
; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_OFFEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN]] ; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_OFFEN_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret float %ret ret float %ret
} }
define amdgpu_ps float @buffer_atomic_fadd_f32_idxen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps float @buffer_atomic_fadd_f32_idxen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_idxen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_idxen_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_IDXEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_IDXEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_idxen_rtn ; GFX11-LABEL: name: buffer_atomic_fadd_f32_idxen_rtn
; GFX11: bb.0 (%ir-block.0): ; GFX11: bb.0 (%ir-block.0):
; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX11-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_IDXEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_IDXEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN]] ; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_IDXEN_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret float %ret ret float %ret
} }
define amdgpu_ps float @buffer_atomic_fadd_f32_bothen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps float @buffer_atomic_fadd_f32_bothen_rtn(float %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_bothen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f32_bothen_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
; GFX11-LABEL: name: buffer_atomic_fadd_f32_bothen_rtn ; GFX11-LABEL: name: buffer_atomic_fadd_f32_bothen_rtn
; GFX11: bb.0 (%ir-block.0): ; GFX11: bb.0 (%ir-block.0):
; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4 ; GFX11-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4
; GFX11-NEXT: {{ $}} ; GFX11-NEXT: {{ $}}
; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX11-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX11-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX11-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX11-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX11-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX11-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX11-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX11-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX11-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX11-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; GFX11-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX11-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX11-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX11-NEXT: [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN]] ; GFX11-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_ADD_F32_BOTHEN_RTN]]
; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX11-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret float %ret ret float %ret
} }
declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg) declare float @llvm.amdgcn.raw.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32 immarg)
declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg) declare float @llvm.amdgcn.struct.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/buffer-atomic-fadd.f64.ll

Show All 11 Linesdefine amdgpu_ps void @buffer_atomic_fadd_f64_offset_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]] ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]]
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_OFFSET killed [[COPY7]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_OFFSET killed [[COPY7]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f64_offen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f64_offen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_offen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_offen_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $vgpr1, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr2, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $vgpr1, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr2, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]] ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]]
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_OFFEN killed [[COPY8]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_OFFEN killed [[COPY8]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f64_idxen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f64_idxen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_idxen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_idxen_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $vgpr1, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr2, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $vgpr1, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr2, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]] ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]]
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_IDXEN killed [[COPY8]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_IDXEN killed [[COPY8]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_f64_bothen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_f64_bothen_no_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_bothen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_bothen_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $vgpr1, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr2, $vgpr3, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $vgpr1, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr2, $vgpr3, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr3 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr3
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY7]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]] ; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]]
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_BOTHEN killed [[COPY9]], killed [[REG_SEQUENCE2]], killed [[REG_SEQUENCE]], [[COPY]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_ADD_F64_BOTHEN killed [[COPY9]], killed [[REG_SEQUENCE2]], killed [[REG_SEQUENCE]], [[COPY]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2) %ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
ret void ret void
} }
define amdgpu_ps double @buffer_atomic_fadd_f64_offset_rtn(double %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps double @buffer_atomic_fadd_f64_offset_rtn(double %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_offset_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_f64_offset_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $vgpr1, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $vgpr1, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]] ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]]
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_OFFSET_RTN [[COPY7]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_OFFSET_RTN [[COPY7]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN]].sub0 ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN]].sub0
; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN]].sub1 ; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFSET_RTN]].sub1
; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[COPY8]] ; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[COPY8]]
; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[COPY9]] ; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[COPY9]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $sgpr0, $sgpr1 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $sgpr0, $sgpr1
%ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret double %ret ret double %ret
} }
Show All 9 Linesdefine amdgpu_ps double @buffer_atomic_fadd_f64_offen_rtn(double %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]] ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]]
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_OFFEN_RTN [[COPY8]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_OFFEN_RTN [[COPY8]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN]].sub0 ; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN]].sub0
; GFX90A_GFX940-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN]].sub1 ; GFX90A_GFX940-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_OFFEN_RTN]].sub1
; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[COPY9]] ; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[COPY9]]
; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[COPY10]] ; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[COPY10]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $sgpr0, $sgpr1 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $sgpr0, $sgpr1
%ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret double %ret ret double %ret
} }
Show All 9 Linesdefine amdgpu_ps double @buffer_atomic_fadd_f64_idxen_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY6]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY7]], %subreg.sub0, [[COPY6]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]] ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]]
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_IDXEN_RTN [[COPY8]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_IDXEN_RTN [[COPY8]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN]].sub0 ; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN]].sub0
; GFX90A_GFX940-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN]].sub1 ; GFX90A_GFX940-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_IDXEN_RTN]].sub1
; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[COPY9]] ; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[COPY9]]
; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[COPY10]] ; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[COPY10]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $sgpr0, $sgpr1 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $sgpr0, $sgpr1
%ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret double %ret ret double %ret
} }
Show All 11 Linesdefine amdgpu_ps double @buffer_atomic_fadd_f64_bothen_rtn(double %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY7]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[COPY8]], %subreg.sub0, [[COPY7]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]] ; GFX90A_GFX940-NEXT: [[COPY9:%[0-9]+]]:vreg_64_align2 = COPY [[REG_SEQUENCE1]]
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN [[COPY9]], killed [[REG_SEQUENCE2]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN:%[0-9]+]]:vreg_64_align2 = BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN [[COPY9]], killed [[REG_SEQUENCE2]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s64), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN]].sub0 ; GFX90A_GFX940-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN]].sub0
; GFX90A_GFX940-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN]].sub1 ; GFX90A_GFX940-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[BUFFER_ATOMIC_ADD_F64_BOTHEN_RTN]].sub1
; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[COPY10]] ; GFX90A_GFX940-NEXT: $sgpr0 = COPY [[COPY10]]
; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[COPY11]] ; GFX90A_GFX940-NEXT: $sgpr1 = COPY [[COPY11]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $sgpr0, $sgpr1 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $sgpr0, $sgpr1
%ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0) %ret = call double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 0)
ret double %ret ret double %ret
} }
declare double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double, <4 x i32>, i32, i32, i32 immarg) declare double @llvm.amdgcn.raw.buffer.atomic.fadd.f64(double, <4 x i32>, i32, i32, i32 immarg)
declare double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double, <4 x i32>, i32, i32, i32, i32 immarg) declare double @llvm.amdgcn.struct.buffer.atomic.fadd.f64(double, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/buffer-atomic-fadd.v2f16-no-rtn.ll

Show All 9 Linesdefine amdgpu_ps void @buffer_atomic_fadd_v2f16_offset_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offset_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offset_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFSET [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 4095, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 4095, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_v2f16_offen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_v2f16_offen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_offen_no_rtn ; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_offen_no_rtn
; GFX908: bb.0 (%ir-block.0): ; GFX908: bb.0 (%ir-block.0):
; GFX908-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX908-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX908-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offen_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_OFFEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_v2f16_idxen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_v2f16_idxen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_idxen_no_rtn ; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_idxen_no_rtn
; GFX908: bb.0 (%ir-block.0): ; GFX908: bb.0 (%ir-block.0):
; GFX908-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX908-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX908-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_idxen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_idxen_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_IDXEN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret void ret void
} }
define amdgpu_ps void @buffer_atomic_fadd_v2f16_bothen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps void @buffer_atomic_fadd_v2f16_bothen_no_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_bothen_no_rtn ; GFX908-LABEL: name: buffer_atomic_fadd_v2f16_bothen_no_rtn
; GFX908: bb.0 (%ir-block.0): ; GFX908: bb.0 (%ir-block.0):
; GFX908-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4 ; GFX908-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4
; GFX908-NEXT: {{ $}} ; GFX908-NEXT: {{ $}}
; GFX908-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX908-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX908-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX908-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX908-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX908-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX908-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX908-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX908-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX908-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX908-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX908-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX908-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX908-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX908-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX908-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; GFX908-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX908-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX908-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX908-NEXT: S_ENDPGM 0 ; GFX908-NEXT: S_ENDPGM 0
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_bothen_no_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_bothen_no_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: BUFFER_ATOMIC_PK_ADD_F16_BOTHEN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 2, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: S_ENDPGM 0 ; GFX90A_GFX940-NEXT: S_ENDPGM 0
%ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2) %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
ret void ret void
} }
declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg) declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg)
declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg) declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/buffer-atomic-fadd.v2f16-rtn.ll

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -march=amdgcn -mcpu=gfx90a -verify-machineinstrs -stop-after=amdgpu-isel < %s | FileCheck -check-prefix=GFX90A_GFX940 %s ; RUN: llc -march=amdgcn -mcpu=gfx90a -verify-machineinstrs -stop-after=amdgpu-isel < %s | FileCheck -check-prefix=GFX90A_GFX940 %s
; RUN: llc -march=amdgcn -mcpu=gfx940 -verify-machineinstrs -stop-after=amdgpu-isel < %s | FileCheck -check-prefix=GFX90A_GFX940 %s ; RUN: llc -march=amdgcn -mcpu=gfx940 -verify-machineinstrs -stop-after=amdgpu-isel < %s | FileCheck -check-prefix=GFX90A_GFX940 %s
define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_offset_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) { define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_offset_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offset_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offset_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY4]], %subreg.sub0, [[COPY3]], %subreg.sub1, [[COPY2]], %subreg.sub2, [[COPY1]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN [[COPY5]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_OFFSET_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 0, i32 %soffset, i32 0)
ret <2 x half> %ret ret <2 x half> %ret
} }
define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_offen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_offen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_offen_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_OFFEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %voffset, i32 %soffset, i32 0)
ret <2 x half> %ret ret <2 x half> %ret
} }
define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_idxen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) { define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_idxen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_idxen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_idxen_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY5]], %subreg.sub0, [[COPY4]], %subreg.sub1, [[COPY3]], %subreg.sub2, [[COPY2]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN [[COPY6]], [[COPY1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 1, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_IDXEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0) %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 0, i32 %soffset, i32 0)
ret <2 x half> %ret ret <2 x half> %ret
} }
define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_bothen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) { define amdgpu_ps <2 x half> @buffer_atomic_fadd_v2f16_bothen_rtn(<2 x half> %val, <4 x i32> inreg %rsrc, i32 %vindex, i32 %voffset, i32 inreg %soffset) {
; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_bothen_rtn ; GFX90A_GFX940-LABEL: name: buffer_atomic_fadd_v2f16_bothen_rtn
; GFX90A_GFX940: bb.0 (%ir-block.0): ; GFX90A_GFX940: bb.0 (%ir-block.0):
; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4 ; GFX90A_GFX940-NEXT: liveins: $vgpr0, $sgpr0, $sgpr1, $sgpr2, $sgpr3, $vgpr1, $vgpr2, $sgpr4
; GFX90A_GFX940-NEXT: {{ $}} ; GFX90A_GFX940-NEXT: {{ $}}
; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4 ; GFX90A_GFX940-NEXT: [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr4
; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2 ; GFX90A_GFX940-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1 ; GFX90A_GFX940-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3 ; GFX90A_GFX940-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr3
; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2 ; GFX90A_GFX940-NEXT: [[COPY4:%[0-9]+]]:sgpr_32 = COPY $sgpr2
; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1 ; GFX90A_GFX940-NEXT: [[COPY5:%[0-9]+]]:sgpr_32 = COPY $sgpr1
; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GFX90A_GFX940-NEXT: [[COPY6:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GFX90A_GFX940-NEXT: [[COPY7:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[COPY6]], %subreg.sub0, [[COPY5]], %subreg.sub1, [[COPY4]], %subreg.sub2, [[COPY3]], %subreg.sub3
; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1 ; GFX90A_GFX940-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE [[COPY2]], %subreg.sub0, [[COPY1]], %subreg.sub1
; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 3, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GFX90A_GFX940-NEXT: [[BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN:%[0-9]+]]:vgpr_32 = BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN [[COPY7]], killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], [[COPY]], 0, 3, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN]] ; GFX90A_GFX940-NEXT: $vgpr0 = COPY [[BUFFER_ATOMIC_PK_ADD_F16_BOTHEN_RTN]]
; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0 ; GFX90A_GFX940-NEXT: SI_RETURN_TO_EPILOG $vgpr0
%ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2) %ret = call <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half> %val, <4 x i32> %rsrc, i32 %vindex, i32 %voffset, i32 %soffset, i32 2)
ret <2 x half> %ret ret <2 x half> %ret
} }
declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg) declare <2 x half> @llvm.amdgcn.raw.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32 immarg)
declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg) declare <2 x half> @llvm.amdgcn.struct.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/buffer-intrinsics-mmo-offsets.ll

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -march=amdgcn -mcpu=gfx908 -verify-machineinstrs -stop-after=amdgpu-isel -o - %s | FileCheck -check-prefix=GCN %s ; RUN: llc -march=amdgcn -mcpu=gfx908 -verify-machineinstrs -stop-after=amdgpu-isel -o - %s | FileCheck -check-prefix=GCN %s
define amdgpu_cs void @mmo_offsets0(ptr addrspace(6) inreg noalias align(16) dereferenceable(18446744073709551615) %arg0, i32 %arg1) { define amdgpu_cs void @mmo_offsets0(ptr addrspace(6) inreg noalias align(16) dereferenceable(18446744073709551615) %arg0, i32 %arg1) {
; GCN-LABEL: name: mmo_offsets0 ; GCN-LABEL: name: mmo_offsets0
; GCN: bb.0.bb.0: ; GCN: bb.0.bb.0:
; GCN-NEXT: liveins: $sgpr0, $vgpr0 ; GCN-NEXT: liveins: $sgpr0, $vgpr0
; GCN-NEXT: {{ $}} ; GCN-NEXT: {{ $}}
; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0 ; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GCN-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr0 ; GCN-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr0
; GCN-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 ; GCN-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_64 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[S_MOV_B32_]], %subreg.sub1 ; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_64 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[S_MOV_B32_]], %subreg.sub1
; GCN-NEXT: [[S_LOAD_DWORDX4_IMM:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM killed [[REG_SEQUENCE]], 0, 0 :: (dereferenceable invariant load (s128) from %ir.arg0, addrspace 6) ; GCN-NEXT: [[S_LOAD_DWORDX4_IMM:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM killed [[REG_SEQUENCE]], 0, 0 :: (dereferenceable invariant load (s128) from %ir.arg0, addrspace 6)
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFSET:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 16, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFSET:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 16, align 1, addrspace 8)
; GCN-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1, implicit $exec ; GCN-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1, implicit $exec
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 16, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 32, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 32, align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFEN_exact killed [[BUFFER_LOAD_DWORDX4_OFFEN]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFEN_exact killed [[BUFFER_LOAD_DWORDX4_OFFEN]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 32, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFSET:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 48, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFSET:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 48, align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 48, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 64, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 64, align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFEN]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 64, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 80, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 80, align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY]], %subreg.sub1 ; GCN-NEXT: [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY]], %subreg.sub0, [[COPY]], %subreg.sub1
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 96, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 96, align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 96, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1065353216, implicit $exec ; GCN-NEXT: [[V_MOV_B32_e32_1:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 1065353216, implicit $exec
; GCN-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[V_MOV_B32_e32_1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 112, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_F32_OFFSET [[V_MOV_B32_e32_1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 112, align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[V_MOV_B32_e32_1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_F32_OFFEN [[V_MOV_B32_e32_1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[V_MOV_B32_e32_1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[V_MOV_B32_e32_1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[V_MOV_B32_e32_1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_F32_IDXEN [[V_MOV_B32_e32_1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 112, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFSET1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 128, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 128, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFSET1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 128, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 128, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 64 ; GCN-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 64
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFSET2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_1]], 64, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 128, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFSET2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_1]], 64, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 128, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_2:%[0-9]+]]:sreg_32 = S_MOV_B32 128 ; GCN-NEXT: [[S_MOV_B32_2:%[0-9]+]]:sreg_32 = S_MOV_B32 128
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFSET3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 128, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFSET3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 128, align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFSET4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[COPY2]], 128, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFSET4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFSET [[S_LOAD_DWORDX4_IMM]], [[COPY2]], 128, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFSET1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 144, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 144, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFSET1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 144, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 144, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_3:%[0-9]+]]:sreg_32 = S_MOV_B32 72 ; GCN-NEXT: [[S_MOV_B32_3:%[0-9]+]]:sreg_32 = S_MOV_B32 72
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFSET2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_3]], 72, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 144, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFSET2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_3]], 72, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 144, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_4:%[0-9]+]]:sreg_32 = S_MOV_B32 144 ; GCN-NEXT: [[S_MOV_B32_4:%[0-9]+]]:sreg_32 = S_MOV_B32 144
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFSET3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 144, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFSET3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 144, align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFEN1:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY3:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFSET4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[COPY3]], 144, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_OFFSET4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_OFFSET [[S_LOAD_DWORDX4_IMM]], [[COPY3]], 144, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 160, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 160, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 160, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 160, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_5:%[0-9]+]]:sreg_32 = S_MOV_B32 80 ; GCN-NEXT: [[S_MOV_B32_5:%[0-9]+]]:sreg_32 = S_MOV_B32 80
; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_5]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 160, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_5]], 80, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 160, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_6:%[0-9]+]]:sreg_32 = S_MOV_B32 160 ; GCN-NEXT: [[S_MOV_B32_6:%[0-9]+]]:sreg_32 = S_MOV_B32 160
; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 160, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 160, align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_6]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY4:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[COPY4]], 160, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_OFFSET [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[COPY4]], 160, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 176, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 176, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 176, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 176, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_7:%[0-9]+]]:sreg_32 = S_MOV_B32 88 ; GCN-NEXT: [[S_MOV_B32_7:%[0-9]+]]:sreg_32 = S_MOV_B32 88
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_7]], 88, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 176, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_7]], 88, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 176, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_8:%[0-9]+]]:sreg_32 = S_MOV_B32 176 ; GCN-NEXT: [[S_MOV_B32_8:%[0-9]+]]:sreg_32 = S_MOV_B32 176
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_8]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 176, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_8]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 176, align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_8]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_8]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY5:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[COPY5]], 176, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_OFFSET [[REG_SEQUENCE1]], [[S_LOAD_DWORDX4_IMM]], [[COPY5]], 176, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 192, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 192, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 192, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 192, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_9:%[0-9]+]]:sreg_32 = S_MOV_B32 96 ; GCN-NEXT: [[S_MOV_B32_9:%[0-9]+]]:sreg_32 = S_MOV_B32 96
; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET2]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_9]], 96, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 192, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET2]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_9]], 96, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 192, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_10:%[0-9]+]]:sreg_32 = S_MOV_B32 192 ; GCN-NEXT: [[S_MOV_B32_10:%[0-9]+]]:sreg_32 = S_MOV_B32 192
; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET3]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_10]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 192, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET3]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_10]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 192, align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFEN_exact killed [[BUFFER_LOAD_DWORDX4_OFFEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_10]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFEN_exact killed [[BUFFER_LOAD_DWORDX4_OFFEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_10]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY6:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET4]], [[S_LOAD_DWORDX4_IMM]], [[COPY6]], 192, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_OFFSET_exact killed [[BUFFER_LOAD_DWORDX4_OFFSET4]], [[S_LOAD_DWORDX4_IMM]], [[COPY6]], 192, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 208, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 208, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET1]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 208, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 208, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_11:%[0-9]+]]:sreg_32 = S_MOV_B32 104 ; GCN-NEXT: [[S_MOV_B32_11:%[0-9]+]]:sreg_32 = S_MOV_B32 104
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET2]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_11]], 104, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 208, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET2]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_11]], 104, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 208, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_12:%[0-9]+]]:sreg_32 = S_MOV_B32 208 ; GCN-NEXT: [[S_MOV_B32_12:%[0-9]+]]:sreg_32 = S_MOV_B32 208
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET3]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_12]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 208, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET3]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_12]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 208, align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_12]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFEN1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_12]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY7:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET4]], [[S_LOAD_DWORDX4_IMM]], [[COPY7]], 208, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_OFFSET_exact killed [[BUFFER_LOAD_FORMAT_XYZW_OFFSET4]], [[S_LOAD_DWORDX4_IMM]], [[COPY7]], 208, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY8]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 224, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY8]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 224, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_13:%[0-9]+]]:sreg_32 = S_MOV_B32 112 ; GCN-NEXT: [[S_MOV_B32_13:%[0-9]+]]:sreg_32 = S_MOV_B32 112
; GCN-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY9]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_13]], 112, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 224, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY9]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_13]], 112, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 224, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_14:%[0-9]+]]:sreg_32 = S_MOV_B32 224 ; GCN-NEXT: [[S_MOV_B32_14:%[0-9]+]]:sreg_32 = S_MOV_B32 224
; GCN-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY10:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY10]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_14]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 224, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY10]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_14]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 224, align 1, addrspace 8)
; GCN-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[COPY]], %subreg.sub1 ; GCN-NEXT: [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[COPY]], %subreg.sub1
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_BOTHEN [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_14]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_BOTHEN [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_14]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY11:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[COPY12:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY12:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN5:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY11]], [[S_LOAD_DWORDX4_IMM]], [[COPY12]], 224, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN5:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY11]], [[S_LOAD_DWORDX4_IMM]], [[COPY12]], 224, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN6:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN6:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN7:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_IDXEN7:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 224, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[COPY13:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY13:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY13]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 240, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN2:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY13]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 240, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_15:%[0-9]+]]:sreg_32 = S_MOV_B32 120 ; GCN-NEXT: [[S_MOV_B32_15:%[0-9]+]]:sreg_32 = S_MOV_B32 120
; GCN-NEXT: [[COPY14:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY14:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY14]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_15]], 120, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 240, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN3:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY14]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_15]], 120, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 240, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_16:%[0-9]+]]:sreg_32 = S_MOV_B32 240 ; GCN-NEXT: [[S_MOV_B32_16:%[0-9]+]]:sreg_32 = S_MOV_B32 240
; GCN-NEXT: [[COPY15:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY15:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY15]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_16]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 240, align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN4:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY15]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_16]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128) from unknown-address + 240, align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_16]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_BOTHEN [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_16]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[COPY16:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY16:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[COPY17:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY17:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN5:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY16]], [[S_LOAD_DWORDX4_IMM]], [[COPY17]], 240, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN5:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY16]], [[S_LOAD_DWORDX4_IMM]], [[COPY17]], 240, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN6:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN6:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN7:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_FORMAT_XYZW_IDXEN7:%[0-9]+]]:vreg_128 = BUFFER_LOAD_FORMAT_XYZW_IDXEN [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 240, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[COPY18:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY18:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY18]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 256, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY18]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 256, align 1, addrspace 8)
; GCN-NEXT: [[COPY19:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY19:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY19]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 128, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 256, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY19]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_2]], 128, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 256, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_17:%[0-9]+]]:sreg_32 = S_MOV_B32 256 ; GCN-NEXT: [[S_MOV_B32_17:%[0-9]+]]:sreg_32 = S_MOV_B32 256
; GCN-NEXT: [[COPY20:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY20:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY20]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_17]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 256, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY20]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_17]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 256, align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_ADD_BOTHEN [[COPY]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_17]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_BOTHEN [[COPY]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_17]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: [[COPY21:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY21:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[COPY22:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY22:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY21]], [[S_LOAD_DWORDX4_IMM]], [[COPY22]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY21]], [[S_LOAD_DWORDX4_IMM]], [[COPY22]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_ADD_IDXEN [[COPY]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 256, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[COPY23:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY23:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY23]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 272, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY23]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 272, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_18:%[0-9]+]]:sreg_32 = S_MOV_B32 136 ; GCN-NEXT: [[S_MOV_B32_18:%[0-9]+]]:sreg_32 = S_MOV_B32 136
; GCN-NEXT: [[COPY24:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY24:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY24]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_18]], 136, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 272, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY24]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_18]], 136, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 272, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_19:%[0-9]+]]:sreg_32 = S_MOV_B32 272 ; GCN-NEXT: [[S_MOV_B32_19:%[0-9]+]]:sreg_32 = S_MOV_B32 272
; GCN-NEXT: [[COPY25:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY25:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY25]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_19]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 272, align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY25]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_19]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32) on unknown-address + 272, align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_19]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_BOTHEN [[REG_SEQUENCE1]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_19]], 0, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: [[COPY26:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY26:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[COPY27:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY27:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY26]], [[S_LOAD_DWORDX4_IMM]], [[COPY27]], 272, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY26]], [[S_LOAD_DWORDX4_IMM]], [[COPY27]], 272, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 7) ; GCN-NEXT: BUFFER_ATOMIC_CMPSWAP_IDXEN [[REG_SEQUENCE1]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 272, 0, implicit $exec :: (volatile dereferenceable load store (s32), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[COPY28:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY28:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN2]], [[COPY28]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 288, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN2]], [[COPY28]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 288, align 1, addrspace 8)
; GCN-NEXT: [[COPY29:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY29:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN3]], [[COPY29]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 144, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 288, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN3]], [[COPY29]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_4]], 144, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 288, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_20:%[0-9]+]]:sreg_32 = S_MOV_B32 288 ; GCN-NEXT: [[S_MOV_B32_20:%[0-9]+]]:sreg_32 = S_MOV_B32 288
; GCN-NEXT: [[COPY30:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY30:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN4]], [[COPY30]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_20]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 288, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN4]], [[COPY30]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_20]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 288, align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_DWORDX4_BOTHEN_exact killed [[BUFFER_LOAD_DWORDX4_BOTHEN]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_20]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_BOTHEN_exact killed [[BUFFER_LOAD_DWORDX4_BOTHEN]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_20]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: [[COPY31:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY31:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[COPY32:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY32:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN5]], [[COPY31]], [[S_LOAD_DWORDX4_IMM]], [[COPY32]], 288, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN5]], [[COPY31]], [[S_LOAD_DWORDX4_IMM]], [[COPY32]], 288, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN6]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN6]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN7]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX4_IDXEN_exact killed [[BUFFER_LOAD_DWORDX4_IDXEN7]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 288, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */ ; GCN-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; GCN-NEXT: [[COPY33:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY33:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN2]], [[COPY33]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 304, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN2]], [[COPY33]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 304, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_21:%[0-9]+]]:sreg_32 = S_MOV_B32 152 ; GCN-NEXT: [[S_MOV_B32_21:%[0-9]+]]:sreg_32 = S_MOV_B32 152
; GCN-NEXT: [[COPY34:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY34:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN3]], [[COPY34]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_21]], 152, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 304, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN3]], [[COPY34]], [[S_LOAD_DWORDX4_IMM]], killed [[S_MOV_B32_21]], 152, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 304, align 1, addrspace 8)
; GCN-NEXT: [[S_MOV_B32_22:%[0-9]+]]:sreg_32 = S_MOV_B32 304 ; GCN-NEXT: [[S_MOV_B32_22:%[0-9]+]]:sreg_32 = S_MOV_B32 304
; GCN-NEXT: [[COPY35:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY35:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN4]], [[COPY35]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_22]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 304, align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN4]], [[COPY35]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_22]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128) into unknown-address + 304, align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_BOTHEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_22]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_BOTHEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_BOTHEN]], [[REG_SEQUENCE2]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_22]], 0, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: [[COPY36:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]] ; GCN-NEXT: [[COPY36:%[0-9]+]]:vgpr_32 = COPY [[S_MOV_B32_]]
; GCN-NEXT: [[COPY37:%[0-9]+]]:sreg_32 = COPY [[COPY]] ; GCN-NEXT: [[COPY37:%[0-9]+]]:sreg_32 = COPY [[COPY]]
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN5]], [[COPY36]], [[S_LOAD_DWORDX4_IMM]], [[COPY37]], 304, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN5]], [[COPY36]], [[S_LOAD_DWORDX4_IMM]], [[COPY37]], 304, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN6]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN6]], [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN7]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_FORMAT_XYZW_IDXEN_exact killed [[BUFFER_LOAD_FORMAT_XYZW_IDXEN7]], [[COPY]], [[S_LOAD_DWORDX4_IMM]], [[S_MOV_B32_]], 304, 0, 0, implicit $exec :: (dereferenceable store (s128), align 1, addrspace 8)
; GCN-NEXT: S_ENDPGM 0 ; GCN-NEXT: S_ENDPGM 0
bb.0: bb.0:
%tmp0 = load <4 x i32>, ptr addrspace(6) %arg0, align 16, !invariant.load !0 %tmp0 = load <4 x i32>, ptr addrspace(6) %arg0, align 16, !invariant.load !0
%buffer0 = call nsz <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %tmp0, i32 0, i32 16, i1 false, i1 false) #0 %buffer0 = call nsz <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %tmp0, i32 0, i32 16, i1 false, i1 false) #0
%buffer1 = call nsz <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %tmp0, i32 0, i32 %arg1, i1 false, i1 false) #0 %buffer1 = call nsz <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %tmp0, i32 0, i32 %arg1, i1 false, i1 false) #0
%buffer2 = call nsz <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %tmp0, i32 1, i32 16, i1 false, i1 false) #0 %buffer2 = call nsz <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %tmp0, i32 1, i32 16, i1 false, i1 false) #0
%buffer3 = call nsz <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %tmp0, i32 %arg1, i32 16, i1 false, i1 false) #0 %buffer3 = call nsz <4 x float> @llvm.amdgcn.buffer.load.v4f32(<4 x i32> %tmp0, i32 %arg1, i32 16, i1 false, i1 false) #0
▲ Show 20 Lines • Show All 182 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/cgp-addressing-modes.ll

