People kzhuravl

kzhuravl (Konstantin Zhuravlyov)
User

Projects

User does not belong to any projects.

User Details

User Since
Feb 16 2016, 12:13 PM (135 w, 2 d)

Recent Activity
View All

Mon, Sep 17

kzhuravl accepted D52123: [AMDGPU] Initialize instruction itinerary from GCNSubtarget.

LGTM

Mon, Sep 17, 7:58 AM

Fri, Sep 14

kzhuravl committed rL342270: AMDGPU: Clear the bits before they are being set in program resource registers.
AMDGPU: Clear the bits before they are being set in program resource registers
Fri, Sep 14, 1:03 PM

Wed, Sep 12

kzhuravl committed rL342077: AMDGPU: Print all kernel descriptor directives (including the ones with default….
AMDGPU: Print all kernel descriptor directives (including the ones with default…
Wed, Sep 12, 1:27 PM
This revision was not accepted when it landed; it landed in state Needs Review.
Wed, Sep 12, 1:27 PM
kzhuravl committed rL342069: AMDGPU: Re-apply r341982 after fixing the layering issue.
AMDGPU: Re-apply r341982 after fixing the layering issue
Wed, Sep 12, 11:52 AM

Tue, Sep 11

kzhuravl created D51954: AMDGPU: Print all kernel descriptor directives (including the ones with default values).
Tue, Sep 11, 3:27 PM
kzhuravl committed rL341982: AMDGPU: Move isa version and EF_AMDGPU_MACH_* determination.
AMDGPU: Move isa version and EF_AMDGPU_MACH_* determination
Tue, Sep 11, 12:00 PM
kzhuravl closed D51890: AMDGPU: Move isa version and EF_AMDGPU_MACH_* determination into TargetParser..
Tue, Sep 11, 12:00 PM

Mon, Sep 10

kzhuravl created D51890: AMDGPU: Move isa version and EF_AMDGPU_MACH_* determination into TargetParser..
Mon, Sep 10, 3:01 PM

Fri, Sep 7

kzhuravl accepted D51790: AMDGPU: Don't abort on unknown addrspace argument.

LGTM

Fri, Sep 7, 9:06 AM

Aug 16 2018

kzhuravl accepted D50843: AMDGPU: Correct errors in device table.

LGTM

Aug 16 2018, 1:04 PM

Aug 9 2018

kzhuravl accepted D50493: AMDGPU: Add another missing builtin.

LGTM

Aug 9 2018, 10:15 AM

Aug 8 2018

kzhuravl accepted D50376: AMDGPU: Fix enabling denormals by default on pre-VI targets.

LGTM

Aug 8 2018, 6:05 AM

Aug 6 2018

kzhuravl accepted D50320: AMDGPU: Add builtin for s_dcache_inv_vol.

LGTM

Aug 6 2018, 10:44 AM
kzhuravl accepted D50321: AMDGPU: Add builtin for s_dcache_wb.

LGTM

Aug 6 2018, 10:43 AM

Aug 3 2018

kzhuravl accepted D50265: AMDGPU: Add feature vi-insts.

LGTM

Aug 3 2018, 11:05 AM

Jul 31 2018

kzhuravl committed rC338471: AMDGPU: Add clamp bit to dot builtins.
AMDGPU: Add clamp bit to dot builtins
Jul 31 2018, 6:32 PM
kzhuravl committed rL338471: AMDGPU: Add clamp bit to dot builtins.
AMDGPU: Add clamp bit to dot builtins
Jul 31 2018, 6:32 PM
kzhuravl closed D50011: AMDGPU: Add clamp bit to dot builtins.
Jul 31 2018, 6:32 PM
kzhuravl committed rL338470: AMDGPU: Add clamp bit to dot intrinsics.
AMDGPU: Add clamp bit to dot intrinsics
Jul 31 2018, 6:32 PM
kzhuravl closed D49874: AMDGPU: Add clamp bit to dot intrinsics .
Jul 31 2018, 6:31 PM
kzhuravl added a comment to D49874: AMDGPU: Add clamp bit to dot intrinsics .