; RUN: opt -S -codegenprepare -mtriple=amdgcn-unknown-unknown -mcpu=tahiti < %s | FileCheck -check-prefix=OPT -check-prefix=OPT-SI -check-prefix=OPT-SICIVI %s ; RUN: opt -S -codegenprepare -mtriple=amdgcn-unknown-unknown -mcpu=tahiti < %s | FileCheck -check-prefix=OPT -check-prefix=OPT-SI -check-prefix=OPT-SICIVI %s
; RUN: opt -S -codegenprepare -mtriple=amdgcn-unknown-unknown -mcpu=bonaire < %s | FileCheck -check-prefix=OPT -check-prefix=OPT-CI -check-prefix=OPT-SICIVI %s ; RUN: opt -S -codegenprepare -mtriple=amdgcn-unknown-unknown -mcpu=bonaire < %s | FileCheck -check-prefix=OPT -check-prefix=OPT-CI -check-prefix=OPT-SICIVI %s
; RUN: opt -S -codegenprepare -mtriple=amdgcn-unknown-unknown -mcpu=tonga -mattr=-flat-for-global < %s | FileCheck -check-prefix=OPT -check-prefix=OPT-VI -check-prefix=OPT-SICIVI %s ; RUN: opt -S -codegenprepare -mtriple=amdgcn-unknown-unknown -mcpu=tonga -mattr=-flat-for-global < %s | FileCheck -check-prefix=OPT -check-prefix=OPT-VI -check-prefix=OPT-SICIVI %s
; RUN: opt -S -codegenprepare -mtriple=amdgcn-unknown-unknown -mcpu=gfx900 < %s | FileCheck -check-prefix=OPT -check-prefix=OPT-GFX9 %s ; RUN: opt -S -codegenprepare -mtriple=amdgcn-unknown-unknown -mcpu=gfx900 < %s | FileCheck -check-prefix=OPT -check-prefix=OPT-GFX9 %s
; RUN: llc -march=amdgcn -mcpu=tahiti -mattr=-promote-alloca -amdgpu-scalarize-global-loads=false -amdgpu-sroa=0 < %s | FileCheck -check-prefix=GCN -check-prefix=SI -check-prefix=SICIVI %s ; RUN: llc -march=amdgcn -mcpu=tahiti -mattr=-promote-alloca -amdgpu-scalarize-global-loads=false -amdgpu-sroa=0 < %s | FileCheck -check-prefix=GCN -check-prefix=SI -check-prefix=SICIVI %s
; RUN: llc -march=amdgcn -mcpu=bonaire -mattr=-promote-alloca -amdgpu-scalarize-global-loads=false -amdgpu-sroa=0 < %s | FileCheck -check-prefix=GCN -check-prefix=CI -check-prefix=SICIVI %s ; RUN: llc -march=amdgcn -mcpu=bonaire -mattr=-promote-alloca -amdgpu-scalarize-global-loads=false -amdgpu-sroa=0 < %s | FileCheck -check-prefix=GCN -check-prefix=CI -check-prefix=SICIVI %s
; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -amdgpu-scalarize-global-loads=false -mattr=-promote-alloca -amdgpu-sroa=0 < %s | FileCheck -check-prefix=GCN -check-prefix=VI -check-prefix=SICIVI %s ; RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -amdgpu-scalarize-global-loads=false -mattr=-promote-alloca -amdgpu-sroa=0 < %s | FileCheck -check-prefix=GCN -check-prefix=VI -check-prefix=SICIVI %s
; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=-promote-alloca -amdgpu-scalarize-global-loads=false -amdgpu-sroa=0 < %s | FileCheck -check-prefix=GCN -check-prefix=GFX9 %s ; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=-promote-alloca -amdgpu-scalarize-global-loads=false -amdgpu-sroa=0 < %s | FileCheck -check-prefix=GCN -check-prefix=GFX9 %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; OPT-LABEL: @test_sink_global_small_offset_i32( ; OPT-LABEL: @test_sink_global_small_offset_i32(
; OPT-CI-NOT: getelementptr i32, ptr addrspace(1) %in ; OPT-CI-NOT: getelementptr i32, ptr addrspace(1) %in
; OPT-VI: getelementptr i32, ptr addrspace(1) %in ; OPT-VI: getelementptr i32, ptr addrspace(1) %in
; OPT: br i1 ; OPT: br i1
; OPT-CI: getelementptr i8, ; OPT-CI: getelementptr i8,
; GCN-LABEL: {{^}}test_sink_global_small_offset_i32: ; GCN-LABEL: {{^}}test_sink_global_small_offset_i32:
▲ Show 20 Lines • Show All 783 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/extract_subvector_vec4_vec3.ll

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 < %s -stop-after=amdgpu-isel | FileCheck -check-prefix=GCN %s ; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 < %s -stop-after=amdgpu-isel | FileCheck -check-prefix=GCN %s
; We want to see a BUFFER_LOAD, some register shuffling, and a BUFFER_STORE. ; We want to see a BUFFER_LOAD, some register shuffling, and a BUFFER_STORE.
; Specifically, we do not want to see a BUFFER_STORE that says "store into ; Specifically, we do not want to see a BUFFER_STORE that says "store into
; stack" in the middle. ; stack" in the middle.
define amdgpu_hs void @main(ptr addrspace(6) inreg %arg) { define amdgpu_hs void @main(ptr addrspace(6) inreg %arg) {
; GCN-LABEL: name: main ; GCN-LABEL: name: main
; GCN: bb.0.main_body: ; GCN: bb.0.main_body:
; GCN-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0 ; GCN-NEXT: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
; GCN-NEXT: [[DEF:%[0-9]+]]:sreg_32 = IMPLICIT_DEF ; GCN-NEXT: [[DEF:%[0-9]+]]:sreg_32 = IMPLICIT_DEF
; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY [[DEF]] ; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY [[DEF]]
; GCN-NEXT: [[DEF1:%[0-9]+]]:sgpr_128 = IMPLICIT_DEF ; GCN-NEXT: [[DEF1:%[0-9]+]]:sgpr_128 = IMPLICIT_DEF
; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY]], [[DEF1]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 7) ; GCN-NEXT: [[BUFFER_LOAD_DWORDX4_OFFEN:%[0-9]+]]:vreg_128 = BUFFER_LOAD_DWORDX4_OFFEN [[COPY]], [[DEF1]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 1, addrspace 8)
; GCN-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub2 ; GCN-NEXT: [[COPY1:%[0-9]+]]:sgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub2
; GCN-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub1 ; GCN-NEXT: [[COPY2:%[0-9]+]]:sgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub1
; GCN-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub0 ; GCN-NEXT: [[COPY3:%[0-9]+]]:sgpr_32 = COPY [[BUFFER_LOAD_DWORDX4_OFFEN]].sub0
; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_96 = REG_SEQUENCE killed [[COPY3]], %subreg.sub0, killed [[COPY2]], %subreg.sub1, killed [[COPY1]], %subreg.sub2 ; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:sgpr_96 = REG_SEQUENCE killed [[COPY3]], %subreg.sub0, killed [[COPY2]], %subreg.sub1, killed [[COPY1]], %subreg.sub2
; GCN-NEXT: [[COPY4:%[0-9]+]]:vreg_96 = COPY [[REG_SEQUENCE]] ; GCN-NEXT: [[COPY4:%[0-9]+]]:vreg_96 = COPY [[REG_SEQUENCE]]
; GCN-NEXT: [[DEF2:%[0-9]+]]:sreg_32 = IMPLICIT_DEF ; GCN-NEXT: [[DEF2:%[0-9]+]]:sreg_32 = IMPLICIT_DEF
; GCN-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY [[DEF2]] ; GCN-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY [[DEF2]]
; GCN-NEXT: [[DEF3:%[0-9]+]]:sgpr_128 = IMPLICIT_DEF ; GCN-NEXT: [[DEF3:%[0-9]+]]:sgpr_128 = IMPLICIT_DEF
; GCN-NEXT: BUFFER_STORE_DWORDX3_OFFEN_exact killed [[COPY4]], [[COPY5]], [[DEF3]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (s96), align 1, addrspace 7) ; GCN-NEXT: BUFFER_STORE_DWORDX3_OFFEN_exact killed [[COPY4]], [[COPY5]], [[DEF3]], [[S_MOV_B32_]], 0, 0, 0, implicit $exec :: (dereferenceable store (s96), align 1, addrspace 8)
; GCN-NEXT: S_ENDPGM 0 ; GCN-NEXT: S_ENDPGM 0
main_body: main_body:
%tmp25 = call <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32> undef, i32 undef, i32 0, i32 0) %tmp25 = call <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32> undef, i32 undef, i32 0, i32 0)
%tmp27 = bitcast <4 x float> %tmp25 to <16 x i8> %tmp27 = bitcast <4 x float> %tmp25 to <16 x i8>
%tmp28 = shufflevector <16 x i8> %tmp27, <16 x i8> undef, <12 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11> %tmp28 = shufflevector <16 x i8> %tmp27, <16 x i8> undef, <12 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
%tmp29 = bitcast <12 x i8> %tmp28 to <3 x i32> %tmp29 = bitcast <12 x i8> %tmp28 to <3 x i32>
call void @llvm.amdgcn.raw.buffer.store.v3i32(<3 x i32> %tmp29, <4 x i32> undef, i32 undef, i32 0, i32 0) #3 call void @llvm.amdgcn.raw.buffer.store.v3i32(<3 x i32> %tmp29, <4 x i32> undef, i32 undef, i32 0, i32 0) #3
ret void ret void
} }
declare void @llvm.amdgcn.raw.buffer.store.v3i32(<3 x i32>, <4 x i32>, i32, i32, i32 immarg) declare void @llvm.amdgcn.raw.buffer.store.v3i32(<3 x i32>, <4 x i32>, i32, i32, i32 immarg)
declare <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32>, i32, i32, i32 immarg) declare <4 x float> @llvm.amdgcn.raw.buffer.load.v4f32(<4 x i32>, i32, i32, i32 immarg)

llvm/test/CodeGen/AMDGPU/force-alwaysinline-lds-global-address.ll

; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=amdgpu-always-inline -amdgpu-enable-lower-module-lds=false %s | FileCheck --check-prefix=ALL %s ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=amdgpu-always-inline -amdgpu-enable-lower-module-lds=false %s | FileCheck --check-prefix=ALL %s
; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=amdgpu-always-inline -amdgpu-enable-lower-module-lds=false %s | FileCheck --check-prefix=ALL %s ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=amdgpu-always-inline -amdgpu-enable-lower-module-lds=false %s | FileCheck --check-prefix=ALL %s
; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -amdgpu-stress-function-calls -passes=amdgpu-always-inline -amdgpu-enable-lower-module-lds=false %s | FileCheck --check-prefix=ALL %s ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -amdgpu-stress-function-calls -passes=amdgpu-always-inline -amdgpu-enable-lower-module-lds=false %s | FileCheck --check-prefix=ALL %s
; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -amdgpu-stress-function-calls -passes=amdgpu-always-inline -amdgpu-enable-lower-module-lds=false %s | FileCheck --check-prefix=ALL %s ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -amdgpu-stress-function-calls -passes=amdgpu-always-inline -amdgpu-enable-lower-module-lds=false %s | FileCheck --check-prefix=ALL %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
@lds0 = addrspace(3) global i32 undef, align 4 @lds0 = addrspace(3) global i32 undef, align 4
@lds1 = addrspace(3) global [512 x i32] undef, align 4 @lds1 = addrspace(3) global [512 x i32] undef, align 4
@nested.lds.address = addrspace(1) global ptr addrspace(3) @lds0, align 4 @nested.lds.address = addrspace(1) global ptr addrspace(3) @lds0, align 4
@gds0 = addrspace(2) global i32 undef, align 4 @gds0 = addrspace(2) global i32 undef, align 4
@alias.lds0 = alias i32, ptr addrspace(3) @lds0 @alias.lds0 = alias i32, ptr addrspace(3) @lds0
@lds.cycle = addrspace(3) global i32 ptrtoint (ptr addrspace(3) @lds.cycle to i32), align 4 @lds.cycle = addrspace(3) global i32 ptrtoint (ptr addrspace(3) @lds.cycle to i32), align 4
▲ Show 20 Lines • Show All 82 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/loop-idiom.ll

; RUN: opt -passes=loop-idiom -S < %s -march=r600 -mcpu=redwood | FileCheck --check-prefix=R600 --check-prefix=FUNC %s ; RUN: opt -passes=loop-idiom -S < %s -march=r600 -mcpu=redwood | FileCheck --check-prefix=R600 --check-prefix=FUNC %s
; RUN: opt -passes=loop-idiom -S < %s -march=amdgcn -mcpu=tahiti -verify-machineinstrs| FileCheck --check-prefix=SI --check-prefix=FUNC %s ; RUN: opt -passes=loop-idiom -S < %s -march=amdgcn -mcpu=tahiti -verify-machineinstrs| FileCheck --check-prefix=SI --check-prefix=FUNC %s
; RUN: opt -passes=loop-idiom -S < %s -march=amdgcn -mcpu=tonga -verify-machineinstrs| FileCheck --check-prefix=SI --check-prefix=FUNC %s ; RUN: opt -passes=loop-idiom -S < %s -march=amdgcn -mcpu=tonga -verify-machineinstrs| FileCheck --check-prefix=SI --check-prefix=FUNC %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; Make sure loop-idiom doesn't create memcpy or memset. There are no library ; Make sure loop-idiom doesn't create memcpy or memset. There are no library
; implementations of these for R600. ; implementations of these for R600.
; FUNC: @no_memcpy ; FUNC: @no_memcpy
; R600-NOT: {{^}}llvm.memcpy ; R600-NOT: {{^}}llvm.memcpy
; SI-NOT: {{^}}llvm.memcpy ; SI-NOT: {{^}}llvm.memcpy
define amdgpu_kernel void @no_memcpy(ptr addrspace(3) %in, i32 %size) { define amdgpu_kernel void @no_memcpy(ptr addrspace(3) %in, i32 %size) {
Show All 39 Lines

llvm/test/CodeGen/AMDGPU/mdt-preserving-crash.ll

; RUN: llc < %s ; RUN: llc < %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
@_RSENC_gDcd_______________________________ = external protected addrspace(1) externally_initialized global [4096 x i8], align 16 @_RSENC_gDcd_______________________________ = external protected addrspace(1) externally_initialized global [4096 x i8], align 16
define protected amdgpu_kernel void @_RSENC_PRInit__________________________________() local_unnamed_addr #0 { define protected amdgpu_kernel void @_RSENC_PRInit__________________________________() local_unnamed_addr #0 {
entry: entry:
%runtimeVersionCopy = alloca [128 x i8], align 16, addrspace(5) %runtimeVersionCopy = alloca [128 x i8], align 16, addrspace(5)
%licenseVersionCopy = alloca [128 x i8], align 16, addrspace(5) %licenseVersionCopy = alloca [128 x i8], align 16, addrspace(5)
▲ Show 20 Lines • Show All 144 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/noop-shader-O0.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -O0 -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s ; RUN: llc -O0 -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
; Ensure NOOP shaders compile at OptNone. ; Ensure NOOP shaders compile at OptNone.
; Confirm registers reserved in SIMachineFunctionInfo are those expected during ; Confirm registers reserved in SIMachineFunctionInfo are those expected during
; lowering, even when e.g. spilling is required due to being at OptNone. ; lowering, even when e.g. spilling is required due to being at OptNone.
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
target triple = "amdgcn-amd-amdpal" target triple = "amdgcn-amd-amdpal"
define amdgpu_vs void @noop_vs() { define amdgpu_vs void @noop_vs() {
; GCN-LABEL: noop_vs: ; GCN-LABEL: noop_vs:
; GCN: ; %bb.0: ; %entry ; GCN: ; %bb.0: ; %entry
; GCN-NEXT: s_endpgm ; GCN-NEXT: s_endpgm
entry: entry:
ret void ret void
▲ Show 20 Lines • Show All 49 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/nullptr-long-address-spaces.ll

  • This file was added.
; XFAIL: *
; RUN: llc < %s -march=amdgcn -mtriple=amdgcn-- -verify-machineinstrs | FileCheck -check-prefixes=CHECK,GCN %s
; RUN: llc < %s -march=r600 -mtriple=r600-- -verify-machineinstrs | FileCheck -check-prefixes=CHECK,R600 %s
; This is a temporary xfail, as the assembly printer is broken when dealing with
; lowerConstant() trying to return a value of size greater than 8 bytes.
; CHECK-LABEL: nullptr7:
; The exact form of the GCN output depends on how the printer gets fixed.
; GCN-NEXT: .zeroes 5
; R600-NEXT: .long 0
@nullptr7 = global ptr addrspace(7) addrspacecast (ptr null to ptr addrspace(7))
; CHECK-LABEL: nullptr8:
; The exact form of the GCN output depends on how the printer gets fixed.
; GCN-NEXT: .zeroes 4
; R600-NEXT: .long 0
@nullptr8 = global ptr addrspace(8) addrspacecast (ptr null to ptr addrspace(7))

llvm/test/CodeGen/AMDGPU/nullptr.ll

Show All 23 Lines
; CHECK-LABEL: nullptr_region: ; CHECK-LABEL: nullptr_region:
; CHECK-NEXT: .long -1 ; CHECK-NEXT: .long -1
@nullptr_region = global ptr addrspace(2) addrspacecast (ptr null to ptr addrspace(2)) @nullptr_region = global ptr addrspace(2) addrspacecast (ptr null to ptr addrspace(2))
; CHECK-LABEL: nullptr6: ; CHECK-LABEL: nullptr6:
; R600-NEXT: .long 0 ; R600-NEXT: .long 0
@nullptr6 = global ptr addrspace(6) addrspacecast (ptr null to ptr addrspace(6)) @nullptr6 = global ptr addrspace(6) addrspacecast (ptr null to ptr addrspace(6))
; CHECK-LABEL: nullptr7: ; FIXME: AsmPrinter can't handle 128-bit constants
; R600-NEXT: .long 0 ; FIXME-LABEL: nullptr7:
@nullptr7 = global ptr addrspace(7) addrspacecast (ptr null to ptr addrspace(7)) ; FIXME-R600-NEXT: .long 0
; FIXME @nullptr7 = global ptr addrspace(7) addrspacecast (ptr null to ptr addrspace(7))
; CHECK-LABEL: nullptr8:
; R600-NEXT: .long 0 ; FIXME-LABEL: nullptr8:
@nullptr8 = global ptr addrspace(8) addrspacecast (ptr null to ptr addrspace(8)) ; FIXME-R600-NEXT: .long 0
; FIXME @nullptr8 = global ptr addrspace(8) addrspacecast (ptr null to ptr addrspace(8))
; CHECK-LABEL: nullptr9: ; CHECK-LABEL: nullptr9:
; R600-NEXT: .long 0 ; R600-NEXT: .long 0
@nullptr9 = global ptr addrspace(9) addrspacecast (ptr null to ptr addrspace(9)) @nullptr9 = global ptr addrspace(9) addrspacecast (ptr null to ptr addrspace(9))
; CHECK-LABEL: nullptr10: ; CHECK-LABEL: nullptr10:
; R600-NEXT: .long 0 ; R600-NEXT: .long 0
@nullptr10 = global ptr addrspace(10) addrspacecast (ptr null to ptr addrspace(10)) @nullptr10 = global ptr addrspace(10) addrspacecast (ptr null to ptr addrspace(10))
▲ Show 20 Lines • Show All 73 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/promote-alloca-lifetime.ll

; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -passes=amdgpu-promote-alloca %s | FileCheck -check-prefix=OPT %s ; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -passes=amdgpu-promote-alloca %s | FileCheck -check-prefix=OPT %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
declare void @llvm.lifetime.start.p5(i64, ptr addrspace(5) nocapture) #0 declare void @llvm.lifetime.start.p5(i64, ptr addrspace(5) nocapture) #0
declare void @llvm.lifetime.end.p5(i64, ptr addrspace(5) nocapture) #0 declare void @llvm.lifetime.end.p5(i64, ptr addrspace(5) nocapture) #0
; OPT-LABEL: @use_lifetime_promotable_lds( ; OPT-LABEL: @use_lifetime_promotable_lds(
; OPT-NOT: alloca i32 ; OPT-NOT: alloca i32
; OPT-NOT: llvm.lifetime ; OPT-NOT: llvm.lifetime
; OPT: store i32 %tmp3, ptr addrspace(3) ; OPT: store i32 %tmp3, ptr addrspace(3)
Show All 26 Lines

llvm/test/CodeGen/AMDGPU/promote-alloca-to-lds-select.ll

; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -passes=amdgpu-promote-alloca < %s | FileCheck %s ; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -passes=amdgpu-promote-alloca < %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; CHECK-LABEL: @lds_promoted_alloca_select_invalid_pointer_operand( ; CHECK-LABEL: @lds_promoted_alloca_select_invalid_pointer_operand(
; CHECK: %alloca = alloca i32 ; CHECK: %alloca = alloca i32
; CHECK: select i1 undef, ptr addrspace(5) undef, ptr addrspace(5) %alloca ; CHECK: select i1 undef, ptr addrspace(5) undef, ptr addrspace(5) %alloca
define amdgpu_kernel void @lds_promoted_alloca_select_invalid_pointer_operand() #0 { define amdgpu_kernel void @lds_promoted_alloca_select_invalid_pointer_operand() #0 {
%alloca = alloca i32, align 4, addrspace(5) %alloca = alloca i32, align 4, addrspace(5)
%select = select i1 undef, ptr addrspace(5) undef, ptr addrspace(5) %alloca %select = select i1 undef, ptr addrspace(5) undef, ptr addrspace(5) %alloca
store i32 0, ptr addrspace(5) %select, align 4 store i32 0, ptr addrspace(5) %select, align 4
▲ Show 20 Lines • Show All 121 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/sgpr-copy-local-cse.ll

; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx906 -verify-machineinstrs -o - %s | FileCheck %s ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx906 -verify-machineinstrs -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
; CHECK-LABEL: {{^}}t0: ; CHECK-LABEL: {{^}}t0:
; CHECK: s_load_dwordx2 s[[[PTR_LO:[0-9]+]]:[[PTR_HI:[0-9]+]]], s[4:5], 0x0 ; CHECK: s_load_dwordx2 s[[[PTR_LO:[0-9]+]]:[[PTR_HI:[0-9]+]]], s[4:5], 0x0
; CHECK: v_mov_b32_e32 v{{[0-9]+}}, s[[PTR_HI]] ; CHECK: v_mov_b32_e32 v{{[0-9]+}}, s[[PTR_HI]]
; There should be no redundant copies from PTR_HI. ; There should be no redundant copies from PTR_HI.
; CHECK-NOT: v_mov_b32_e32 v{{[0-9]+}}, s[[PTR_HI]] ; CHECK-NOT: v_mov_b32_e32 v{{[0-9]+}}, s[[PTR_HI]]
define protected amdgpu_kernel void @t0(ptr addrspace(1) %p, i32 %i0, i32 %j0, i32 %k0) { define protected amdgpu_kernel void @t0(ptr addrspace(1) %p, i32 %i0, i32 %j0, i32 %k0) {
Show All 16 Lines