Ping.

Jul 31 2018, 8:30 AM
kzhuravl added inline comments to D50024: [AMDGPU] Support idot2 pattern..
Jul 31 2018, 8:30 AM

Jul 30 2018

kzhuravl added a dependency for D50011: AMDGPU: Add clamp bit to dot builtins: D49874: AMDGPU: Add clamp bit to dot intrinsics .
Jul 30 2018, 2:18 PM
kzhuravl added a dependent revision for D49874: AMDGPU: Add clamp bit to dot intrinsics : D50011: AMDGPU: Add clamp bit to dot builtins.
Jul 30 2018, 2:18 PM
kzhuravl created D50011: AMDGPU: Add clamp bit to dot builtins.
Jul 30 2018, 2:17 PM
kzhuravl added inline comments to D49874: AMDGPU: Add clamp bit to dot intrinsics .
Jul 30 2018, 2:17 PM
kzhuravl updated the diff for D49874: AMDGPU: Add clamp bit to dot intrinsics .

Address review feedback.

Jul 30 2018, 2:16 PM

Jul 26 2018

kzhuravl created D49874: AMDGPU: Add clamp bit to dot intrinsics .
Jul 26 2018, 2:12 PM

Jul 20 2018

kzhuravl committed rC337612: AMDGPU: Switch default dwarf version to 2.
AMDGPU: Switch default dwarf version to 2
Jul 20 2018, 1:51 PM
kzhuravl committed rL337612: AMDGPU: Switch default dwarf version to 2.
AMDGPU: Switch default dwarf version to 2
Jul 20 2018, 1:51 PM
kzhuravl closed D49613: AMDGPU: Switch default dwarf version to 2.
Jul 20 2018, 1:51 PM
kzhuravl created D49613: AMDGPU: Switch default dwarf version to 2.
Jul 20 2018, 1:27 PM

Jul 17 2018

kzhuravl added a comment to D49448: [AMDGPU] Fix VGPR spills where offset doesn't fit in 12 bits.

Needs a test

Jul 17 2018, 1:25 PM

Jul 11 2018

kzhuravl committed rL336825: AMDGPU/NFC: Use already available explicit kernarg.
AMDGPU/NFC: Use already available explicit kernarg
Jul 11 2018, 10:32 AM

Jul 10 2018

kzhuravl committed rL336697: AMDGPU: Make hidden argument metadata consistent with.
AMDGPU: Make hidden argument metadata consistent with
Jul 10 2018, 9:18 AM
kzhuravl closed D49096: AMDGPU: Make hidden argument metadata consistent with amdgpu-implicitarg-num-bytes attribute.
Jul 10 2018, 9:17 AM
kzhuravl committed rL336689: AMDGPU/NFC: Fix typo in test name.
AMDGPU/NFC: Fix typo in test name
Jul 10 2018, 8:59 AM

Jul 9 2018

kzhuravl updated the diff for D49096: AMDGPU: Make hidden argument metadata consistent with amdgpu-implicitarg-num-bytes attribute.

Add missing space.

Jul 9 2018, 1:07 PM
kzhuravl updated the diff for D49096: AMDGPU: Make hidden argument metadata consistent with amdgpu-implicitarg-num-bytes attribute.

Drop checks for OpenCL. I was not aware that it was decided to possibly allow this for non-OpenCL languages.

Jul 9 2018, 1:06 PM
kzhuravl created D49096: AMDGPU: Make hidden argument metadata consistent with amdgpu-implicitarg-num-bytes attribute.
Jul 9 2018, 12:09 PM
kzhuravl accepted D49089: [AMDGPU][Waitcnt] fix "comparison of integers of different signs" build error.