llvm/test/CodeGen/AMDGPU/splitkit-getsubrangeformask.ll

Show First 20 Lines • Show All 43 Lines • ▼ Show 20 Linesdefine amdgpu_gs void @_amdgpu_gs_main(i32 inreg %primShaderTableAddrLow, <31 x i32> inreg %userData) {
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM1:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM undef %74:sreg_64, 0, 0 :: (invariant load (s128) from `ptr addrspace(4) undef`, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM1:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM undef %74:sreg_64, 0, 0 :: (invariant load (s128) from `ptr addrspace(4) undef`, addrspace 4)
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM undef %132:sgpr_128, 0, 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM undef %132:sgpr_128, 0, 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: KILL undef %74:sreg_64 ; CHECK-NEXT: KILL undef %74:sreg_64
; CHECK-NEXT: KILL undef %132:sgpr_128 ; CHECK-NEXT: KILL undef %132:sgpr_128
; CHECK-NEXT: KILL %130.sub0, %130.sub1 ; CHECK-NEXT: KILL %130.sub0, %130.sub1
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM1:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM [[S_LOAD_DWORDX4_IMM]], 0, 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM1:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM [[S_LOAD_DWORDX4_IMM]], 0, 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec ; CHECK-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
; CHECK-NEXT: undef %302.sub1:sgpr_128 = S_MOV_B32 0 ; CHECK-NEXT: undef %302.sub1:sgpr_128 = S_MOV_B32 0
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], undef %89:sgpr_128, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], undef %89:sgpr_128, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN1:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM1]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN1:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM1]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: KILL undef %89:sgpr_128 ; CHECK-NEXT: KILL undef %89:sgpr_128
; CHECK-NEXT: [[S_SUB_I32_:%[0-9]+]]:sreg_32 = S_SUB_I32 [[S_BUFFER_LOAD_DWORD_IMM]], 29, implicit-def dead $scc ; CHECK-NEXT: [[S_SUB_I32_:%[0-9]+]]:sreg_32 = S_SUB_I32 [[S_BUFFER_LOAD_DWORD_IMM]], 29, implicit-def dead $scc
; CHECK-NEXT: [[S_SUB_I32_1:%[0-9]+]]:sreg_32 = S_SUB_I32 [[S_BUFFER_LOAD_DWORD_IMM]], 30, implicit-def dead $scc ; CHECK-NEXT: [[S_SUB_I32_1:%[0-9]+]]:sreg_32 = S_SUB_I32 [[S_BUFFER_LOAD_DWORD_IMM]], 30, implicit-def dead $scc
; CHECK-NEXT: [[S_SUB_I32_2:%[0-9]+]]:sreg_32 = S_SUB_I32 [[S_BUFFER_LOAD_DWORD_IMM1]], 31, implicit-def dead $scc ; CHECK-NEXT: [[S_SUB_I32_2:%[0-9]+]]:sreg_32 = S_SUB_I32 [[S_BUFFER_LOAD_DWORD_IMM1]], 31, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_U32_:%[0-9]+]]:sreg_32 = S_ADD_U32 [[COPY5]], 64, implicit-def $scc ; CHECK-NEXT: [[S_ADD_U32_:%[0-9]+]]:sreg_32 = S_ADD_U32 [[COPY5]], 64, implicit-def $scc
; CHECK-NEXT: [[S_ADDC_U32_:%[0-9]+]]:sreg_32 = S_ADDC_U32 undef %54:sreg_32, 0, implicit-def dead $scc, implicit $scc ; CHECK-NEXT: [[S_ADDC_U32_:%[0-9]+]]:sreg_32 = S_ADDC_U32 undef %54:sreg_32, 0, implicit-def dead $scc, implicit $scc
; CHECK-NEXT: undef %149.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_]], [[S_LSHL_B32_]], implicit-def $scc ; CHECK-NEXT: undef %149.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_]], [[S_LSHL_B32_]], implicit-def $scc
; CHECK-NEXT: %149.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_]], [[S_ASHR_I32_]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %149.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_]], [[S_ASHR_I32_]], implicit-def dead $scc, implicit $scc
Show All 34 Linesdefine amdgpu_gs void @_amdgpu_gs_main(i32 inreg %primShaderTableAddrLow, <31 x i32> inreg %userData) {
; CHECK-NEXT: undef %293.sub0:sreg_64 = S_ADD_U32 [[COPY7]], [[S_LSHL_B32_2]], implicit-def $scc ; CHECK-NEXT: undef %293.sub0:sreg_64 = S_ADD_U32 [[COPY7]], [[S_LSHL_B32_2]], implicit-def $scc
; CHECK-NEXT: %293.sub1:sreg_64 = S_ADDC_U32 undef %45:sreg_32, [[S_ASHR_I32_2]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %293.sub1:sreg_64 = S_ADDC_U32 undef %45:sreg_32, [[S_ASHR_I32_2]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_LSHL_B32_]], 16, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_LSHL_B32_]], 16, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_1:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_LSHL_B32_2]], 16, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_1:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_LSHL_B32_2]], 16, implicit-def dead $scc
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR %302, [[S_ADD_I32_]], 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR %302, [[S_ADD_I32_]], 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR1:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR %302, undef %314:sreg_32, 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR1:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR %302, undef %314:sreg_32, 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR2:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR %302, [[S_ADD_I32_1]], 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR2:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR %302, [[S_ADD_I32_1]], 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM2:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM %302, 16, 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM2:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM %302, 16, 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET undef %118:sgpr_128, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET undef %118:sgpr_128, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR3:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR undef %369:sgpr_128, undef %370:sreg_32, 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR3:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR undef %369:sgpr_128, undef %370:sreg_32, 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM3:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM undef %380:sgpr_128, 16, 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM3:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM undef %380:sgpr_128, 16, 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM3:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %156, 0, 0 :: (invariant load (s128) from %ir.92, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM3:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %156, 0, 0 :: (invariant load (s128) from %ir.92, addrspace 4)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM4:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %163, 0, 0 :: (invariant load (s128) from %ir.97, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM4:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %163, 0, 0 :: (invariant load (s128) from %ir.97, addrspace 4)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM5:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %176, 0, 0 :: (invariant load (s128) from %ir.104, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM5:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %176, 0, 0 :: (invariant load (s128) from %ir.104, addrspace 4)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM6:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %183, 0, 0 :: (invariant load (s128) from %ir.109, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM6:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %183, 0, 0 :: (invariant load (s128) from %ir.109, addrspace 4)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM7:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %190, 0, 0 :: (invariant load (s128) from %ir.114, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM7:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %190, 0, 0 :: (invariant load (s128) from %ir.114, addrspace 4)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN2:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM2]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN2:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM2]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR4:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR undef %364:sgpr_128, [[S_ADD_I32_]], 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR4:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR undef %364:sgpr_128, [[S_ADD_I32_]], 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR5:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR undef %375:sgpr_128, [[S_ADD_I32_1]], 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR5:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR undef %375:sgpr_128, [[S_ADD_I32_1]], 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[S_ADD_I32_2:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR]], -98, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_2:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR]], -98, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_3:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR1]], -114, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_3:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR1]], -114, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_4:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR2]], -130, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_4:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR2]], -130, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_5:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM2]], -178, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_5:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM2]], -178, implicit-def dead $scc
; CHECK-NEXT: undef %327.sub0:sreg_64 = S_ADD_U32 [[COPY8]], [[S_LSHL_B32_]], implicit-def $scc ; CHECK-NEXT: undef %327.sub0:sreg_64 = S_ADD_U32 [[COPY8]], [[S_LSHL_B32_]], implicit-def $scc
; CHECK-NEXT: %327.sub1:sreg_64 = S_ADDC_U32 undef %42:sreg_32, [[S_ASHR_I32_]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %327.sub1:sreg_64 = S_ADDC_U32 undef %42:sreg_32, [[S_ASHR_I32_]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: undef %335.sub0:sreg_64 = S_ADD_U32 [[COPY9]], [[S_LSHL_B32_]], implicit-def $scc ; CHECK-NEXT: undef %335.sub0:sreg_64 = S_ADD_U32 [[COPY9]], [[S_LSHL_B32_]], implicit-def $scc
; CHECK-NEXT: %335.sub1:sreg_64 = S_ADDC_U32 undef %39:sreg_32, [[S_ASHR_I32_]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %335.sub1:sreg_64 = S_ADDC_U32 undef %39:sreg_32, [[S_ASHR_I32_]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: undef %343.sub0:sreg_64 = S_ADD_U32 [[COPY9]], [[S_LSHL_B32_1]], implicit-def $scc ; CHECK-NEXT: undef %343.sub0:sreg_64 = S_ADD_U32 [[COPY9]], [[S_LSHL_B32_1]], implicit-def $scc
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM8:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %200, 0, 0 :: (invariant load (s128) from %ir.121, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM8:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %200, 0, 0 :: (invariant load (s128) from %ir.121, addrspace 4)
; CHECK-NEXT: %343.sub1:sreg_64 = S_ADDC_U32 undef %39:sreg_32, [[S_ASHR_I32_1]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %343.sub1:sreg_64 = S_ADDC_U32 undef %39:sreg_32, [[S_ASHR_I32_1]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: undef %351.sub0:sreg_64 = S_ADD_U32 [[COPY9]], [[S_LSHL_B32_2]], implicit-def $scc ; CHECK-NEXT: undef %351.sub0:sreg_64 = S_ADD_U32 [[COPY9]], [[S_LSHL_B32_2]], implicit-def $scc
; CHECK-NEXT: %351.sub1:sreg_64 = S_ADDC_U32 undef %39:sreg_32, [[S_ASHR_I32_2]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %351.sub1:sreg_64 = S_ADDC_U32 undef %39:sreg_32, [[S_ASHR_I32_2]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_LSHL_B32_3:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY10]], 4, implicit-def dead $scc ; CHECK-NEXT: [[S_LSHL_B32_3:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY10]], 4, implicit-def dead $scc
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN3:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM3]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN3:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM3]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_ADD_I32_6:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_LSHL_B32_3]], 16, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_6:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_LSHL_B32_3]], 16, implicit-def dead $scc
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR6:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR undef %396:sgpr_128, [[S_ADD_I32_6]], 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_SGPR6:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_SGPR undef %396:sgpr_128, [[S_ADD_I32_6]], 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN4:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM4]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN4:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM4]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM9:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %50, 224, 0 :: (invariant load (s128) from %ir.126, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM9:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %50, 224, 0 :: (invariant load (s128) from %ir.126, addrspace 4)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM10:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %210, 0, 0 :: (invariant load (s128) from %ir.127, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM10:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %210, 0, 0 :: (invariant load (s128) from %ir.127, addrspace 4)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN5:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM5]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN5:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM5]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM11:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %217, 0, 0 :: (invariant load (s128) from %ir.132, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM11:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %217, 0, 0 :: (invariant load (s128) from %ir.132, addrspace 4)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM12:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %224, 0, 0 :: (invariant load (s128) from %ir.137, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM12:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %224, 0, 0 :: (invariant load (s128) from %ir.137, addrspace 4)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN6:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM6]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN6:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM6]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN7:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM7]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN7:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM7]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN8:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM8]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN8:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM8]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_ADD_I32_7:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR4]], -217, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_7:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR4]], -217, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_8:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -233, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_8:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -233, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_9:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR5]], -249, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_9:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR5]], -249, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_10:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM3]], -297, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_10:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM3]], -297, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_11:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -313, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_11:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -313, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_12:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -329, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_12:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -329, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_13:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -345, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_13:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -345, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_14:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR6]], -441, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_14:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR6]], -441, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_U32_3:%[0-9]+]]:sreg_32 = S_ADD_U32 [[COPY1]], 160, implicit-def $scc ; CHECK-NEXT: [[S_ADD_U32_3:%[0-9]+]]:sreg_32 = S_ADD_U32 [[COPY1]], 160, implicit-def $scc
; CHECK-NEXT: [[S_ADDC_U32_3:%[0-9]+]]:sreg_32 = S_ADDC_U32 undef %36:sreg_32, 0, implicit-def dead $scc, implicit $scc ; CHECK-NEXT: [[S_ADDC_U32_3:%[0-9]+]]:sreg_32 = S_ADDC_U32 undef %36:sreg_32, 0, implicit-def dead $scc, implicit $scc
; CHECK-NEXT: undef %411.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_3]], [[S_LSHL_B32_2]], implicit-def $scc ; CHECK-NEXT: undef %411.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_3]], [[S_LSHL_B32_2]], implicit-def $scc
; CHECK-NEXT: %411.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_3]], [[S_ASHR_I32_2]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %411.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_3]], [[S_ASHR_I32_2]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_LSHL_B32_4:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY11]], 4, implicit-def dead $scc ; CHECK-NEXT: [[S_LSHL_B32_4:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY11]], 4, implicit-def dead $scc
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN9:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM10]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN9:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM10]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_ASHR_I32_4:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[S_LSHL_B32_4]], 31, implicit-def dead $scc ; CHECK-NEXT: [[S_ASHR_I32_4:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[S_LSHL_B32_4]], 31, implicit-def dead $scc
; CHECK-NEXT: undef %425.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_3]], [[S_LSHL_B32_4]], implicit-def $scc ; CHECK-NEXT: undef %425.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_3]], [[S_LSHL_B32_4]], implicit-def $scc
; CHECK-NEXT: %425.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_3]], [[S_ASHR_I32_4]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %425.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_3]], [[S_ASHR_I32_4]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_ADD_U32_4:%[0-9]+]]:sreg_32 = S_ADD_U32 %56.sub0, 168, implicit-def $scc ; CHECK-NEXT: [[S_ADD_U32_4:%[0-9]+]]:sreg_32 = S_ADD_U32 %56.sub0, 168, implicit-def $scc
; CHECK-NEXT: [[S_ADDC_U32_4:%[0-9]+]]:sreg_32 = S_ADDC_U32 undef %57:sreg_32, 0, implicit-def dead $scc, implicit $scc ; CHECK-NEXT: [[S_ADDC_U32_4:%[0-9]+]]:sreg_32 = S_ADDC_U32 undef %57:sreg_32, 0, implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM13:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %241, 0, 0 :: (invariant load (s128) from %ir.147, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM13:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %241, 0, 0 :: (invariant load (s128) from %ir.147, addrspace 4)
; CHECK-NEXT: [[S_LSHL_B32_5:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY4]], 3, implicit-def dead $scc ; CHECK-NEXT: [[S_LSHL_B32_5:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY4]], 3, implicit-def dead $scc
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN10:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM11]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN10:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM11]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_ASHR_I32_5:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[S_LSHL_B32_5]], 31, implicit-def dead $scc ; CHECK-NEXT: [[S_ASHR_I32_5:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[S_LSHL_B32_5]], 31, implicit-def dead $scc
; CHECK-NEXT: undef %441.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_4]], [[S_LSHL_B32_5]], implicit-def $scc ; CHECK-NEXT: undef %441.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_4]], [[S_LSHL_B32_5]], implicit-def $scc
; CHECK-NEXT: %441.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_4]], [[S_ASHR_I32_5]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %441.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_4]], [[S_ASHR_I32_5]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_LOAD_DWORD_IMM:%[0-9]+]]:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM %441, 0, 0 :: (invariant load (s32) from %ir.269, align 8, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORD_IMM:%[0-9]+]]:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM %441, 0, 0 :: (invariant load (s32) from %ir.269, align 8, addrspace 4)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM14:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %253, 0, 0 :: (invariant load (s128) from %ir.154, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM14:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %253, 0, 0 :: (invariant load (s128) from %ir.154, addrspace 4)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN11:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM12]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN11:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM12]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM15:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %261, 0, 0 :: (invariant load (s128) from %ir.159, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM15:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %261, 0, 0 :: (invariant load (s128) from %ir.159, addrspace 4)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN12:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM9]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN12:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM9]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN13:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM13]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN13:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM13]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: %71.sub3:sgpr_128 = S_MOV_B32 553734060 ; CHECK-NEXT: %71.sub3:sgpr_128 = S_MOV_B32 553734060
; CHECK-NEXT: %71.sub2:sgpr_128 = S_MOV_B32 -1 ; CHECK-NEXT: %71.sub2:sgpr_128 = S_MOV_B32 -1
; CHECK-NEXT: [[COPY13:%[0-9]+]]:sgpr_128 = COPY %71 ; CHECK-NEXT: [[COPY13:%[0-9]+]]:sgpr_128 = COPY %71
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM16:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %273, 0, 0 :: (invariant load (s128) from %ir.167, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM16:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %273, 0, 0 :: (invariant load (s128) from %ir.167, addrspace 4)
; CHECK-NEXT: [[COPY13]].sub1:sgpr_128 = COPY %302.sub1 ; CHECK-NEXT: [[COPY13]].sub1:sgpr_128 = COPY %302.sub1
; CHECK-NEXT: [[COPY13]].sub0:sgpr_128 = COPY [[S_LOAD_DWORD_IMM]] ; CHECK-NEXT: [[COPY13]].sub0:sgpr_128 = COPY [[S_LOAD_DWORD_IMM]]
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM4:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM [[COPY13]], 0, 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM4:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM [[COPY13]], 0, 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN14:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM14]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN14:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM14]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN15:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM15]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN15:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM15]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM17:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %286, 0, 0 :: (invariant load (s128) from %ir.175, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM17:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %286, 0, 0 :: (invariant load (s128) from %ir.175, addrspace 4)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM18:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %293, 0, 0 :: (invariant load (s128) from %ir.180, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM18:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %293, 0, 0 :: (invariant load (s128) from %ir.180, addrspace 4)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN16:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM16]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN16:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM16]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_LSHL_B32_6:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY3]], 3, implicit-def dead $scc ; CHECK-NEXT: [[S_LSHL_B32_6:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY3]], 3, implicit-def dead $scc
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET1:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET [[S_LOAD_DWORDX4_IMM1]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET1:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET [[S_LOAD_DWORDX4_IMM1]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_ASHR_I32_6:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[S_LSHL_B32_6]], 31, implicit-def dead $scc ; CHECK-NEXT: [[S_ASHR_I32_6:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[S_LSHL_B32_6]], 31, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_15:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM4]], -467, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_15:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM4]], -467, implicit-def dead $scc
; CHECK-NEXT: undef %453.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_4]], [[S_LSHL_B32_6]], implicit-def $scc ; CHECK-NEXT: undef %453.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_4]], [[S_LSHL_B32_6]], implicit-def $scc
; CHECK-NEXT: %453.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_4]], [[S_ASHR_I32_6]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %453.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_4]], [[S_ASHR_I32_6]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM %453, 0, 0 :: (invariant load (s64) from %ir.277, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX2_IMM:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM %453, 0, 0 :: (invariant load (s64) from %ir.277, addrspace 4)
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET2:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET [[S_LOAD_DWORDX4_IMM17]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET2:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET [[S_LOAD_DWORDX4_IMM17]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET3:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET [[S_LOAD_DWORDX4_IMM18]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_DWORD_OFFSET3:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET [[S_LOAD_DWORDX4_IMM18]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM19:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %327, 0, 0 :: (invariant load (s128) from %ir.202, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM19:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %327, 0, 0 :: (invariant load (s128) from %ir.202, addrspace 4)
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM20:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %335, 0, 0 :: (invariant load (s128) from %ir.208, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM20:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %335, 0, 0 :: (invariant load (s128) from %ir.208, addrspace 4)
; CHECK-NEXT: [[COPY14:%[0-9]+]]:sgpr_128 = COPY %71 ; CHECK-NEXT: [[COPY14:%[0-9]+]]:sgpr_128 = COPY %71
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM21:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %343, 0, 0 :: (invariant load (s128) from %ir.213, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM21:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %343, 0, 0 :: (invariant load (s128) from %ir.213, addrspace 4)
; CHECK-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32 = S_AND_B32 [[S_LOAD_DWORDX2_IMM]].sub1, 65535, implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B32_:%[0-9]+]]:sreg_32 = S_AND_B32 [[S_LOAD_DWORDX2_IMM]].sub1, 65535, implicit-def dead $scc
; CHECK-NEXT: [[COPY14]].sub0:sgpr_128 = COPY [[S_LOAD_DWORDX2_IMM]].sub0 ; CHECK-NEXT: [[COPY14]].sub0:sgpr_128 = COPY [[S_LOAD_DWORDX2_IMM]].sub0
; CHECK-NEXT: [[COPY14]].sub1:sgpr_128 = COPY [[S_AND_B32_]] ; CHECK-NEXT: [[COPY14]].sub1:sgpr_128 = COPY [[S_AND_B32_]]
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM5:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM [[COPY14]], 0, 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM5:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM [[COPY14]], 0, 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM22:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %351, 0, 0 :: (invariant load (s128) from %ir.218, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM22:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %351, 0, 0 :: (invariant load (s128) from %ir.218, addrspace 4)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN17:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM19]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN17:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM19]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN18:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM20]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN18:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM20]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_LSHL_B32_7:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY2]], 3, implicit-def dead $scc ; CHECK-NEXT: [[S_LSHL_B32_7:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY2]], 3, implicit-def dead $scc
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN19:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM21]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN19:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM21]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_ASHR_I32_7:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[S_LSHL_B32_7]], 31, implicit-def dead $scc ; CHECK-NEXT: [[S_ASHR_I32_7:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[S_LSHL_B32_7]], 31, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_16:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM5]], -468, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_16:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM5]], -468, implicit-def dead $scc
; CHECK-NEXT: undef %468.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_4]], [[S_LSHL_B32_7]], implicit-def $scc ; CHECK-NEXT: undef %468.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_4]], [[S_LSHL_B32_7]], implicit-def $scc
; CHECK-NEXT: %468.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_4]], [[S_ASHR_I32_7]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %468.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_4]], [[S_ASHR_I32_7]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN20:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM22]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN20:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM22]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[S_LOAD_DWORDX2_IMM1:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM %468, 0, 0 :: (invariant load (s64) from %ir.287, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX2_IMM1:%[0-9]+]]:sreg_64_xexec = S_LOAD_DWORDX2_IMM %468, 0, 0 :: (invariant load (s64) from %ir.287, addrspace 4)
; CHECK-NEXT: [[COPY15:%[0-9]+]]:sgpr_128 = COPY %71 ; CHECK-NEXT: [[COPY15:%[0-9]+]]:sgpr_128 = COPY %71
; CHECK-NEXT: [[S_AND_B32_1:%[0-9]+]]:sreg_32 = S_AND_B32 [[S_LOAD_DWORDX2_IMM1]].sub1, 65535, implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B32_1:%[0-9]+]]:sreg_32 = S_AND_B32 [[S_LOAD_DWORDX2_IMM1]].sub1, 65535, implicit-def dead $scc
; CHECK-NEXT: [[COPY15]].sub0:sgpr_128 = COPY [[S_LOAD_DWORDX2_IMM1]].sub0 ; CHECK-NEXT: [[COPY15]].sub0:sgpr_128 = COPY [[S_LOAD_DWORDX2_IMM1]].sub0
; CHECK-NEXT: [[COPY15]].sub1:sgpr_128 = COPY [[S_AND_B32_1]] ; CHECK-NEXT: [[COPY15]].sub1:sgpr_128 = COPY [[S_AND_B32_1]]
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM6:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM [[COPY15]], 0, 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM6:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM [[COPY15]], 0, 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM23:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %411, 0, 0 :: (invariant load (s128) from %ir.253, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM23:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %411, 0, 0 :: (invariant load (s128) from %ir.253, addrspace 4)
; CHECK-NEXT: [[S_LOAD_DWORD_IMM1:%[0-9]+]]:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM undef %488:sreg_64, 0, 0 :: (invariant load (s32) from `ptr addrspace(4) undef`, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORD_IMM1:%[0-9]+]]:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM undef %488:sreg_64, 0, 0 :: (invariant load (s32) from `ptr addrspace(4) undef`, addrspace 4)
; CHECK-NEXT: KILL %411.sub0, %411.sub1 ; CHECK-NEXT: KILL %411.sub0, %411.sub1
; CHECK-NEXT: KILL undef %488:sreg_64 ; CHECK-NEXT: KILL undef %488:sreg_64
; CHECK-NEXT: KILL [[COPY15]].sub0_sub1, [[COPY15]].sub2_sub3 ; CHECK-NEXT: KILL [[COPY15]].sub0_sub1, [[COPY15]].sub2_sub3
; CHECK-NEXT: [[S_LSHL_B32_8:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY12]], 3, implicit-def dead $scc ; CHECK-NEXT: [[S_LSHL_B32_8:%[0-9]+]]:sreg_32 = S_LSHL_B32 [[COPY12]], 3, implicit-def dead $scc
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM24:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %425, 0, 0 :: (invariant load (s128) from %ir.261, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM24:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %425, 0, 0 :: (invariant load (s128) from %ir.261, addrspace 4)
; CHECK-NEXT: [[S_ASHR_I32_8:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[S_LSHL_B32_8]], 31, implicit-def dead $scc ; CHECK-NEXT: [[S_ASHR_I32_8:%[0-9]+]]:sreg_32_xm0 = S_ASHR_I32 [[S_LSHL_B32_8]], 31, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_17:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM6]], -469, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_17:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM6]], -469, implicit-def dead $scc
; CHECK-NEXT: undef %485.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_4]], [[S_LSHL_B32_8]], implicit-def $scc ; CHECK-NEXT: undef %485.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_4]], [[S_LSHL_B32_8]], implicit-def $scc
; CHECK-NEXT: %485.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_4]], [[S_ASHR_I32_8]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %485.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_4]], [[S_ASHR_I32_8]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_LOAD_DWORD_IMM2:%[0-9]+]]:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM %485, 0, 0 :: (invariant load (s32) from %ir.298, align 8, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORD_IMM2:%[0-9]+]]:sreg_32_xm0_xexec = S_LOAD_DWORD_IMM %485, 0, 0 :: (invariant load (s32) from %ir.298, align 8, addrspace 4)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN21:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM23]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN21:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM23]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN22:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM24]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN22:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM24]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: KILL [[S_LOAD_DWORDX4_IMM24]] ; CHECK-NEXT: KILL [[S_LOAD_DWORDX4_IMM24]]
; CHECK-NEXT: KILL [[S_LOAD_DWORDX4_IMM23]] ; CHECK-NEXT: KILL [[S_LOAD_DWORDX4_IMM23]]
; CHECK-NEXT: [[S_AND_B32_2:%[0-9]+]]:sreg_32 = S_AND_B32 [[S_LOAD_DWORD_IMM1]], 65535, implicit-def dead $scc ; CHECK-NEXT: [[S_AND_B32_2:%[0-9]+]]:sreg_32 = S_AND_B32 [[S_LOAD_DWORD_IMM1]], 65535, implicit-def dead $scc
; CHECK-NEXT: [[COPY16:%[0-9]+]]:sgpr_128 = COPY %71 ; CHECK-NEXT: [[COPY16:%[0-9]+]]:sgpr_128 = COPY %71
; CHECK-NEXT: [[COPY16]].sub1:sgpr_128 = COPY [[S_AND_B32_2]] ; CHECK-NEXT: [[COPY16]].sub1:sgpr_128 = COPY [[S_AND_B32_2]]
; CHECK-NEXT: [[COPY16]].sub0:sgpr_128 = COPY [[S_LOAD_DWORD_IMM2]] ; CHECK-NEXT: [[COPY16]].sub0:sgpr_128 = COPY [[S_LOAD_DWORD_IMM2]]
; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM7:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM [[COPY16]], 0, 0 :: (dereferenceable invariant load (s32)) ; CHECK-NEXT: [[S_BUFFER_LOAD_DWORD_IMM7:%[0-9]+]]:sreg_32_xm0_xexec = S_BUFFER_LOAD_DWORD_IMM [[COPY16]], 0, 0 :: (dereferenceable invariant load (s32))
; CHECK-NEXT: [[S_ADD_I32_18:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM]], -474, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_18:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM]], -474, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_19:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -475, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_19:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -475, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_20:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -491, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_20:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -491, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_21:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -507, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_21:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -507, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_22:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -539, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_22:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_SGPR3]], -539, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_I32_23:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM7]], -473, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_23:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM7]], -473, implicit-def dead $scc
; CHECK-NEXT: [[S_ADD_U32_5:%[0-9]+]]:sreg_32 = S_ADD_U32 [[COPY]], 96, implicit-def $scc ; CHECK-NEXT: [[S_ADD_U32_5:%[0-9]+]]:sreg_32 = S_ADD_U32 [[COPY]], 96, implicit-def $scc
; CHECK-NEXT: [[S_ADDC_U32_5:%[0-9]+]]:sreg_32 = S_ADDC_U32 undef %33:sreg_32, 0, implicit-def dead $scc, implicit $scc ; CHECK-NEXT: [[S_ADDC_U32_5:%[0-9]+]]:sreg_32 = S_ADDC_U32 undef %33:sreg_32, 0, implicit-def dead $scc, implicit $scc
; CHECK-NEXT: undef %514.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_5]], [[S_LSHL_B32_]], implicit-def $scc ; CHECK-NEXT: undef %514.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_5]], [[S_LSHL_B32_]], implicit-def $scc
; CHECK-NEXT: %514.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_5]], [[S_ASHR_I32_]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %514.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_5]], [[S_ASHR_I32_]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM25:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %514, 0, 0 :: (invariant load (s128) from %ir.316, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM25:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %514, 0, 0 :: (invariant load (s128) from %ir.316, addrspace 4)
; CHECK-NEXT: undef %522.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_5]], [[S_LSHL_B32_1]], implicit-def $scc ; CHECK-NEXT: undef %522.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_5]], [[S_LSHL_B32_1]], implicit-def $scc
; CHECK-NEXT: %522.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_5]], [[S_ASHR_I32_1]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %522.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_5]], [[S_ASHR_I32_1]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM26:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %522, 0, 0 :: (invariant load (s128) from %ir.321, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM26:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %522, 0, 0 :: (invariant load (s128) from %ir.321, addrspace 4)
; CHECK-NEXT: undef %530.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_5]], [[S_LSHL_B32_2]], implicit-def $scc ; CHECK-NEXT: undef %530.sub0:sreg_64 = S_ADD_U32 [[S_ADD_U32_5]], [[S_LSHL_B32_2]], implicit-def $scc
; CHECK-NEXT: %530.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_5]], [[S_ASHR_I32_2]], implicit-def dead $scc, implicit $scc ; CHECK-NEXT: %530.sub1:sreg_64 = S_ADDC_U32 [[S_ADDC_U32_5]], [[S_ASHR_I32_2]], implicit-def dead $scc, implicit $scc
; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM27:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %530, 0, 0 :: (invariant load (s128) from %ir.326, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX4_IMM27:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM %530, 0, 0 :: (invariant load (s128) from %ir.326, addrspace 4)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN23:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM25]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN23:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM25]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN24:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM26]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN24:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM26]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN25:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM27]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 7) ; CHECK-NEXT: [[BUFFER_LOAD_FORMAT_X_IDXEN25:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_FORMAT_X_IDXEN [[V_MOV_B32_e32_]], [[S_LOAD_DWORDX4_IMM27]], 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), align 1, addrspace 8)
; CHECK-NEXT: KILL [[S_LOAD_DWORDX4_IMM27]] ; CHECK-NEXT: KILL [[S_LOAD_DWORDX4_IMM27]]
; CHECK-NEXT: KILL [[S_LOAD_DWORDX4_IMM25]] ; CHECK-NEXT: KILL [[S_LOAD_DWORDX4_IMM25]]
; CHECK-NEXT: KILL [[V_MOV_B32_e32_]] ; CHECK-NEXT: KILL [[V_MOV_B32_e32_]]
; CHECK-NEXT: KILL [[S_LOAD_DWORDX4_IMM26]] ; CHECK-NEXT: KILL [[S_LOAD_DWORDX4_IMM26]]
; CHECK-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 -2, [[BUFFER_LOAD_FORMAT_X_IDXEN]], 0, implicit $exec ; CHECK-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 -2, [[BUFFER_LOAD_FORMAT_X_IDXEN]], 0, implicit $exec
; CHECK-NEXT: [[V_ADD_U32_e64_1:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 -1, [[BUFFER_LOAD_FORMAT_X_IDXEN1]], 0, implicit $exec ; CHECK-NEXT: [[V_ADD_U32_e64_1:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 -1, [[BUFFER_LOAD_FORMAT_X_IDXEN1]], 0, implicit $exec
; CHECK-NEXT: [[V_ADD_U32_e64_2:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 -3, [[BUFFER_LOAD_FORMAT_X_IDXEN]], 0, implicit $exec ; CHECK-NEXT: [[V_ADD_U32_e64_2:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 -3, [[BUFFER_LOAD_FORMAT_X_IDXEN]], 0, implicit $exec
; CHECK-NEXT: [[V_OR_B32_e64_:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 [[V_ADD_U32_e64_]], [[V_ADD_U32_e64_1]], implicit $exec ; CHECK-NEXT: [[V_OR_B32_e64_:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 [[V_ADD_U32_e64_]], [[V_ADD_U32_e64_1]], implicit $exec
▲ Show 20 Lines • Show All 105 Lines • ▼ Show 20 Linesdefine amdgpu_gs void @_amdgpu_gs_main(i32 inreg %primShaderTableAddrLow, <31 x i32> inreg %userData) {
; CHECK-NEXT: [[V_ADD_U32_e64_30:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 -593, [[BUFFER_LOAD_FORMAT_X_IDXEN]], 0, implicit $exec ; CHECK-NEXT: [[V_ADD_U32_e64_30:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 -593, [[BUFFER_LOAD_FORMAT_X_IDXEN]], 0, implicit $exec
; CHECK-NEXT: [[V_OR_B32_e64_65:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 [[V_OR_B32_e64_64]], [[V_ADD_U32_e64_29]], implicit $exec ; CHECK-NEXT: [[V_OR_B32_e64_65:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 [[V_OR_B32_e64_64]], [[V_ADD_U32_e64_29]], implicit $exec
; CHECK-NEXT: [[S_LOAD_DWORDX8_IMM:%[0-9]+]]:sgpr_256 = S_LOAD_DWORDX8_IMM undef %564:sreg_64, 0, 0 :: (invariant load (s256) from `ptr addrspace(4) undef`, addrspace 4) ; CHECK-NEXT: [[S_LOAD_DWORDX8_IMM:%[0-9]+]]:sgpr_256 = S_LOAD_DWORDX8_IMM undef %564:sreg_64, 0, 0 :: (invariant load (s256) from `ptr addrspace(4) undef`, addrspace 4)
; CHECK-NEXT: [[V_OR_B32_e64_66:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 [[V_OR_B32_e64_65]], [[V_ADD_U32_e64_30]], implicit $exec ; CHECK-NEXT: [[V_OR_B32_e64_66:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 [[V_OR_B32_e64_65]], [[V_ADD_U32_e64_30]], implicit $exec
; CHECK-NEXT: [[S_ADD_I32_24:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM8]], -594, implicit-def dead $scc ; CHECK-NEXT: [[S_ADD_I32_24:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BUFFER_LOAD_DWORD_IMM8]], -594, implicit-def dead $scc
; CHECK-NEXT: [[V_OR_B32_e64_67:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 [[S_ADD_I32_24]], [[V_OR_B32_e64_66]], implicit $exec ; CHECK-NEXT: [[V_OR_B32_e64_67:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 [[S_ADD_I32_24]], [[V_OR_B32_e64_66]], implicit $exec
; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U32_e64 0, [[V_OR_B32_e64_67]], implicit $exec ; CHECK-NEXT: [[V_CMP_EQ_U32_e64_:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U32_e64 0, [[V_OR_B32_e64_67]], implicit $exec
; CHECK-NEXT: undef %693.sub3:vreg_128 = V_CNDMASK_B32_e64 0, 0, 0, 1, [[V_CMP_EQ_U32_e64_]], implicit $exec ; CHECK-NEXT: undef %693.sub3:vreg_128 = V_CNDMASK_B32_e64 0, 0, 0, 1, [[V_CMP_EQ_U32_e64_]], implicit $exec
; CHECK-NEXT: IMAGE_STORE_V4_V2_gfx10 %693, undef %578:vreg_64, [[S_LOAD_DWORDX8_IMM]], 15, 1, -1, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), addrspace 7) ; CHECK-NEXT: IMAGE_STORE_V4_V2_gfx10 %693, undef %578:vreg_64, [[S_LOAD_DWORDX8_IMM]], 15, 1, -1, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), addrspace 8)
; CHECK-NEXT: S_ENDPGM 0 ; CHECK-NEXT: S_ENDPGM 0
.expVert: .expVert:
%0 = extractelement <31 x i32> %userData, i64 2 %0 = extractelement <31 x i32> %userData, i64 2
%1 = extractelement <31 x i32> %userData, i64 3 %1 = extractelement <31 x i32> %userData, i64 3
%2 = extractelement <31 x i32> %userData, i64 4 %2 = extractelement <31 x i32> %userData, i64 4
%3 = extractelement <31 x i32> %userData, i64 7 %3 = extractelement <31 x i32> %userData, i64 7
%4 = extractelement <31 x i32> %userData, i64 8 %4 = extractelement <31 x i32> %userData, i64 8
%5 = extractelement <31 x i32> %userData, i64 9 %5 = extractelement <31 x i32> %userData, i64 9
▲ Show 20 Lines • Show All 401 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/unroll.ll

; RUN: opt -mtriple=amdgcn-- -passes='loop-unroll,simplifycfg,sroa' %s -S -o - | FileCheck %s ; RUN: opt -mtriple=amdgcn-- -passes='loop-unroll,simplifycfg,sroa' %s -S -o - | FileCheck %s
; RUN: opt -mtriple=r600-- -passes='loop-unroll,simplifycfg,sroa' %s -S -o - | FileCheck %s ; RUN: opt -mtriple=r600-- -passes='loop-unroll,simplifycfg,sroa' %s -S -o - | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; This test contains a simple loop that initializes an array declared in ; This test contains a simple loop that initializes an array declared in
; private memory. We want to make sure these kinds of loops are always ; private memory. We want to make sure these kinds of loops are always
; unrolled, because private memory is slow. ; unrolled, because private memory is slow.
; CHECK-LABEL: @private_memory ; CHECK-LABEL: @private_memory
; CHECK-NOT: alloca ; CHECK-NOT: alloca
; CHECK: store i32 5, ptr addrspace(1) %out ; CHECK: store i32 5, ptr addrspace(1) %out
▲ Show 20 Lines • Show All 122 Lines • Show Last 20 Lines

llvm/test/CodeGen/AMDGPU/unsupported-image-a16.ll

; RUN: not --crash llc -global-isel=0 -march=amdgcn -mcpu=fiji -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=SDAG-ERR %s ; RUN: not --crash llc -global-isel=0 -march=amdgcn -mcpu=fiji -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=SDAG-ERR %s
; RUN: not --crash llc -global-isel=1 -global-isel-abort=1 -march=amdgcn -mcpu=fiji -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=GISEL-ERR %s ; RUN: not --crash llc -global-isel=1 -global-isel-abort=1 -march=amdgcn -mcpu=fiji -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=GISEL-ERR %s
; Make sure this doesn't assert on targets without the r128-16 ; Make sure this doesn't assert on targets without the r128-16
; feature, and instead generates a selection error. ; feature, and instead generates a selection error.
; SDAG-ERR: LLVM ERROR: Cannot select: intrinsic %llvm.amdgcn.image.load.1d ; SDAG-ERR: LLVM ERROR: Cannot select: intrinsic %llvm.amdgcn.image.load.1d
; GISEL-ERR: LLVM ERROR: unable to legalize instruction: %{{[0-9]+}}:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, %{{[0-9]+}}:_(s16), %{{[0-9]+}}:_(<8 x s32>), 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) (in function: load_1d) ; GISEL-ERR: LLVM ERROR: unable to legalize instruction: %{{[0-9]+}}:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.load.1d), 15, %{{[0-9]+}}:_(s16), %{{[0-9]+}}:_(<8 x s32>), 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8) (in function: load_1d)
define amdgpu_ps <4 x float> @load_1d(<8 x i32> inreg %rsrc, <2 x i16> %coords) { define amdgpu_ps <4 x float> @load_1d(<8 x i32> inreg %rsrc, <2 x i16> %coords) {
main_body: main_body:
%s = extractelement <2 x i16> %coords, i32 0 %s = extractelement <2 x i16> %coords, i32 0
%v = call <4 x float> @llvm.amdgcn.image.load.1d.v4f32.i16(i32 15, i16 %s, <8 x i32> %rsrc, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.load.1d.v4f32.i16(i32 15, i16 %s, <8 x i32> %rsrc, i32 0, i32 0)
ret <4 x float> %v ret <4 x float> %v
} }
declare <4 x float> @llvm.amdgcn.image.load.1d.v4f32.i16(i32 immarg, i16, <8 x i32>, i32 immarg, i32 immarg) #0 declare <4 x float> @llvm.amdgcn.image.load.1d.v4f32.i16(i32 immarg, i16, <8 x i32>, i32 immarg, i32 immarg) #0
attributes #0 = { nounwind readonly } attributes #0 = { nounwind readonly }

llvm/test/CodeGen/AMDGPU/unsupported-image-g16.ll

; RUN: not --crash llc -global-isel=0 -march=amdgcn -mcpu=fiji -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=SDAG-ERR %s ; RUN: not --crash llc -global-isel=0 -march=amdgcn -mcpu=fiji -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=SDAG-ERR %s
; RUN: not --crash llc -global-isel=0 -march=amdgcn -mcpu=gfx900 -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=SDAG-ERR %s ; RUN: not --crash llc -global-isel=0 -march=amdgcn -mcpu=gfx900 -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=SDAG-ERR %s
; RUN: not --crash llc -global-isel=1 -march=amdgcn -mcpu=fiji -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=GISEL-ERR %s ; RUN: not --crash llc -global-isel=1 -march=amdgcn -mcpu=fiji -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=GISEL-ERR %s
; RUN: not --crash llc -global-isel=1 -march=amdgcn -mcpu=gfx900 -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=GISEL-ERR %s ; RUN: not --crash llc -global-isel=1 -march=amdgcn -mcpu=gfx900 -verify-machineinstrs -o /dev/null %s 2>&1 | FileCheck -check-prefix=GISEL-ERR %s
; Make sure this doesn't assert on targets without the g16 feature, and instead ; Make sure this doesn't assert on targets without the g16 feature, and instead
; generates a selection error. ; generates a selection error.
; SDAG-ERR: LLVM ERROR: Cannot select: intrinsic %llvm.amdgcn.image.sample.d.1d ; SDAG-ERR: LLVM ERROR: Cannot select: intrinsic %llvm.amdgcn.image.sample.d.1d
; GISEL-ERR: LLVM ERROR: unable to legalize instruction: %{{[0-9]+}}:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, %{{[0-9]+}}:_(s16), %{{[0-9]+}}:_(s16), %{{[0-9]+}}:_(s32), %{{[0-9]+}}:_(<8 x s32>), %{{[0-9]+}}:_(<4 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 7) (in function: sample_d_1d) ; GISEL-ERR: LLVM ERROR: unable to legalize instruction: %{{[0-9]+}}:_(<4 x s32>) = G_AMDGPU_INTRIN_IMAGE_LOAD intrinsic(@llvm.amdgcn.image.sample.d.1d), 15, %{{[0-9]+}}:_(s16), %{{[0-9]+}}:_(s16), %{{[0-9]+}}:_(s32), %{{[0-9]+}}:_(<8 x s32>), %{{[0-9]+}}:_(<4 x s32>), 0, 0, 0 :: (dereferenceable load (<4 x s32>), addrspace 8) (in function: sample_d_1d)
define amdgpu_ps <4 x float> @sample_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, float %s) { define amdgpu_ps <4 x float> @sample_d_1d(<8 x i32> inreg %rsrc, <4 x i32> inreg %samp, half %dsdh, half %dsdv, float %s) {
main_body: main_body:
%v = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f32(i32 15, half %dsdh, half %dsdv, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0) %v = call <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f32(i32 15, half %dsdh, half %dsdv, float %s, <8 x i32> %rsrc, <4 x i32> %samp, i1 0, i32 0, i32 0)
ret <4 x float> %v ret <4 x float> %v
} }
declare <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f32(i32, half, half, float, <8 x i32>, <4 x i32>, i1, i32, i32) #0 declare <4 x float> @llvm.amdgcn.image.sample.d.1d.v4f32.f16.f32(i32, half, half, float, <8 x i32>, <4 x i32>, i1, i32, i32) #0
attributes #0 = { nounwind readonly } attributes #0 = { nounwind readonly }

llvm/test/CodeGen/AMDGPU/vgpr-liverange-ir.ll

Show First 20 Lines • Show All 631 Lines • ▼ Show 20 Linesdefine protected amdgpu_kernel void @nested_waterfalls(ptr addrspace(1) %tex.coerce) local_unnamed_addr {
; SI-NEXT: [[V_CMP_EQ_U64_e64_5:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 killed [[REG_SEQUENCE7]], [[GLOBAL_LOAD_DWORDX4_2]].sub2_sub3, implicit $exec ; SI-NEXT: [[V_CMP_EQ_U64_e64_5:%[0-9]+]]:sreg_32_xm0_xexec = V_CMP_EQ_U64_e64 killed [[REG_SEQUENCE7]], [[GLOBAL_LOAD_DWORDX4_2]].sub2_sub3, implicit $exec
; SI-NEXT: [[S_AND_B32_3:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 killed [[V_CMP_EQ_U64_e64_4]], killed [[V_CMP_EQ_U64_e64_5]], implicit-def dead $scc ; SI-NEXT: [[S_AND_B32_3:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_B32 killed [[V_CMP_EQ_U64_e64_4]], killed [[V_CMP_EQ_U64_e64_5]], implicit-def dead $scc
; SI-NEXT: [[REG_SEQUENCE8:%[0-9]+]]:sgpr_128 = REG_SEQUENCE killed [[V_READFIRSTLANE_B32_8]], %subreg.sub0, killed [[V_READFIRSTLANE_B32_9]], %subreg.sub1, killed [[V_READFIRSTLANE_B32_10]], %subreg.sub2, killed [[V_READFIRSTLANE_B32_11]], %subreg.sub3 ; SI-NEXT: [[REG_SEQUENCE8:%[0-9]+]]:sgpr_128 = REG_SEQUENCE killed [[V_READFIRSTLANE_B32_8]], %subreg.sub0, killed [[V_READFIRSTLANE_B32_9]], %subreg.sub1, killed [[V_READFIRSTLANE_B32_10]], %subreg.sub2, killed [[V_READFIRSTLANE_B32_11]], %subreg.sub3
; SI-NEXT: [[S_AND_SAVEEXEC_B32_1:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_3]], implicit-def $exec, implicit-def dead $scc, implicit $exec ; SI-NEXT: [[S_AND_SAVEEXEC_B32_1:%[0-9]+]]:sreg_32_xm0_xexec = S_AND_SAVEEXEC_B32 killed [[S_AND_B32_3]], implicit-def $exec, implicit-def dead $scc, implicit $exec
; SI-NEXT: {{ $}} ; SI-NEXT: {{ $}}
; SI-NEXT: bb.5: ; SI-NEXT: bb.5:
; SI-NEXT: successors: %bb.4(0x40000000), %bb.6(0x40000000) ; SI-NEXT: successors: %bb.4(0x40000000), %bb.6(0x40000000)
; SI-NEXT: {{ $}} ; SI-NEXT: {{ $}}
; SI-NEXT: [[IMAGE_SAMPLE_V1_V2_gfx10_:%[0-9]+]]:vgpr_32 = IMAGE_SAMPLE_V1_V2_gfx10 undef %22:vreg_64, [[REG_SEQUENCE5]], killed [[REG_SEQUENCE8]], 1, 1, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), addrspace 7) ; SI-NEXT: [[IMAGE_SAMPLE_V1_V2_gfx10_:%[0-9]+]]:vgpr_32 = IMAGE_SAMPLE_V1_V2_gfx10 undef %22:vreg_64, [[REG_SEQUENCE5]], killed [[REG_SEQUENCE8]], 1, 1, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s32), addrspace 8)
; SI-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, killed [[S_AND_SAVEEXEC_B32_1]], implicit-def dead $scc ; SI-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, killed [[S_AND_SAVEEXEC_B32_1]], implicit-def dead $scc
; SI-NEXT: SI_WATERFALL_LOOP %bb.4, implicit $exec ; SI-NEXT: SI_WATERFALL_LOOP %bb.4, implicit $exec
; SI-NEXT: {{ $}} ; SI-NEXT: {{ $}}
; SI-NEXT: bb.6: ; SI-NEXT: bb.6:
; SI-NEXT: successors: %bb.2(0x40000000), %bb.7(0x40000000) ; SI-NEXT: successors: %bb.2(0x40000000), %bb.7(0x40000000)
; SI-NEXT: {{ $}} ; SI-NEXT: {{ $}}
; SI-NEXT: $exec_lo = S_MOV_B32 killed [[S_MOV_B32_1]] ; SI-NEXT: $exec_lo = S_MOV_B32 killed [[S_MOV_B32_1]]
; SI-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, killed [[S_AND_SAVEEXEC_B32_]], implicit-def dead $scc ; SI-NEXT: $exec_lo = S_XOR_B32_term $exec_lo, killed [[S_AND_SAVEEXEC_B32_]], implicit-def dead $scc
Show All 31 Lines

llvm/test/CodeGen/MIR/AMDGPU/custom-pseudo-source-values.ll

; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 -stop-after finalize-isel -o %t.mir %s ; RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 -stop-after finalize-isel -o %t.mir %s
; RUN: llc -run-pass=none -verify-machineinstrs %t.mir -o - | FileCheck %s ; RUN: llc -run-pass=none -verify-machineinstrs %t.mir -o - | FileCheck %s
; Test that custom pseudo source values can be round trip serialized through MIR. ; Test that custom pseudo source values can be round trip serialized through MIR.
; CHECK-LABEL: {{^}}name: shader ; CHECK-LABEL: {{^}}name: shader
; CHECK: %[[#]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET killed %17, %18, 4, 0, 0, implicit $exec :: (dereferenceable load (s32) from unknown-address + 4, align 1, addrspace 7) ; CHECK: %[[#]]:vgpr_32 = BUFFER_LOAD_DWORD_OFFSET killed %17, %18, 4, 0, 0, implicit $exec :: (dereferenceable load (s32) from unknown-address + 4, align 1, addrspace 8)
; CHECK: IMAGE_STORE_V4_V3_nsa_gfx10 killed %[[#]], %[[#]], %[[#]], %[[#]], killed %[[#]], 15, 2, -1, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), addrspace 7) ; CHECK: IMAGE_STORE_V4_V3_nsa_gfx10 killed %[[#]], %[[#]], %[[#]], %[[#]], killed %[[#]], 15, 2, -1, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable store (s128), addrspace 8)
; CHECK: DS_GWS_BARRIER %[[#]], 63, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource") ; CHECK: DS_GWS_BARRIER %[[#]], 63, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource")
define amdgpu_cs void @shader(i32 %arg0, i32 %arg1, <8 x i32> inreg %arg2, <4 x i32> inreg %arg3) { define amdgpu_cs void @shader(i32 %arg0, i32 %arg1, <8 x i32> inreg %arg2, <4 x i32> inreg %arg3) {
%bload0 = call i32 @llvm.amdgcn.raw.buffer.load.i32(<4 x i32> %arg3, i32 4, i32 0, i32 0) %bload0 = call i32 @llvm.amdgcn.raw.buffer.load.i32(<4 x i32> %arg3, i32 4, i32 0, i32 0)
%bload1 = call i32 @llvm.amdgcn.raw.buffer.load.i32(<4 x i32> %arg3, i32 8, i32 0, i32 0) %bload1 = call i32 @llvm.amdgcn.raw.buffer.load.i32(<4 x i32> %arg3, i32 8, i32 0, i32 0)
%bload2 = call i32 @llvm.amdgcn.raw.buffer.load.i32(<4 x i32> %arg3, i32 12, i32 0, i32 0) %bload2 = call i32 @llvm.amdgcn.raw.buffer.load.i32(<4 x i32> %arg3, i32 12, i32 0, i32 0)
%bload3 = call i32 @llvm.amdgcn.raw.buffer.load.i32(<4 x i32> %arg3, i32 16, i32 0, i32 0) %bload3 = call i32 @llvm.amdgcn.raw.buffer.load.i32(<4 x i32> %arg3, i32 16, i32 0, i32 0)
%bload0.f = bitcast i32 %bload0 to float %bload0.f = bitcast i32 %bload0 to float
%bload1.f = bitcast i32 %bload1 to float %bload1.f = bitcast i32 %bload1 to float
Show All 18 Lines

llvm/test/Instrumentation/AddressSanitizer/AMDGPU/adaptive_constant_global_redzones.ll

; RUN: opt < %s -passes=asan -S | FileCheck %s ; RUN: opt < %s -passes=asan -S | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
; Here we check that the global redzone sizes grow with the object size ; Here we check that the global redzone sizes grow with the object size
; for objects in constant address space. ; for objects in constant address space.
@G10 = addrspace(4) global [10 x i8] zeroinitializer, align 1 @G10 = addrspace(4) global [10 x i8] zeroinitializer, align 1
; CHECK: @G10 = addrspace(4) global { [10 x i8], [22 x i8] } ; CHECK: @G10 = addrspace(4) global { [10 x i8], [22 x i8] }
▲ Show 20 Lines • Show All 48 Lines • Show Last 20 Lines

llvm/test/Instrumentation/AddressSanitizer/AMDGPU/adaptive_global_redzones.ll

; RUN: opt < %s -passes=asan -S | FileCheck %s ; RUN: opt < %s -passes=asan -S | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
; Here we check that the global redzone sizes grow with the object size ; Here we check that the global redzone sizes grow with the object size
; for objects in global address space. ; for objects in global address space.
@G10 = addrspace(1) global [10 x i8] zeroinitializer, align 1 @G10 = addrspace(1) global [10 x i8] zeroinitializer, align 1
; CHECK: @G10 = addrspace(1) global { [10 x i8], [22 x i8] } ; CHECK: @G10 = addrspace(1) global { [10 x i8], [22 x i8] }
▲ Show 20 Lines • Show All 48 Lines • Show Last 20 Lines

llvm/test/Instrumentation/AddressSanitizer/AMDGPU/asan_do_not_instrument_lds.ll

; RUN: opt < %s -passes=asan -S | FileCheck %s ; RUN: opt < %s -passes=asan -S | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
; Memory access to lds are not instrumented ; Memory access to lds are not instrumented
@count = addrspace(3) global [100 x i32] undef, align 16 @count = addrspace(3) global [100 x i32] undef, align 16
define protected amdgpu_kernel void @lds_store(i32 %i) sanitize_address { define protected amdgpu_kernel void @lds_store(i32 %i) sanitize_address {
entry: entry:
Show All 17 Lines

llvm/test/Instrumentation/AddressSanitizer/AMDGPU/asan_do_not_instrument_scratch.ll

; RUN: opt < %s -passes=asan -S | FileCheck %s ; RUN: opt < %s -passes=asan -S | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
; Memory access to scratch are not instrumented ; Memory access to scratch are not instrumented
define protected amdgpu_kernel void @scratch_store(i32 %i) sanitize_address { define protected amdgpu_kernel void @scratch_store(i32 %i) sanitize_address {
entry: entry:
; CHECK-NOT: call * __asan_report ; CHECK-NOT: call * __asan_report
%c = alloca i32, align 4, addrspace(5) %c = alloca i32, align 4, addrspace(5)
Show All 11 Lines

llvm/test/Instrumentation/AddressSanitizer/AMDGPU/asan_instrument_constant_address_space.ll