LGTM

Jul 9 2018, 10:37 AM · Restricted Project

Jun 27 2018

kzhuravl accepted D48639: AMDGPU: Fix AMDGPUCodeGenPrepare using uninitialized AMDGPUAS struct.
Jun 27 2018, 7:12 AM
kzhuravl added a comment to D48639: AMDGPU: Fix AMDGPUCodeGenPrepare using uninitialized AMDGPUAS struct.

Not sure how this wasn't noticed before.

Jun 27 2018, 7:12 AM

Jun 26 2018

kzhuravl committed rL335707: AMDGPU/NFC: Fix typo in comment.
AMDGPU/NFC: Fix typo in comment
Jun 26 2018, 10:40 PM
kzhuravl committed rL335706: AMDHSA: Rename RESERVED -> RESERVED0, mark gfx9-specific field.
AMDHSA: Rename RESERVED -> RESERVED0, mark gfx9-specific field
Jun 26 2018, 10:23 PM
kzhuravl committed rL335705: AMDHSA/NFC: Address missed review feedback from.
AMDHSA/NFC: Address missed review feedback from
Jun 26 2018, 10:00 PM
kzhuravl committed rL335669: AMDGPU: Silence unused warnings in waitcnt insertion pass in release build.
AMDGPU: Silence unused warnings in waitcnt insertion pass in release build
Jun 26 2018, 2:38 PM
kzhuravl closed D48607: AMDGPU: Silence unused warnings in waitcnt insertion pass in release build.
Jun 26 2018, 2:38 PM
kzhuravl added inline comments to D48607: AMDGPU: Silence unused warnings in waitcnt insertion pass in release build.
Jun 26 2018, 2:07 PM
kzhuravl created D48607: AMDGPU: Silence unused warnings in waitcnt insertion pass in release build.
Jun 26 2018, 2:07 PM

Jun 22 2018

kzhuravl committed rL335378: AMDHSA: Put old assembler docs back.
AMDHSA: Put old assembler docs back
Jun 22 2018, 12:28 PM
kzhuravl closed D48497: AMDHSA: Put old assembler docs back.
Jun 22 2018, 12:28 PM
kzhuravl added a comment to D47736: AMDHSA Code Object v3 assembler syntax update.
In D47736#1140911, @artem.tamazov wrote:
In D47736#1140852, @kzhuravl wrote:
In D47736#1140847, @artem.tamazov wrote:

Is it so that the updated documentation is valid only if code-object-v3 is specified?

Yes

Is the V3 code object compatible with the latest ROCm release?

Not yet

So this change could confuse AMDGPU asm programmers.

Jun 22 2018, 10:57 AM
kzhuravl created D48497: AMDHSA: Put old assembler docs back.
Jun 22 2018, 10:57 AM
kzhuravl added a comment to D47736: AMDHSA Code Object v3 assembler syntax update.
In D47736#1140847, @artem.tamazov wrote:

Is it so that the updated documentation is valid only if code-object-v3 is specified?

Jun 22 2018, 10:12 AM
kzhuravl accepted D48486: AMDGPU: Respect align argument parameter.

LGTM

Jun 22 2018, 9:00 AM

Jun 21 2018

kzhuravl committed rL335288: AMDGPU: Remove ability to reserve VGPRs for debugger.
AMDGPU: Remove ability to reserve VGPRs for debugger
Jun 21 2018, 1:33 PM
kzhuravl closed D48234: AMDGPU: Remove ability to reserve VGPRs for debugger.
Jun 21 2018, 1:32 PM
kzhuravl committed rC335287: AMDGPU: Remove amdgpu-debugger-reserve-regs feature.
AMDGPU: Remove amdgpu-debugger-reserve-regs feature
Jun 21 2018, 1:32 PM
kzhuravl committed rL335287: AMDGPU: Remove amdgpu-debugger-reserve-regs feature.
AMDGPU: Remove amdgpu-debugger-reserve-regs feature
Jun 21 2018, 1:32 PM
kzhuravl added a comment to D48435: AMDGPU: Add patterns for i32/i64 local atomic load/store.
In D48435#1139714, @t-tye wrote:

Now this is implemented it may be worth converting the memorylegalizer tests from MIR to IR tests.

Jun 21 2018, 11:46 AM
kzhuravl committed rL335267: AMDGPU/AMDHSA: Remove GridWorkGroupCountX/Y/Z.
AMDGPU/AMDHSA: Remove GridWorkGroupCountX/Y/Z
Jun 21 2018, 11:40 AM
kzhuravl closed D48191: AMDGPU/AMDHSA: Remove GridWorkGroupCountX/Y/Z.
Jun 21 2018, 11:40 AM
kzhuravl accepted D47736: AMDHSA Code Object v3 assembler syntax update.

LGTM

Jun 21 2018, 11:13 AM
kzhuravl accepted D48435: AMDGPU: Add patterns for i32/i64 local atomic load/store.

LGTM. Orderings are already handled correctly.

Jun 21 2018, 8:56 AM

Jun 20 2018

kzhuravl accepted D48385: AMDGPU: Fix scalar_to_vector for v4i16/v4f16.

LGTM

Jun 20 2018, 12:12 PM

Jun 15 2018

kzhuravl created D48234: AMDGPU: Remove ability to reserve VGPRs for debugger.
Jun 15 2018, 1:25 PM
kzhuravl added inline comments to D47736: AMDHSA Code Object v3 assembler syntax update.
Jun 15 2018, 7:31 AM

Jun 14 2018

kzhuravl added inline comments to D47736: AMDHSA Code Object v3 assembler syntax update.
Jun 14 2018, 2:29 PM
kzhuravl updated subscribers of D48191: AMDGPU/AMDHSA: Remove GridWorkGroupCountX/Y/Z.
Jun 14 2018, 1:24 PM
kzhuravl created D48191: AMDGPU/AMDHSA: Remove GridWorkGroupCountX/Y/Z.
Jun 14 2018, 1:11 PM
kzhuravl added inline comments to D47736: AMDHSA Code Object v3 assembler syntax update.
Jun 14 2018, 10:03 AM

Jun 13 2018

kzhuravl added inline comments to D47736: AMDHSA Code Object v3 assembler syntax update.
Jun 13 2018, 6:50 PM
kzhuravl added a comment to D48103: [AMDGPU] Update code object metadata format documentation.

This seems like a duplicate of D47549?

Jun 13 2018, 6:35 PM
kzhuravl accepted D47549: [AMDGPU] Update code object metadata format documentation.

looks good to me

Jun 13 2018, 6:34 PM
kzhuravl accepted D48101: [AMDGPU] Document the AMDGPU LLVM attributes.

looks good to me

Jun 13 2018, 6:33 PM

Jun 12 2018

kzhuravl closed D42257: AMDGPU/MemoryModel: Fix monotonic atomic loads.

Done.

Jun 12 2018, 11:43 AM
kzhuravl added a comment to D47736: AMDHSA Code Object v3 assembler syntax update.

Missing codegen tests.

Jun 12 2018, 11:42 AM
kzhuravl closed D47566: AMDHSA: Code object v3 updates.

rL334519

Jun 12 2018, 11:42 AM
kzhuravl committed rL334521: AMDHSA/NFC: Code object v3 updates (additional):.
AMDHSA/NFC: Code object v3 updates (additional):
Jun 12 2018, 11:38 AM
kzhuravl committed rL334519: AMDHSA: Code object v3 updates.
AMDHSA: Code object v3 updates
Jun 12 2018, 11:07 AM