; RUN: opt < %s -passes=asan -S | FileCheck %s ; RUN: opt < %s -passes=asan -S | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
@x = addrspace(4) global [2 x i32] zeroinitializer, align 4 @x = addrspace(4) global [2 x i32] zeroinitializer, align 4
define protected amdgpu_kernel void @constant_load(i64 %i) sanitize_address { define protected amdgpu_kernel void @constant_load(i64 %i) sanitize_address {
entry: entry:
; CHECK-LABEL: @constant_load ; CHECK-LABEL: @constant_load
; CHECK-NOT: load ; CHECK-NOT: load
Show All 26 Lines

llvm/test/Instrumentation/AddressSanitizer/AMDGPU/asan_instrument_generic_address_space.ll

; RUN: opt < %s -passes=asan -S | FileCheck %s ; RUN: opt < %s -passes=asan -S | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
define protected amdgpu_kernel void @generic_store(ptr addrspace(1) %p, i32 %i) sanitize_address { define protected amdgpu_kernel void @generic_store(ptr addrspace(1) %p, i32 %i) sanitize_address {
entry: entry:
; CHECK-LABEL: @generic_store ; CHECK-LABEL: @generic_store
; CHECK-NOT: store ; CHECK-NOT: store
; CHECK: %[[GENERIC_ADDR:[^ ]*]] = addrspacecast ptr addrspace(1) %p to ptr ; CHECK: %[[GENERIC_ADDR:[^ ]*]] = addrspacecast ptr addrspace(1) %p to ptr
; CHECK: call i1 @llvm.amdgcn.is.shared(ptr %[[GENERIC_ADDR]]) ; CHECK: call i1 @llvm.amdgcn.is.shared(ptr %[[GENERIC_ADDR]])
▲ Show 20 Lines • Show All 66 Lines • Show Last 20 Lines

llvm/test/Instrumentation/AddressSanitizer/AMDGPU/asan_instrument_global_address_space.ll

; RUN: opt < %s -passes=asan -S | FileCheck %s ; RUN: opt < %s -passes=asan -S | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
define protected amdgpu_kernel void @global_store(ptr addrspace(1) %p, i32 %i) sanitize_address { define protected amdgpu_kernel void @global_store(ptr addrspace(1) %p, i32 %i) sanitize_address {
entry: entry:
; CHECK-LABEL: @global_store ; CHECK-LABEL: @global_store
; CHECK-NOT: store ; CHECK-NOT: store
; ;
; CHECK: %[[STORE_ADDR:[^ ]*]] = ptrtoint ptr addrspace(1) %p to i64 ; CHECK: %[[STORE_ADDR:[^ ]*]] = ptrtoint ptr addrspace(1) %p to i64
▲ Show 20 Lines • Show All 53 Lines • Show Last 20 Lines

llvm/test/Instrumentation/AddressSanitizer/AMDGPU/global_metadata_addrspacecasts.ll

; RUN: opt < %s -passes=asan -S | FileCheck %s ; RUN: opt < %s -passes=asan -S | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
@g = addrspace(1) global [1 x i32] zeroinitializer, align 4 @g = addrspace(1) global [1 x i32] zeroinitializer, align 4
;CHECK: llvm.asan.globals ;CHECK: llvm.asan.globals
!llvm.asan.globals = !{!0, !1} !llvm.asan.globals = !{!0, !1}
!0 = !{ptr addrspace(1) @g, null, !"name", i1 false, i1 false} !0 = !{ptr addrspace(1) @g, null, !"name", i1 false, i1 false}
!1 = !{ptr addrspacecast (ptr addrspace(1) @g to ptr), null, !"name", i1 false, i1 false} !1 = !{ptr addrspacecast (ptr addrspace(1) @g to ptr), null, !"name", i1 false, i1 false}

llvm/test/Instrumentation/AddressSanitizer/AMDGPU/no_redzones_in_lds_globals.ll

; RUN: opt < %s -passes=asan -S | FileCheck %s ; RUN: opt < %s -passes=asan -S | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
@G10 = addrspace(3) global [10 x i8] zeroinitializer, align 1 @G10 = addrspace(3) global [10 x i8] zeroinitializer, align 1
; CHECK-NOT: @G10 = addrspace(3) global { [10 x i8], [* x i8] } ; CHECK-NOT: @G10 = addrspace(3) global { [10 x i8], [* x i8] }

llvm/test/Instrumentation/AddressSanitizer/AMDGPU/no_redzones_in_scratch_globals.ll

; RUN: opt < %s -passes=asan -S | FileCheck %s ; RUN: opt < %s -passes=asan -S | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
@G10 = addrspace(5) global [10 x i8] zeroinitializer, align 1 @G10 = addrspace(5) global [10 x i8] zeroinitializer, align 1
; CHECK-NOT: @G10 = addrspace(5) global { [10 x i8], [* x i8] } ; CHECK-NOT: @G10 = addrspace(5) global { [10 x i8], [* x i8] }

llvm/test/Transforms/AlignmentFromAssumptions/amdgpu-crash.ll

; Test that we don't crash. ; Test that we don't crash.
; RUN: opt < %s -passes=alignment-from-assumptions -S ; RUN: opt < %s -passes=alignment-from-assumptions -S
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7"
%"core::str::CharIndices.29.66.90.114.138.149.165.173.181.197.205.213.229.387.398" = type { [0 x i64], i64, [0 x i64], { ptr, ptr }, [0 x i64] } %"core::str::CharIndices.29.66.90.114.138.149.165.173.181.197.205.213.229.387.398" = type { [0 x i64], i64, [0 x i64], { ptr, ptr }, [0 x i64] }
%"unwind::libunwind::_Unwind_Exception.9.51.75.99.123.147.163.171.179.195.203.211.227.385.396" = type { [0 x i64], i64, [0 x i64], ptr, [0 x i64], [6 x i64], [0 x i64] } %"unwind::libunwind::_Unwind_Exception.9.51.75.99.123.147.163.171.179.195.203.211.227.385.396" = type { [0 x i64], i64, [0 x i64], ptr, [0 x i64], [6 x i64], [0 x i64] }
%"unwind::libunwind::_Unwind_Context.10.52.76.100.124.148.164.172.180.196.204.212.228.386.397" = type { [0 x i8] } %"unwind::libunwind::_Unwind_Context.10.52.76.100.124.148.164.172.180.196.204.212.228.386.397" = type { [0 x i8] }
define void @"_ZN44_$LT$$RF$T$u20$as$u20$core..fmt..Display$GT$3fmt17h7b1d039c7ff5e1feE"() { define void @"_ZN44_$LT$$RF$T$u20$as$u20$core..fmt..Display$GT$3fmt17h7b1d039c7ff5e1feE"() {
start: start:
%_15.i.i = alloca %"core::str::CharIndices.29.66.90.114.138.149.165.173.181.197.205.213.229.387.398", align 8, addrspace(5) %_15.i.i = alloca %"core::str::CharIndices.29.66.90.114.138.149.165.173.181.197.205.213.229.387.398", align 8, addrspace(5)
Show All 19 Lines

llvm/test/Transforms/AtomicExpand/AMDGPU/expand-atomic-i16.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -mtriple=amdgcn-amd-amdhsa -S -atomic-expand %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -S -atomic-expand %s | FileCheck %s
; RUN: opt -mtriple=r600-mesa-mesa3d -S -atomic-expand %s | FileCheck %s ; RUN: opt -mtriple=r600-mesa-mesa3d -S -atomic-expand %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
define i16 @test_atomicrmw_xchg_i16_global(ptr addrspace(1) %ptr, i16 %value) { define i16 @test_atomicrmw_xchg_i16_global(ptr addrspace(1) %ptr, i16 %value) {
; CHECK-LABEL: @test_atomicrmw_xchg_i16_global( ; CHECK-LABEL: @test_atomicrmw_xchg_i16_global(
; CHECK-NEXT: [[ALIGNEDADDR:%.*]] = call ptr addrspace(1) @llvm.ptrmask.p1.i64(ptr addrspace(1) [[PTR:%.*]], i64 -4) ; CHECK-NEXT: [[ALIGNEDADDR:%.*]] = call ptr addrspace(1) @llvm.ptrmask.p1.i64(ptr addrspace(1) [[PTR:%.*]], i64 -4)
; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr addrspace(1) [[PTR]] to i64 ; CHECK-NEXT: [[TMP1:%.*]] = ptrtoint ptr addrspace(1) [[PTR]] to i64
; CHECK-NEXT: [[PTRLSB:%.*]] = and i64 [[TMP1]], 3 ; CHECK-NEXT: [[PTRLSB:%.*]] = and i64 [[TMP1]], 3
; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[PTRLSB]], 3 ; CHECK-NEXT: [[TMP2:%.*]] = shl i64 [[PTRLSB]], 3
; CHECK-NEXT: [[SHIFTAMT:%.*]] = trunc i64 [[TMP2]] to i32 ; CHECK-NEXT: [[SHIFTAMT:%.*]] = trunc i64 [[TMP2]] to i32
▲ Show 20 Lines • Show All 888 Lines • Show Last 20 Lines

llvm/test/Transforms/EarlyCSE/AMDGPU/memrealtime.ll

; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes='early-cse<memssa>' -earlycse-debug-hash < %s | FileCheck %s ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes='early-cse<memssa>' -earlycse-debug-hash < %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; CHECK-LABEL: @memrealtime( ; CHECK-LABEL: @memrealtime(
; CHECK: call i64 @llvm.amdgcn.s.memrealtime() ; CHECK: call i64 @llvm.amdgcn.s.memrealtime()
; CHECK: call i64 @llvm.amdgcn.s.memrealtime() ; CHECK: call i64 @llvm.amdgcn.s.memrealtime()
define amdgpu_kernel void @memrealtime(i64 %cycles) #0 { define amdgpu_kernel void @memrealtime(i64 %cycles) #0 {
entry: entry:
%0 = tail call i64 @llvm.amdgcn.s.memrealtime() %0 = tail call i64 @llvm.amdgcn.s.memrealtime()
%cmp3 = icmp sgt i64 %cycles, 0 %cmp3 = icmp sgt i64 %cycles, 0
Show All 33 Lines

llvm/test/Transforms/IndVarSimplify/AMDGPU/no-widen-to-i64.ll

; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -passes=indvars %s | FileCheck %s ; RUN: opt -S -mtriple=amdgcn-unknown-amdhsa -passes=indvars %s | FileCheck %s
; Bug 21148 ; Bug 21148
; Induction variables should not be widened for 64-bit integers, ; Induction variables should not be widened for 64-bit integers,
; despite being a legal type. ; despite being a legal type.
; ;
; The cost of basic arithmetic instructions on a 64-bit integer are ; The cost of basic arithmetic instructions on a 64-bit integer are
; twice as expensive as that on a 32-bit integer, or split into 2 ; twice as expensive as that on a 32-bit integer, or split into 2
; 32-bit components. ; 32-bit components.
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; CHECK-LABEL: @indvar_32_bit( ; CHECK-LABEL: @indvar_32_bit(
; CHECK-NOT: sext i32 ; CHECK-NOT: sext i32
; CHECK: phi i32 ; CHECK: phi i32
define amdgpu_kernel void @indvar_32_bit(i32 %n, ptr nocapture %output) { define amdgpu_kernel void @indvar_32_bit(i32 %n, ptr nocapture %output) {
entry: entry:
%cmp5 = icmp sgt i32 %n, 0 %cmp5 = icmp sgt i32 %n, 0
br i1 %cmp5, label %for.body.preheader, label %for.end br i1 %cmp5, label %for.body.preheader, label %for.end
▲ Show 20 Lines • Show All 78 Lines • Show Last 20 Lines

llvm/test/Transforms/InferAddressSpaces/AMDGPU/noop-ptrint-pair.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -S -o - -passes=infer-address-spaces %s | FileCheck -check-prefixes=COMMON,AMDGCN %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -S -o - -passes=infer-address-spaces %s | FileCheck -check-prefixes=COMMON,AMDGCN %s
; RUN: opt -S -o - -passes=infer-address-spaces -assume-default-is-flat-addrspace %s | FileCheck -check-prefixes=COMMON,NOTTI %s ; RUN: opt -S -o - -passes=infer-address-spaces -assume-default-is-flat-addrspace %s | FileCheck -check-prefixes=COMMON,NOTTI %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7"
; COMMON-LABEL: @noop_ptrint_pair( ; COMMON-LABEL: @noop_ptrint_pair(
; AMDGCN-NEXT: store i32 0, ptr addrspace(1) %{{.*}} ; AMDGCN-NEXT: store i32 0, ptr addrspace(1) %{{.*}}
; AMDGCN-NEXT: ret void ; AMDGCN-NEXT: ret void
; NOTTI-NEXT: %1 = ptrtoint ptr addrspace(1) %x.coerce to i64 ; NOTTI-NEXT: %1 = ptrtoint ptr addrspace(1) %x.coerce to i64
; NOTTI-NEXT: %2 = inttoptr i64 %1 to ptr ; NOTTI-NEXT: %2 = inttoptr i64 %1 to ptr
; NOTTI-NEXT: store i32 0, ptr %2 ; NOTTI-NEXT: store i32 0, ptr %2
; NOTTI-NEXT: ret void ; NOTTI-NEXT: ret void
▲ Show 20 Lines • Show All 99 Lines • Show Last 20 Lines

llvm/test/Transforms/InferAddressSpaces/AMDGPU/ptrmask.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=infer-address-spaces,instsimplify %s | FileCheck %s ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=infer-address-spaces,instsimplify %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
define i8 @ptrmask_cast_local_to_flat(ptr addrspace(3) %src.ptr, i64 %mask) { define i8 @ptrmask_cast_local_to_flat(ptr addrspace(3) %src.ptr, i64 %mask) {
; CHECK-LABEL: @ptrmask_cast_local_to_flat( ; CHECK-LABEL: @ptrmask_cast_local_to_flat(
; CHECK-NEXT: [[CAST:%.*]] = addrspacecast ptr addrspace(3) [[SRC_PTR:%.*]] to ptr ; CHECK-NEXT: [[CAST:%.*]] = addrspacecast ptr addrspace(3) [[SRC_PTR:%.*]] to ptr
; CHECK-NEXT: [[MASKED:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[CAST]], i64 [[MASK:%.*]]) ; CHECK-NEXT: [[MASKED:%.*]] = call ptr @llvm.ptrmask.p0.i64(ptr [[CAST]], i64 [[MASK:%.*]])
; CHECK-NEXT: [[LOAD:%.*]] = load i8, ptr [[MASKED]], align 1 ; CHECK-NEXT: [[LOAD:%.*]] = load i8, ptr [[MASKED]], align 1
; CHECK-NEXT: ret i8 [[LOAD]] ; CHECK-NEXT: ret i8 [[LOAD]]
; ;
▲ Show 20 Lines • Show All 360 Lines • Show Last 20 Lines

llvm/test/Transforms/InferAddressSpaces/X86/noop-ptrint-pair.ll

; RUN: opt -mtriple=x86_64-unknown-unknown -S -o - -passes=infer-address-spaces -assume-default-is-flat-addrspace %s | FileCheck %s ; RUN: opt -mtriple=x86_64-unknown-unknown -S -o - -passes=infer-address-spaces -assume-default-is-flat-addrspace %s | FileCheck %s
; Check that assert in X86TargetMachine::isNoopAddrSpaceCast is not triggered. ; Check that assert in X86TargetMachine::isNoopAddrSpaceCast is not triggered.
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7"
; CHECK-LABEL: @noop_ptrint_pair( ; CHECK-LABEL: @noop_ptrint_pair(
; CHECK: addrspacecast ptr addrspace(1) %x to ptr addrspace(4) ; CHECK: addrspacecast ptr addrspace(1) %x to ptr addrspace(4)
; CHECK-NEXT: ptrtoint ptr addrspace(4) %{{.*}} to i64 ; CHECK-NEXT: ptrtoint ptr addrspace(4) %{{.*}} to i64
; CHECK-NEXT: inttoptr i64 %{{.*}} to ptr addrspace(4) ; CHECK-NEXT: inttoptr i64 %{{.*}} to ptr addrspace(4)
define void @noop_ptrint_pair(ptr addrspace(1) %x) { define void @noop_ptrint_pair(ptr addrspace(1) %x) {
%1 = addrspacecast ptr addrspace(1) %x to ptr addrspace(4) %1 = addrspacecast ptr addrspace(1) %x to ptr addrspace(4)
%2 = ptrtoint ptr addrspace(4) %1 to i64 %2 = ptrtoint ptr addrspace(4) %1 to i64
%3 = inttoptr i64 %2 to ptr addrspace(4) %3 = inttoptr i64 %2 to ptr addrspace(4)
ret void ret void
} }

llvm/test/Transforms/Inline/AMDGPU/amdgpu-inline-alloca-argument.ll

; RUN: opt -mtriple=amdgcn--amdhsa -S -passes=inline -inline-threshold=0 < %s | FileCheck %s ; RUN: opt -mtriple=amdgcn--amdhsa -S -passes=inline -inline-threshold=0 < %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
define void @use_flat_ptr_arg(ptr nocapture %p) { define void @use_flat_ptr_arg(ptr nocapture %p) {
entry: entry:
%tmp1 = load float, ptr %p, align 4 %tmp1 = load float, ptr %p, align 4
%div = fdiv float 1.000000e+00, %tmp1 %div = fdiv float 1.000000e+00, %tmp1
%add0 = fadd float %div, 1.0 %add0 = fadd float %div, 1.0
%add1 = fadd float %add0, 1.0 %add1 = fadd float %add0, 1.0
%add2 = fadd float %add1, 1.0 %add2 = fadd float %add1, 1.0
▲ Show 20 Lines • Show All 58 Lines • Show Last 20 Lines

llvm/test/Transforms/InstCombine/AMDGPU/memcpy-from-constant.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -mtriple=amdgcn-amd-amdhsa -S -amdgpu-aa-wrapper -amdgpu-aa -instcombine -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -S -amdgpu-aa-wrapper -amdgpu-aa -instcombine -o - %s | FileCheck %s
; Make sure the optimization from memcpy-from-global.ll happens, but ; Make sure the optimization from memcpy-from-global.ll happens, but
; the constant source is not a global variable. ; the constant source is not a global variable.
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; Simple memcpy to alloca from constant address space argument. ; Simple memcpy to alloca from constant address space argument.
define i8 @memcpy_constant_arg_ptr_to_alloca(ptr addrspace(4) noalias readonly align 4 dereferenceable(32) %arg, i32 %idx) { define i8 @memcpy_constant_arg_ptr_to_alloca(ptr addrspace(4) noalias readonly align 4 dereferenceable(32) %arg, i32 %idx) {
; CHECK-LABEL: @memcpy_constant_arg_ptr_to_alloca( ; CHECK-LABEL: @memcpy_constant_arg_ptr_to_alloca(
; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[IDX:%.*]] to i64 ; CHECK-NEXT: [[TMP1:%.*]] = sext i32 [[IDX:%.*]] to i64
; CHECK-NEXT: [[GEP:%.*]] = getelementptr [32 x i8], ptr addrspace(4) [[ARG:%.*]], i64 0, i64 [[TMP1]] ; CHECK-NEXT: [[GEP:%.*]] = getelementptr [32 x i8], ptr addrspace(4) [[ARG:%.*]], i64 0, i64 [[TMP1]]
; CHECK-NEXT: [[LOAD:%.*]] = load i8, ptr addrspace(4) [[GEP]], align 1 ; CHECK-NEXT: [[LOAD:%.*]] = load i8, ptr addrspace(4) [[GEP]], align 1
; CHECK-NEXT: ret i8 [[LOAD]] ; CHECK-NEXT: ret i8 [[LOAD]]
▲ Show 20 Lines • Show All 231 Lines • Show Last 20 Lines

llvm/test/Transforms/InstCombine/alloca-in-non-alloca-as.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -passes=instcombine -S | FileCheck %s ; RUN: opt < %s -passes=instcombine -S | FileCheck %s
; Gracefully handle the alloca that is not in the alloca AS (=5) ; Gracefully handle the alloca that is not in the alloca AS (=5)
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
declare void @use(ptr, ptr) declare void @use(ptr, ptr)
define weak amdgpu_kernel void @__omp_offloading_802_ea0109_main_l8(ptr %a) { define weak amdgpu_kernel void @__omp_offloading_802_ea0109_main_l8(ptr %a) {
; CHECK-LABEL: @__omp_offloading_802_ea0109_main_l8( ; CHECK-LABEL: @__omp_offloading_802_ea0109_main_l8(
; CHECK-NEXT: .master: ; CHECK-NEXT: .master:
; CHECK-NEXT: [[TMP0:%.*]] = alloca [8 x i8], align 1 ; CHECK-NEXT: [[TMP0:%.*]] = alloca [8 x i8], align 1
Show All 26 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/aa-metadata.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa,scoped-noalias-aa -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefix=SCOPE %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa,scoped-noalias-aa -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefix=SCOPE %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefix=NOSCOPE %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefix=NOSCOPE %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; This fails to vectorize if the !alias.scope is not used ; This fails to vectorize if the !alias.scope is not used
define amdgpu_kernel void @vectorize_alias_scope(ptr addrspace(1) nocapture %a, ptr addrspace(1) nocapture %b, ptr addrspace(1) nocapture readonly %c) #0 { define amdgpu_kernel void @vectorize_alias_scope(ptr addrspace(1) nocapture %a, ptr addrspace(1) nocapture %b, ptr addrspace(1) nocapture readonly %c) #0 {
; SCOPE-LABEL: @vectorize_alias_scope( ; SCOPE-LABEL: @vectorize_alias_scope(
; SCOPE-NEXT: entry: ; SCOPE-NEXT: entry:
; SCOPE-NEXT: [[LD_C:%.*]] = load float, ptr addrspace(1) [[C:%.*]], align 4, !alias.scope !0 ; SCOPE-NEXT: [[LD_C:%.*]] = load float, ptr addrspace(1) [[C:%.*]], align 4, !alias.scope !0
; SCOPE-NEXT: store <2 x float> zeroinitializer, ptr addrspace(1) [[A:%.*]], align 4, !noalias !0 ; SCOPE-NEXT: store <2 x float> zeroinitializer, ptr addrspace(1) [[A:%.*]], align 4, !noalias !0
Show All 26 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/adjust-alloca-alignment.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -S -passes=load-store-vectorizer --mcpu=hawaii -mattr=-unaligned-access-mode,+max-private-element-size-16 < %s | FileCheck -check-prefixes=CHECK,ALIGNED %s ; RUN: opt -S -passes=load-store-vectorizer --mcpu=hawaii -mattr=-unaligned-access-mode,+max-private-element-size-16 < %s | FileCheck -check-prefixes=CHECK,ALIGNED %s
; RUN: opt -S -passes=load-store-vectorizer --mcpu=hawaii -mattr=+unaligned-access-mode,+unaligned-scratch-access,+max-private-element-size-16 < %s | FileCheck -check-prefixes=CHECK,UNALIGNED %s ; RUN: opt -S -passes=load-store-vectorizer --mcpu=hawaii -mattr=+unaligned-access-mode,+unaligned-scratch-access,+max-private-element-size-16 < %s | FileCheck -check-prefixes=CHECK,UNALIGNED %s
; RUN: opt -S -passes='function(load-store-vectorizer)' --mcpu=hawaii -mattr=-unaligned-access-mode,+max-private-element-size-16 < %s | FileCheck -check-prefixes=CHECK,ALIGNED %s ; RUN: opt -S -passes='function(load-store-vectorizer)' --mcpu=hawaii -mattr=-unaligned-access-mode,+max-private-element-size-16 < %s | FileCheck -check-prefixes=CHECK,ALIGNED %s
; RUN: opt -S -passes='function(load-store-vectorizer)' --mcpu=hawaii -mattr=+unaligned-access-mode,+unaligned-scratch-access,+max-private-element-size-16 < %s | FileCheck -check-prefixes=CHECK,UNALIGNED %s ; RUN: opt -S -passes='function(load-store-vectorizer)' --mcpu=hawaii -mattr=+unaligned-access-mode,+unaligned-scratch-access,+max-private-element-size-16 < %s | FileCheck -check-prefixes=CHECK,UNALIGNED %s
target triple = "amdgcn--" target triple = "amdgcn--"
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
define amdgpu_kernel void @load_unknown_offset_align1_i8(ptr addrspace(1) noalias %out, i32 %offset) #0 { define amdgpu_kernel void @load_unknown_offset_align1_i8(ptr addrspace(1) noalias %out, i32 %offset) #0 {
; ALIGNED-LABEL: @load_unknown_offset_align1_i8( ; ALIGNED-LABEL: @load_unknown_offset_align1_i8(
; ALIGNED-NEXT: [[ALLOCA:%.*]] = alloca [128 x i8], align 1, addrspace(5) ; ALIGNED-NEXT: [[ALLOCA:%.*]] = alloca [128 x i8], align 1, addrspace(5)
; ALIGNED-NEXT: [[PTR0:%.*]] = getelementptr inbounds [128 x i8], ptr addrspace(5) [[ALLOCA]], i32 0, i32 [[OFFSET:%.*]] ; ALIGNED-NEXT: [[PTR0:%.*]] = getelementptr inbounds [128 x i8], ptr addrspace(5) [[ALLOCA]], i32 0, i32 [[OFFSET:%.*]]
; ALIGNED-NEXT: [[VAL0:%.*]] = load i8, ptr addrspace(5) [[PTR0]], align 1 ; ALIGNED-NEXT: [[VAL0:%.*]] = load i8, ptr addrspace(5) [[PTR0]], align 1
; ALIGNED-NEXT: [[PTR1:%.*]] = getelementptr inbounds i8, ptr addrspace(5) [[PTR0]], i32 1 ; ALIGNED-NEXT: [[PTR1:%.*]] = getelementptr inbounds i8, ptr addrspace(5) [[PTR0]], i32 1
; ALIGNED-NEXT: [[VAL1:%.*]] = load i8, ptr addrspace(5) [[PTR1]], align 1 ; ALIGNED-NEXT: [[VAL1:%.*]] = load i8, ptr addrspace(5) [[PTR1]], align 1
▲ Show 20 Lines • Show All 346 Lines • Show Last 20 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/complex-index.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
declare i64 @_Z12get_local_idj(i32) declare i64 @_Z12get_local_idj(i32)
declare i64 @_Z12get_group_idj(i32) declare i64 @_Z12get_group_idj(i32)
declare double @llvm.fmuladd.f64(double, double, double) declare double @llvm.fmuladd.f64(double, double, double)
; CHECK-LABEL: @factorizedVsNonfactorizedAccess( ; CHECK-LABEL: @factorizedVsNonfactorizedAccess(
Show All 40 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/extended-index.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
declare i32 @llvm.amdgcn.workitem.id.x() #1 declare i32 @llvm.amdgcn.workitem.id.x() #1
; CHECK-LABEL: @basic_merge_sext_index( ; CHECK-LABEL: @basic_merge_sext_index(
; CHECK: sext i32 %id.x to i64 ; CHECK: sext i32 %id.x to i64
; CHECK: load <2 x float> ; CHECK: load <2 x float>
; CHECK: store <2 x float> zeroinitializer ; CHECK: store <2 x float> zeroinitializer
define amdgpu_kernel void @basic_merge_sext_index(ptr addrspace(1) nocapture %a, ptr addrspace(1) nocapture %b, ptr addrspace(1) nocapture readonly %c) #0 { define amdgpu_kernel void @basic_merge_sext_index(ptr addrspace(1) nocapture %a, ptr addrspace(1) nocapture %b, ptr addrspace(1) nocapture readonly %c) #0 {
▲ Show 20 Lines • Show All 139 Lines • Show Last 20 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/gep-bitcast.ll

; RUN: opt -S -mtriple=amdgcn--amdhsa -passes=load-store-vectorizer < %s | FileCheck %s ; RUN: opt -S -mtriple=amdgcn--amdhsa -passes=load-store-vectorizer < %s | FileCheck %s
; RUN: opt -S -mtriple=amdgcn--amdhsa -passes='function(load-store-vectorizer)' < %s | FileCheck %s ; RUN: opt -S -mtriple=amdgcn--amdhsa -passes='function(load-store-vectorizer)' < %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; Check that vectorizer can find a GEP through bitcast ; Check that vectorizer can find a GEP through bitcast
; CHECK-LABEL: @vect_zext_bitcast_f32_to_i32_idx ; CHECK-LABEL: @vect_zext_bitcast_f32_to_i32_idx
; CHECK: load <4 x i32> ; CHECK: load <4 x i32>
define void @vect_zext_bitcast_f32_to_i32_idx(ptr addrspace(1) %arg1, i32 %base) { define void @vect_zext_bitcast_f32_to_i32_idx(ptr addrspace(1) %arg1, i32 %base) {
%add1 = add nuw i32 %base, 0 %add1 = add nuw i32 %base, 0
%zext1 = zext i32 %add1 to i64 %zext1 = zext i32 %add1 to i64
%gep1 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 %zext1 %gep1 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 %zext1
▲ Show 20 Lines • Show All 108 Lines • Show Last 20 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/insertion-point.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; Check position of the inserted vector load/store. Vectorized loads should be ; Check position of the inserted vector load/store. Vectorized loads should be
; inserted at the position of the first load in the chain, and stores should be ; inserted at the position of the first load in the chain, and stores should be
; inserted at the position of the last store. ; inserted at the position of the last store.
; CHECK-LABEL: @insert_load_point( ; CHECK-LABEL: @insert_load_point(
; CHECK: %z = add i32 %x, 4 ; CHECK: %z = add i32 %x, 4
; CHECK: load <2 x float> ; CHECK: load <2 x float>
▲ Show 20 Lines • Show All 106 Lines • Show Last 20 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/interleaved-mayalias-store.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; This is NOT OK to vectorize, as either load may alias either store. ; This is NOT OK to vectorize, as either load may alias either store.
; CHECK: load double ; CHECK: load double
; CHECK: store double 0.000000e+00, ptr addrspace(1) %a, ; CHECK: store double 0.000000e+00, ptr addrspace(1) %a,
; CHECK: load double ; CHECK: load double
; CHECK: store double 0.000000e+00, ptr addrspace(1) %a.idx.1 ; CHECK: store double 0.000000e+00, ptr addrspace(1) %a.idx.1
define amdgpu_kernel void @interleave(ptr addrspace(1) nocapture %a, ptr addrspace(1) nocapture %b, ptr addrspace(1) nocapture readonly %c) #0 { define amdgpu_kernel void @interleave(ptr addrspace(1) nocapture %a, ptr addrspace(1) nocapture %b, ptr addrspace(1) nocapture readonly %c) #0 {
Show All 17 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/invariant-load.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; CHECK-LABEL: @interleave ; CHECK-LABEL: @interleave
; CHECK: load <2 x double>, ptr addrspace(1) %{{.}}, align 8{{$}} ; CHECK: load <2 x double>, ptr addrspace(1) %{{.}}, align 8{{$}}
; CHECK: store <2 x double> zeroinitializer ; CHECK: store <2 x double> zeroinitializer
; CHECK: store double %add ; CHECK: store double %add
define amdgpu_kernel void @interleave(ptr addrspace(1) nocapture %a, ptr addrspace(1) nocapture %b, ptr addrspace(1) nocapture readonly %c) #0 { define amdgpu_kernel void @interleave(ptr addrspace(1) nocapture %a, ptr addrspace(1) nocapture %b, ptr addrspace(1) nocapture readonly %c) #0 {
entry: entry:
%a.idx.1 = getelementptr inbounds double, ptr addrspace(1) %a, i64 1 %a.idx.1 = getelementptr inbounds double, ptr addrspace(1) %a, i64 1
Show All 17 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores-private.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-4,-unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT4,ELT4-ALIGNED,ALIGNED,ALL %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-4,-unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT4,ELT4-ALIGNED,ALIGNED,ALL %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-8,-unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT8,ALIGNED,ALL %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-8,-unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT8,ALIGNED,ALL %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-16,-unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT16,ALIGNED,ALL %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-16,-unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT16,ALIGNED,ALL %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-4,+unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT4,ELT4-UNALIGNED,UNALIGNED,ALL %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-4,+unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT4,ELT4-UNALIGNED,UNALIGNED,ALL %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-8,+unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT8,ELT8-UNALIGNED,UNALIGNED,ALL %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-8,+unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT8,ELT8-UNALIGNED,UNALIGNED,ALL %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-16,+unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT16,ELT16-UNALIGNED,UNALIGNED,ALL %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -mattr=+max-private-element-size-16,+unaligned-scratch-access -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=ELT16,ELT16-UNALIGNED,UNALIGNED,ALL %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v4i32 ; ALL-LABEL: @merge_private_store_4_vector_elts_loads_v4i32
; ELT4-ALIGNED: store i32 ; ELT4-ALIGNED: store i32
; ELT4-ALIGNED: store i32 ; ELT4-ALIGNED: store i32
; ELT4-ALIGNED: store i32 ; ELT4-ALIGNED: store i32
; ELT4-ALIGNED: store i32 ; ELT4-ALIGNED: store i32
; ELT8: store <2 x i32> ; ELT8: store <2 x i32>
▲ Show 20 Lines • Show All 195 Lines • Show Last 20 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/merge-stores.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa --mcpu=hawaii -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa --mcpu=hawaii -passes=load-store-vectorizer -S -o - %s | FileCheck %s
; Copy of test/CodeGen/AMDGPU/merge-stores.ll with some additions ; Copy of test/CodeGen/AMDGPU/merge-stores.ll with some additions
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; TODO: Vector element tests ; TODO: Vector element tests
; TODO: Non-zero base offset for load and store combinations ; TODO: Non-zero base offset for load and store combinations
; TODO: Same base addrspacecasted ; TODO: Same base addrspacecasted
define amdgpu_kernel void @merge_global_store_2_constants_i8(ptr addrspace(1) %out) #0 { define amdgpu_kernel void @merge_global_store_2_constants_i8(ptr addrspace(1) %out) #0 {
; CHECK-LABEL: @merge_global_store_2_constants_i8( ; CHECK-LABEL: @merge_global_store_2_constants_i8(
▲ Show 20 Lines • Show All 826 Lines • Show Last 20 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/missing-alignment.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -mtriple=amdgcn-- -mcpu=tonga -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-- -mcpu=tonga -passes=load-store-vectorizer -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
@lds = internal addrspace(3) global [512 x float] undef, align 4 @lds = internal addrspace(3) global [512 x float] undef, align 4
; The original load has an implicit alignment of 4, and should not ; The original load has an implicit alignment of 4, and should not
; increase to an align 8 load. ; increase to an align 8 load.
define amdgpu_kernel void @load_keep_base_alignment_missing_align(ptr addrspace(1) %out) { define amdgpu_kernel void @load_keep_base_alignment_missing_align(ptr addrspace(1) %out) {
; CHECK-LABEL: @load_keep_base_alignment_missing_align( ; CHECK-LABEL: @load_keep_base_alignment_missing_align(
Show All 30 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/multiple_tails.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -mcpu=hawaii -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=GCN,GFX7 %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -mcpu=hawaii -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=GCN,GFX7 %s
; RUN: opt -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=GCN,GFX9 %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -passes=load-store-vectorizer -S -o - %s | FileCheck -check-prefixes=GCN,GFX9 %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; Checks that there is no crash when there are multiple tails ; Checks that there is no crash when there are multiple tails
; for a the same head starting a chain. ; for a the same head starting a chain.
@0 = internal addrspace(3) global [16384 x i32] undef @0 = internal addrspace(3) global [16384 x i32] undef
; GCN-LABEL: @no_crash( ; GCN-LABEL: @no_crash(
; GCN: store <2 x i32> zeroinitializer ; GCN: store <2 x i32> zeroinitializer
; GCN: store i32 0 ; GCN: store i32 0
▲ Show 20 Lines • Show All 61 Lines • Show Last 20 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/no-implicit-float.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; CHECK-LABEL: @no_implicit_float( ; CHECK-LABEL: @no_implicit_float(
; CHECK: store i32 ; CHECK: store i32
; CHECK: store i32 ; CHECK: store i32
; CHECK: store i32 ; CHECK: store i32
; CHECK: store i32 ; CHECK: store i32
define amdgpu_kernel void @no_implicit_float(ptr addrspace(1) %out) #0 { define amdgpu_kernel void @no_implicit_float(ptr addrspace(1) %out) #0 {
%out.gep.1 = getelementptr i32, ptr addrspace(1) %out, i32 1 %out.gep.1 = getelementptr i32, ptr addrspace(1) %out, i32 1
Show All 11 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/optnone.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; CHECK-LABEL: @optnone( ; CHECK-LABEL: @optnone(
; CHECK: store i32 ; CHECK: store i32
; CHECK: store i32 ; CHECK: store i32
define amdgpu_kernel void @optnone(ptr addrspace(1) %out) noinline optnone { define amdgpu_kernel void @optnone(ptr addrspace(1) %out) noinline optnone {
%out.gep.1 = getelementptr i32, ptr addrspace(1) %out, i32 1 %out.gep.1 = getelementptr i32, ptr addrspace(1) %out, i32 1
store i32 123, ptr addrspace(1) %out.gep.1 store i32 123, ptr addrspace(1) %out.gep.1
Show All 13 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/pointer-elements.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
declare i32 @llvm.amdgcn.workitem.id.x() #1 declare i32 @llvm.amdgcn.workitem.id.x() #1
; CHECK-LABEL: @merge_v2p1i8( ; CHECK-LABEL: @merge_v2p1i8(
; CHECK: load <2 x i64> ; CHECK: load <2 x i64>
; CHECK: inttoptr i64 %{{[^ ]+}} to ptr addrspace(1) ; CHECK: inttoptr i64 %{{[^ ]+}} to ptr addrspace(1)
; CHECK: inttoptr i64 %{{[^ ]+}} to ptr addrspace(1) ; CHECK: inttoptr i64 %{{[^ ]+}} to ptr addrspace(1)
; CHECK: store <2 x i64> zeroinitializer ; CHECK: store <2 x i64> zeroinitializer
▲ Show 20 Lines • Show All 284 Lines • Show Last 20 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/selects-inseltpoison.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer,dce -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer,dce -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
define void @base_case(i1 %cnd, ptr addrspace(1) %a, ptr addrspace(1) %b, ptr addrspace(1) %out) { define void @base_case(i1 %cnd, ptr addrspace(1) %a, ptr addrspace(1) %b, ptr addrspace(1) %out) {
; CHECK-LABEL: @base_case ; CHECK-LABEL: @base_case
; CHECK: load <3 x i32> ; CHECK: load <3 x i32>
entry: entry:
%gep1 = getelementptr inbounds i32, ptr addrspace(1) %a, i64 1 %gep1 = getelementptr inbounds i32, ptr addrspace(1) %a, i64 1
%gep2 = getelementptr inbounds i32, ptr addrspace(1) %a, i64 2 %gep2 = getelementptr inbounds i32, ptr addrspace(1) %a, i64 2
%gep4 = getelementptr inbounds i32, ptr addrspace(1) %b, i64 1 %gep4 = getelementptr inbounds i32, ptr addrspace(1) %b, i64 1
▲ Show 20 Lines • Show All 79 Lines • Show Last 20 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/selects.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer,dce -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer,dce -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
define void @base_case(i1 %cnd, ptr addrspace(1) %a, ptr addrspace(1) %b, ptr addrspace(1) %out) { define void @base_case(i1 %cnd, ptr addrspace(1) %a, ptr addrspace(1) %b, ptr addrspace(1) %out) {
; CHECK-LABEL: @base_case ; CHECK-LABEL: @base_case
; CHECK: load <3 x i32> ; CHECK: load <3 x i32>
entry: entry:
%gep1 = getelementptr inbounds i32, ptr addrspace(1) %a, i64 1 %gep1 = getelementptr inbounds i32, ptr addrspace(1) %a, i64 1
%gep2 = getelementptr inbounds i32, ptr addrspace(1) %a, i64 2 %gep2 = getelementptr inbounds i32, ptr addrspace(1) %a, i64 2
%gep4 = getelementptr inbounds i32, ptr addrspace(1) %b, i64 1 %gep4 = getelementptr inbounds i32, ptr addrspace(1) %b, i64 1
▲ Show 20 Lines • Show All 79 Lines • Show Last 20 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/store_with_aliasing_load.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; Check that, in the presence of an aliasing load, the stores preceding the ; Check that, in the presence of an aliasing load, the stores preceding the
; aliasing load are safe to vectorize. ; aliasing load are safe to vectorize.
; CHECK-LABEL: store_vectorize_with_alias ; CHECK-LABEL: store_vectorize_with_alias
; CHECK: store <4 x float> ; CHECK: store <4 x float>
; CHECK: load <4 x float> ; CHECK: load <4 x float>
; CHECK: store <4 x float> ; CHECK: store <4 x float>
Show All 36 Lines

llvm/test/Transforms/LoadStoreVectorizer/AMDGPU/weird-type-accesses.ll

; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s ; RUN: opt -mtriple=amdgcn-amd-amdhsa -passes=load-store-vectorizer -S -o - %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; Checks that we don't merge loads/stores of types smaller than one ; Checks that we don't merge loads/stores of types smaller than one
; byte, or vectors with elements smaller than one byte. ; byte, or vectors with elements smaller than one byte.
%struct.foo = type { i32, i8 } %struct.foo = type { i32, i8 }
declare void @use_i1(i1) declare void @use_i1(i1)
declare void @use_i2(i2) declare void @use_i2(i2)
▲ Show 20 Lines • Show All 186 Lines • Show Last 20 Lines

llvm/test/Transforms/LoopLoadElim/pr46854-adress-spaces.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; REQUIRES: amdgpu-registered-target ; REQUIRES: amdgpu-registered-target
; RUN: opt -passes='require<globals-aa>,loop-simplify,loop-load-elim' -S %s | FileCheck %s ; RUN: opt -passes='require<globals-aa>,loop-simplify,loop-load-elim' -S %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
%struct.foo = type { %struct.pluto, i8, ptr, i32 } %struct.foo = type { %struct.pluto, i8, ptr, i32 }
%struct.pluto = type { i32, i32, i32, ptr, i32, i32, i32 } %struct.pluto = type { i32, i32, i32, ptr, i32, i32, i32 }
%struct.wombat = type { %struct.barney } %struct.wombat = type { %struct.barney }
%struct.barney = type { <2 x float> } %struct.barney = type { <2 x float> }
@global = external protected local_unnamed_addr addrspace(4) externally_initialized global %struct.foo, align 8 @global = external protected local_unnamed_addr addrspace(4) externally_initialized global %struct.foo, align 8
▲ Show 20 Lines • Show All 59 Lines • Show Last 20 Lines

llvm/test/Transforms/LoopStrengthReduce/AMDGPU/atomics.ll

; RUN: opt -S -mtriple=amdgcn-- -mcpu=bonaire -loop-reduce < %s | FileCheck -check-prefix=OPT %s ; RUN: opt -S -mtriple=amdgcn-- -mcpu=bonaire -loop-reduce < %s | FileCheck -check-prefix=OPT %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; Make sure the pointer / address space of AtomicRMW is considered ; Make sure the pointer / address space of AtomicRMW is considered
; OPT-LABEL: @test_local_atomicrmw_addressing_loop_uniform_index_max_offset_i32( ; OPT-LABEL: @test_local_atomicrmw_addressing_loop_uniform_index_max_offset_i32(
; OPT: .lr.ph.preheader: ; OPT: .lr.ph.preheader:
; OPT: %scevgep2 = getelementptr i8, ptr addrspace(3) %arg1, i32 65532 ; OPT: %scevgep2 = getelementptr i8, ptr addrspace(3) %arg1, i32 65532
; OPT: br label %.lr.ph ; OPT: br label %.lr.ph
▲ Show 20 Lines • Show All 162 Lines • Show Last 20 Lines

llvm/test/Transforms/LoopStrengthReduce/AMDGPU/different-addrspace-addressing-mode-loops.ll