Jun 11 2018

kzhuravl committed rLLD334444: AMDGPU/LLD: Handle R_AMDGPU_REL64 relocation.
AMDGPU/LLD: Handle R_AMDGPU_REL64 relocation
Jun 11 2018, 2:47 PM
kzhuravl committed rL334444: AMDGPU/LLD: Handle R_AMDGPU_REL64 relocation.
AMDGPU/LLD: Handle R_AMDGPU_REL64 relocation
Jun 11 2018, 2:47 PM
kzhuravl closed D47734: AMDGPU/LLD: Handle R_AMDGPU_REL64 relocation.
Jun 11 2018, 2:47 PM
kzhuravl committed rL334443: AMDGPU: Add 64-bit relative variant kind.
AMDGPU: Add 64-bit relative variant kind
Jun 11 2018, 2:42 PM
kzhuravl closed D47601: AMDGPU: Add 64-bit relative variant kind.
Jun 11 2018, 2:42 PM

Jun 7 2018

kzhuravl added inline comments to D47566: AMDHSA: Code object v3 updates.
Jun 7 2018, 12:33 PM
kzhuravl accepted D47900: AMDGPU: Error on LDS global address in functions.

LGTM

Jun 7 2018, 12:20 PM
kzhuravl accepted D47504: [AMDGPU] Simplify memory legalizer.

LGTM

Jun 7 2018, 11:52 AM

Jun 5 2018

kzhuravl added inline comments to D47566: AMDHSA: Code object v3 updates.
Jun 5 2018, 5:42 PM
kzhuravl updated the diff for D47566: AMDHSA: Code object v3 updates.

Address review feedback.

Jun 5 2018, 5:42 PM
kzhuravl added inline comments to D47736: AMDHSA Code Object v3 assembler syntax update.
Jun 5 2018, 12:24 PM

Jun 4 2018

kzhuravl added inline comments to D47566: AMDHSA: Code object v3 updates.
Jun 4 2018, 3:24 PM
kzhuravl added a dependent revision for D47601: AMDGPU: Add 64-bit relative variant kind: D47734: AMDGPU/LLD: Handle R_AMDGPU_REL64 relocation.
Jun 4 2018, 1:06 PM
kzhuravl added a dependency for D47734: AMDGPU/LLD: Handle R_AMDGPU_REL64 relocation: D47601: AMDGPU: Add 64-bit relative variant kind.
Jun 4 2018, 1:06 PM
kzhuravl created D47734: AMDGPU/LLD: Handle R_AMDGPU_REL64 relocation.
Jun 4 2018, 11:42 AM
kzhuravl updated the diff for D47566: AMDHSA: Code object v3 updates.

Fix symbol bindings and rebase.

Jun 4 2018, 11:40 AM
kzhuravl updated the diff for D47601: AMDGPU: Add 64-bit relative variant kind.

Rebase.

Jun 4 2018, 11:40 AM
kzhuravl added a comment to D47601: AMDGPU: Add 64-bit relative variant kind.
In D47601#1121159, @t-tye wrote:
In D47601#1120916, @nhaehnle wrote:
In D47601#1120903, @nhaehnle wrote:

What is this actually needed for? Having a relative relocation in a data segment doesn't seem that useful?

Never mind, I see it in the other commit. Could you please update the table of relocation types in AMDGPUUsage.rst? It has an R_AMDGPU_RELATIVE64, I don't know what that's about, but it doesn't mention R_AMDGPU_REL64.

The AMDGPU_REL64 is defined in AMDGPUUsage as:

'''
Relocation Type Kind Value Field Calculation
R_AMDGPU_REL64 Static 5 word64 S + A - P
'''

It is a static relocation used to create PC relative values between sections. For example, there is a field in the kernel descriptor that has to be the delta from the kernel descriptor to the kernel entry point. So if the kernel descriptor is put in a different section to the code, a static relocation record is need. It will be resolved when the executable is linked.

The R_AMDGPU_RELATIVE64 is a dynamic relocation record implemented by most targets to support having absolute addresses in data sections that need to be initialized to the address of a symbol in the same shared object. For example, a variable initialized to the address of another variable which is considered a compile time constant in languages such as C/C++.

Jun 4 2018, 10:25 AM