; RUN: opt -S -mtriple=amdgcn-- -mcpu=bonaire -loop-reduce < %s | FileCheck -check-prefix=OPT %s ; RUN: opt -S -mtriple=amdgcn-- -mcpu=bonaire -loop-reduce < %s | FileCheck -check-prefix=OPT %s
; Test that loops with different maximum offsets for different address ; Test that loops with different maximum offsets for different address
; spaces are correctly handled. ; spaces are correctly handled.
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
; OPT-LABEL: @test_global_addressing_loop_uniform_index_max_offset_i32( ; OPT-LABEL: @test_global_addressing_loop_uniform_index_max_offset_i32(
; OPT: .lr.ph.preheader: ; OPT: .lr.ph.preheader:
; OPT: %scevgep2 = getelementptr i8, ptr addrspace(1) %arg1, i64 4095 ; OPT: %scevgep2 = getelementptr i8, ptr addrspace(1) %arg1, i64 4095
; OPT: br label %.lr.ph ; OPT: br label %.lr.ph
; OPT: {{^}}.lr.ph: ; OPT: {{^}}.lr.ph:
; OPT: %lsr.iv3 = phi ptr addrspace(1) [ %scevgep4, %.lr.ph ], [ %scevgep2, %.lr.ph.preheader ] ; OPT: %lsr.iv3 = phi ptr addrspace(1) [ %scevgep4, %.lr.ph ], [ %scevgep2, %.lr.ph.preheader ]
; OPT: load i8, ptr addrspace(1) %lsr.iv3, align 1 ; OPT: load i8, ptr addrspace(1) %lsr.iv3, align 1
▲ Show 20 Lines • Show All 148 Lines • Show Last 20 Lines

llvm/test/Transforms/LoopStrengthReduce/AMDGPU/lsr-invalid-ptr-extend.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -march=amdgcn -loop-reduce -S < %s | FileCheck %s ; RUN: opt -march=amdgcn -loop-reduce -S < %s | FileCheck %s
; REQUIRES: asserts ; REQUIRES: asserts
; Test that LSR does not attempt to extend a pointer type to an integer type, ; Test that LSR does not attempt to extend a pointer type to an integer type,
; which causes a SCEV analysis assertion. ; which causes a SCEV analysis assertion.
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
@gVar = external hidden local_unnamed_addr addrspace(3) global [1024 x double], align 16 @gVar = external hidden local_unnamed_addr addrspace(3) global [1024 x double], align 16
define amdgpu_kernel void @scaledregtest() local_unnamed_addr { define amdgpu_kernel void @scaledregtest() local_unnamed_addr {
; CHECK-LABEL: @scaledregtest( ; CHECK-LABEL: @scaledregtest(
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
▲ Show 20 Lines • Show All 95 Lines • Show Last 20 Lines

llvm/test/Transforms/LoopStrengthReduce/AMDGPU/lsr-postinc-pos-addrspace.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -march=amdgcn -mcpu=bonaire -loop-reduce -S < %s | FileCheck %s ; RUN: opt -march=amdgcn -mcpu=bonaire -loop-reduce -S < %s | FileCheck %s
; Test various conditions where OptimizeLoopTermCond doesn't look at a ; Test various conditions where OptimizeLoopTermCond doesn't look at a
; memory instruction use and fails to find the address space. ; memory instruction use and fails to find the address space.
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
define amdgpu_kernel void @local_cmp_user(i32 %arg0) nounwind { define amdgpu_kernel void @local_cmp_user(i32 %arg0) nounwind {
; CHECK-LABEL: @local_cmp_user( ; CHECK-LABEL: @local_cmp_user(
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[ARG0:%.*]], 1 ; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[ARG0:%.*]], 1
; CHECK-NEXT: br label [[BB11:%.*]] ; CHECK-NEXT: br label [[BB11:%.*]]
; CHECK: bb11: ; CHECK: bb11:
; CHECK-NEXT: [[LSR_IV1:%.*]] = phi i32 [ [[LSR_IV_NEXT2:%.*]], [[BB:%.*]] ], [ -2, [[ENTRY:%.*]] ] ; CHECK-NEXT: [[LSR_IV1:%.*]] = phi i32 [ [[LSR_IV_NEXT2:%.*]], [[BB:%.*]] ], [ -2, [[ENTRY:%.*]] ]
▲ Show 20 Lines • Show All 156 Lines • Show Last 20 Lines

llvm/test/Transforms/LoopStrengthReduce/AMDGPU/preserve-addrspace-assert.ll

; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -loop-reduce %s | FileCheck %s ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -loop-reduce %s | FileCheck %s
; Test for assert resulting from inconsistent isLegalAddressingMode ; Test for assert resulting from inconsistent isLegalAddressingMode
; answers when the address space was dropped from the query. ; answers when the address space was dropped from the query.
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
%0 = type { i32, double, i32, float } %0 = type { i32, double, i32, float }
; CHECK-LABEL: @lsr_crash_preserve_addrspace_unknown_type( ; CHECK-LABEL: @lsr_crash_preserve_addrspace_unknown_type(
; CHECK: %scevgep1 = getelementptr i8, ptr addrspace(3) %tmp, i32 8 ; CHECK: %scevgep1 = getelementptr i8, ptr addrspace(3) %tmp, i32 8
; CHECK: load double, ptr addrspace(3) %scevgep1 ; CHECK: load double, ptr addrspace(3) %scevgep1
; CHECK: %scevgep = getelementptr i8, ptr addrspace(3) %tmp, i32 16 ; CHECK: %scevgep = getelementptr i8, ptr addrspace(3) %tmp, i32 16
▲ Show 20 Lines • Show All 80 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/attributor_pointer_offset_crash.ll

; RUN: opt < %s -S -passes=openmp-opt ; RUN: opt < %s -S -passes=openmp-opt
; Verify the address space cast doesn't cause a crash ; Verify the address space cast doesn't cause a crash
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
%"struct.(anonymous namespace)::TeamStateTy" = type { %"struct.(anonymous namespace)::ICVStateTy", i32, ptr } %"struct.(anonymous namespace)::TeamStateTy" = type { %"struct.(anonymous namespace)::ICVStateTy", i32, ptr }
%"struct.(anonymous namespace)::ICVStateTy" = type { i32, i32, i32, i32, i32, i32 } %"struct.(anonymous namespace)::ICVStateTy" = type { i32, i32, i32, i32, i32, i32 }
%"struct.(anonymous namespace)::ThreadStateTy" = type { %"struct.(anonymous namespace)::ICVStateTy", ptr } %"struct.(anonymous namespace)::ThreadStateTy" = type { %"struct.(anonymous namespace)::ICVStateTy", ptr }
@_ZN12_GLOBAL__N_19TeamStateE = internal addrspace(3) global %"struct.(anonymous namespace)::TeamStateTy" undef, align 8 @_ZN12_GLOBAL__N_19TeamStateE = internal addrspace(3) global %"struct.(anonymous namespace)::TeamStateTy" undef, align 8
define internal ptr @__kmpc_alloc_shared() { define internal ptr @__kmpc_alloc_shared() {
Show All 17 Lines

llvm/test/Transforms/OpenMP/spmdization_constant_prop.ll

; RUN: opt -S -passes=openmp-opt < %s | FileCheck %s ; RUN: opt -S -passes=openmp-opt < %s | FileCheck %s
; ;
; Verify we change it to SPMD mode but also avoid propagating the old mode (=generic) into the __kmpc_target_init function. ; Verify we change it to SPMD mode but also avoid propagating the old mode (=generic) into the __kmpc_target_init function.
; ;
; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode ; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode
; CHECK: call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 2, i1 false) ; CHECK: call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 2, i1 false)
; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode ; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode
; CHECK: store i32 1, ptr addrspace(3) @IsSPMDMode ; CHECK: store i32 1, ptr addrspace(3) @IsSPMDMode
; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode ; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode
; ;
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
%struct.ident_t = type { i32, i32, i32, i32, ptr } %struct.ident_t = type { i32, i32, i32, i32, ptr }
%struct.DeviceEnvironmentTy = type { i32, i32, i32, i32 } %struct.DeviceEnvironmentTy = type { i32, i32, i32, i32 }
%"struct.(anonymous namespace)::SharedMemorySmartStackTy" = type { [512 x i8], [1024 x i8] } %"struct.(anonymous namespace)::SharedMemorySmartStackTy" = type { [512 x i8], [1024 x i8] }
%"struct.(anonymous namespace)::TeamStateTy" = type { %"struct.(anonymous namespace)::ICVStateTy", i32, ptr } %"struct.(anonymous namespace)::TeamStateTy" = type { %"struct.(anonymous namespace)::ICVStateTy", i32, ptr }
%"struct.(anonymous namespace)::ICVStateTy" = type { i32, i32, i32, i32, i32, i32 } %"struct.(anonymous namespace)::ICVStateTy" = type { i32, i32, i32, i32, i32, i32 }
▲ Show 20 Lines • Show All 308 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/values_in_offload_arrays.alloca.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -S -passes=openmp-opt-cgscc -aa-pipeline=basic-aa -openmp-hide-memory-transfer-latency < %s | FileCheck %s ; RUN: opt -S -passes=openmp-opt-cgscc -aa-pipeline=basic-aa -openmp-hide-memory-transfer-latency < %s | FileCheck %s
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
@.__omp_offloading_heavyComputation.region_id = weak constant i8 0 @.__omp_offloading_heavyComputation.region_id = weak constant i8 0
@.offload_maptypes. = private unnamed_addr constant [2 x i64] [i64 35, i64 35] @.offload_maptypes. = private unnamed_addr constant [2 x i64] [i64 35, i64 35]
%struct.ident_t = type { i32, i32, i32, i32, ptr } %struct.ident_t = type { i32, i32, i32, i32, ptr }
@.str = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00", align 1 @.str = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00", align 1
@0 = private unnamed_addr global %struct.ident_t { i32 0, i32 2, i32 0, i32 0, ptr @.str }, align 8 @0 = private unnamed_addr global %struct.ident_t { i32 0, i32 2, i32 0, i32 0, ptr @.str }, align 8
▲ Show 20 Lines • Show All 83 Lines • Show Last 20 Lines

llvm/test/Transforms/SLPVectorizer/AMDGPU/address-space-ptr-sze-gep-index-assert.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -passes=slp-vectorizer -slp-threshold=-18 < %s | FileCheck %s ; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -passes=slp-vectorizer -slp-threshold=-18 < %s | FileCheck %s
; Make sure there's no SCEV assert when the indexes are for different ; Make sure there's no SCEV assert when the indexes are for different
; sized address spaces ; sized address spaces
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5"
define void @slp_scev_assert(i32 %idx, i64 %tmp3) #0 { define void @slp_scev_assert(i32 %idx, i64 %tmp3) #0 {
; CHECK-LABEL: @slp_scev_assert( ; CHECK-LABEL: @slp_scev_assert(
; CHECK-NEXT: bb: ; CHECK-NEXT: bb:
; CHECK-NEXT: [[TMP:%.*]] = addrspacecast ptr addrspace(5) undef to ptr ; CHECK-NEXT: [[TMP:%.*]] = addrspacecast ptr addrspace(5) undef to ptr
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr addrspace(5) undef, i32 [[IDX:%.*]] ; CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr addrspace(5) undef, i32 [[IDX:%.*]]
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP]], i64 [[TMP3:%.*]] ; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds i8, ptr [[TMP]], i64 [[TMP3:%.*]]
; CHECK-NEXT: store i8 0, ptr addrspace(5) [[TMP2]], align 1 ; CHECK-NEXT: store i8 0, ptr addrspace(5) [[TMP2]], align 1
▲ Show 20 Lines • Show All 155 Lines • Show Last 20 Lines

llvm/test/Transforms/VectorCombine/AMDGPU/as-transition-inseltpoison.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -passes=vector-combine -S -mtriple=amdgcn-amd-amdhsa | FileCheck %s --check-prefixes=CHECK ; RUN: opt < %s -passes=vector-combine -S -mtriple=amdgcn-amd-amdhsa | FileCheck %s --check-prefixes=CHECK
; ModuleID = 'load-as-transition.ll' ; ModuleID = 'load-as-transition.ll'
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
%struct.hoge = type { float } %struct.hoge = type { float }
define protected amdgpu_kernel void @load_from_other_as(ptr nocapture nonnull %resultptr) local_unnamed_addr #0 { define protected amdgpu_kernel void @load_from_other_as(ptr nocapture nonnull %resultptr) local_unnamed_addr #0 {
; CHECK-LABEL: @load_from_other_as( ; CHECK-LABEL: @load_from_other_as(
; CHECK-NEXT: bb: ; CHECK-NEXT: bb:
; CHECK-NEXT: [[A:%.*]] = alloca [[STRUCT_HOGE:%.*]], align 4, addrspace(5) ; CHECK-NEXT: [[A:%.*]] = alloca [[STRUCT_HOGE:%.*]], align 4, addrspace(5)
Show All 20 Lines

llvm/test/Transforms/VectorCombine/AMDGPU/as-transition.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt < %s -passes=vector-combine -S -mtriple=amdgcn-amd-amdhsa | FileCheck %s --check-prefixes=CHECK ; RUN: opt < %s -passes=vector-combine -S -mtriple=amdgcn-amd-amdhsa | FileCheck %s --check-prefixes=CHECK
; ModuleID = 'load-as-transition.ll' ; ModuleID = 'load-as-transition.ll'
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-p7:160:256:256:32-p8:128:128-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
%struct.hoge = type { float } %struct.hoge = type { float }
define protected amdgpu_kernel void @load_from_other_as(ptr nocapture nonnull %resultptr) local_unnamed_addr #0 { define protected amdgpu_kernel void @load_from_other_as(ptr nocapture nonnull %resultptr) local_unnamed_addr #0 {
; CHECK-LABEL: @load_from_other_as( ; CHECK-LABEL: @load_from_other_as(
; CHECK-NEXT: bb: ; CHECK-NEXT: bb:
; CHECK-NEXT: [[A:%.*]] = alloca [[STRUCT_HOGE:%.*]], align 4, addrspace(5) ; CHECK-NEXT: [[A:%.*]] = alloca [[STRUCT_HOGE:%.*]], align 4, addrspace(5)
Show All 20 Lines

llvm/unittests/Bitcode/DataLayoutUpgradeTest.cpp

Show All 24 Lines EXPECT_EQ(DL1, "e-m:e-p:32:32-p270:32:32-p271:32:32-p272:64:64-i64:64"
"-f80:128-n8:16:32:64-S128"); "-f80:128-n8:16:32:64-S128");
EXPECT_EQ(DL2, "e-m:w-p:32:32-p270:32:32-p271:32:32-p272:64:64-i64:64" EXPECT_EQ(DL2, "e-m:w-p:32:32-p270:32:32-p271:32:32-p272:64:64-i64:64"
"-f80:128-n8:16:32-S32"); "-f80:128-n8:16:32-S32");
EXPECT_EQ(DL3, "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128" EXPECT_EQ(DL3, "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-i128:128"
"-n32:64-S128"); "-n32:64-S128");
// Check that AMDGPU targets add -G1 if it's not present. // Check that AMDGPU targets add -G1 if it's not present.
EXPECT_EQ(UpgradeDataLayoutString("e-p:32:32", "r600"), "e-p:32:32-G1"); EXPECT_EQ(UpgradeDataLayoutString("e-p:32:32", "r600"), "e-p:32:32-G1");
EXPECT_EQ(UpgradeDataLayoutString("e-p:64:64", "amdgcn"), "e-p:64:64-G1"); // and that ANDGCN adds p7 and p8 as well.
EXPECT_EQ(UpgradeDataLayoutString("e-p:64:64", "amdgcn"),
"e-p:64:64-G1-p7:160:256:256:32-p8:128:128-ni:7");
EXPECT_EQ(UpgradeDataLayoutString("e-p:64:64-G1", "amdgcn"),
"e-p:64:64-G1-p7:160:256:256:32-p8:128:128-ni:7");
// but that r600 does not.
EXPECT_EQ(UpgradeDataLayoutString("e-p:32:32-G1", "r600"), "e-p:32:32-G1");
// Check that RISCV64 upgrades -n64 to -n32:64. // Check that RISCV64 upgrades -n64 to -n32:64.
EXPECT_EQ(UpgradeDataLayoutString("e-m:e-p:64:64-i64:64-i128:128-n64-S128", EXPECT_EQ(UpgradeDataLayoutString("e-m:e-p:64:64-i64:64-i128:128-n64-S128",
"riscv64"), "riscv64"),
"e-m:e-p:64:64-i64:64-i128:128-n32:64-S128"); "e-m:e-p:64:64-i64:64-i128:128-n32:64-S128");
} }
TEST(DataLayoutUpgradeTest, NoDataLayoutUpgrade) { TEST(DataLayoutUpgradeTest, NoDataLayoutUpgrade) {
Show All 12 Lines EXPECT_EQ(DL1, "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64"
"-f80:128:128-n8:16:32:64-S128"); "-f80:128:128-n8:16:32:64-S128");
EXPECT_EQ(DL2, "e-p:32:32"); EXPECT_EQ(DL2, "e-p:32:32");
EXPECT_EQ(DL3, "e-m:e-i64:64-n32:64"); EXPECT_EQ(DL3, "e-m:e-i64:64-n32:64");
EXPECT_EQ(DL4, "e-m:o-i64:64-i128:128-n32:64-S128"); EXPECT_EQ(DL4, "e-m:o-i64:64-i128:128-n32:64-S128");
// Check that AMDGPU targets don't add -G1 if there is already a -G flag. // Check that AMDGPU targets don't add -G1 if there is already a -G flag.
EXPECT_EQ(UpgradeDataLayoutString("e-p:32:32-G2", "r600"), "e-p:32:32-G2"); EXPECT_EQ(UpgradeDataLayoutString("e-p:32:32-G2", "r600"), "e-p:32:32-G2");
EXPECT_EQ(UpgradeDataLayoutString("G2", "r600"), "G2"); EXPECT_EQ(UpgradeDataLayoutString("G2", "r600"), "G2");
EXPECT_EQ(UpgradeDataLayoutString("e-p:64:64-G2", "amdgcn"), "e-p:64:64-G2"); EXPECT_EQ(UpgradeDataLayoutString("e-p:64:64-G2", "amdgcn"),
EXPECT_EQ(UpgradeDataLayoutString("G2-e-p:64:64", "amdgcn"), "G2-e-p:64:64"); "e-p:64:64-G2-p7:160:256:256:32-p8:128:128-ni:7");
EXPECT_EQ(UpgradeDataLayoutString("e-p:64:64-G0", "amdgcn"), "e-p:64:64-G0"); EXPECT_EQ(UpgradeDataLayoutString("G2-e-p:64:64", "amdgcn"),
"G2-e-p:64:64-p7:160:256:256:32-p8:128:128-ni:7");
EXPECT_EQ(UpgradeDataLayoutString("e-p:64:64-G0", "amdgcn"),
"e-p:64:64-G0-p7:160:256:256:32-p8:128:128-ni:7");
// Check that AMDGCN targets don't add already declared address space 7.
EXPECT_EQ(UpgradeDataLayoutString("e-p:64:64-p7:64:64", "amdgcn"),
"e-p:64:64-p7:64:64-G1-p8:128:128-ni:7");
EXPECT_EQ(UpgradeDataLayoutString("p7:64:64-G2-e-p:64:64", "amdgcn"),
"p7:64:64-G2-e-p:64:64-p8:128:128-ni:7");
EXPECT_EQ(UpgradeDataLayoutString("e-p:64:64-p7:64:64-G1", "amdgcn"),
"e-p:64:64-p7:64:64-G1-p8:128:128-ni:7");
} }
TEST(DataLayoutUpgradeTest, EmptyDataLayout) { TEST(DataLayoutUpgradeTest, EmptyDataLayout) {
std::string DL1 = UpgradeDataLayoutString("", "x86_64-unknown-linux-gnu"); std::string DL1 = UpgradeDataLayoutString("", "x86_64-unknown-linux-gnu");
std::string DL2 = UpgradeDataLayoutString( std::string DL2 = UpgradeDataLayoutString(
"e-m:e-p:32:32-i64:64-f80:128-n8:16:32:64-S128", ""); "e-m:e-p:32:32-i64:64-f80:128-n8:16:32:64-S128", "");
EXPECT_EQ(DL1, ""); EXPECT_EQ(DL1, "");
EXPECT_EQ(DL2, "e-m:e-p:32:32-i64:64-f80:128-n8:16:32:64-S128"); EXPECT_EQ(DL2, "e-m:e-p:32:32-i64:64-f80:128-n8:16:32:64-S128");
// Check that AMDGPU targets add G1 if it's not present. // Check that AMDGPU targets add G1 if it's not present.
EXPECT_EQ(UpgradeDataLayoutString("", "r600"), "G1"); EXPECT_EQ(UpgradeDataLayoutString("", "r600"), "G1");
EXPECT_EQ(UpgradeDataLayoutString("", "amdgcn"), "G1"); EXPECT_EQ(UpgradeDataLayoutString("", "amdgcn"),
"G1-p7:160:256:256:32-p8:128:128-ni:7");
} }
} // end namespace } // end namespace
llvm/test/CodeGen/AMDGPU/GlobalISel/buffer-atomic-fadd.v2f16-rtn.ll