This is an archive of the discontinued LLVM Phabricator instance.

Differential D79455

[mlir][StandardToLLVM] Add support for lowering FPToSIOp to LLVM.
ClosedPublic

Authored by hanchung on May 5 2020, 4:39 PM.

Details

Reviewers
rriddle
asaadaldien
ftynse
Commits
rG3f07cab3124d: [mlir][StandardToLLVM] Add support for lowering FPToSIOp to LLVM.
Summary

Depends On D79374

Diff Detail

Event Timeline

hanchung created this revision.May 5 2020, 4:39 PM
Herald added a reviewer: ftynse. · View Herald TranscriptMay 5 2020, 4:39 PM
Herald added a project: Restricted Project. · View Herald Transcript
Herald added subscribers: llvm-commits, Kayjukh, frgossen and 12 others. · View Herald Transcript
asaadaldien accepted this revision.May 5 2020, 5:02 PM
This revision is now accepted and ready to land.May 5 2020, 5:02 PM
Harbormaster failed remote builds in B55864: Diff 262259!May 5 2020, 5:17 PM
ftynse accepted this revision.May 6 2020, 12:50 AM
Closed by commit rG3f07cab3124d: [mlir][StandardToLLVM] Add support for lowering FPToSIOp to LLVM. (authored by hanchung). · Explain WhyMay 11 2020, 1:34 AM
This revision was automatically updated to reflect the committed changes.
rriddle added inline comments.May 27 2020, 12:12 PM
mlir/lib/Conversion/StandardToLLVM/StandardToLLVM.cpp
2131

Can we start using using directives for these instead?

using FPToSILowering = OneToOneConvertToLLVMPattern<FPToSIOp, LLVM::FPToSIOp>;?

Herald added a subscriber: jurahul. · View Herald TranscriptMay 27 2020, 12:12 PM

Revision Contents

PathSize
mlir/
lib/
Conversion/
StandardToLLVM/
6 lines
test/
Conversion/
StandardToLLVM/
14 lines

Diff 262259

mlir/lib/Conversion/StandardToLLVM/StandardToLLVM.cpp

Show First 20 Lines • Show All 2,122 Lines • ▼ Show 20 Linesstruct SIToFPLowering
using Super::Super; using Super::Super;
}; };
struct FPExtLowering struct FPExtLowering
: public OneToOneConvertToLLVMPattern<FPExtOp, LLVM::FPExtOp> { : public OneToOneConvertToLLVMPattern<FPExtOp, LLVM::FPExtOp> {
using Super::Super; using Super::Super;
}; };
struct FPToSILowering
rriddleUnsubmitted
Not Done
ReplyInline Actions

Can we start using using directives for these instead?

using FPToSILowering = OneToOneConvertToLLVMPattern<FPToSIOp, LLVM::FPToSIOp>;?

rriddle: Can we start using `using` directives for these instead? `using FPToSILowering =…
: public OneToOneConvertToLLVMPattern<FPToSIOp, LLVM::FPToSIOp> {
using Super::Super;
};
struct FPTruncLowering struct FPTruncLowering
: public OneToOneConvertToLLVMPattern<FPTruncOp, LLVM::FPTruncOp> { : public OneToOneConvertToLLVMPattern<FPTruncOp, LLVM::FPTruncOp> {
using Super::Super; using Super::Super;
}; };
struct SignExtendIOpLowering struct SignExtendIOpLowering
: public OneToOneConvertToLLVMPattern<SignExtendIOp, LLVM::SExtOp> { : public OneToOneConvertToLLVMPattern<SignExtendIOp, LLVM::SExtOp> {
using Super::Super; using Super::Super;
▲ Show 20 Lines • Show All 761 Lines • ▼ Show 20 Lines patterns.insert<
DivFOpLowering, DivFOpLowering,
ExpOpLowering, ExpOpLowering,
Exp2OpLowering, Exp2OpLowering,
GenericAtomicRMWOpLowering, GenericAtomicRMWOpLowering,
LogOpLowering, LogOpLowering,
Log10OpLowering, Log10OpLowering,
Log2OpLowering, Log2OpLowering,
FPExtLowering, FPExtLowering,
FPToSILowering,
FPTruncLowering, FPTruncLowering,
IndexCastOpLowering, IndexCastOpLowering,
MulFOpLowering, MulFOpLowering,
MulIOpLowering, MulIOpLowering,
NegFOpLowering, NegFOpLowering,
OrOpLowering, OrOpLowering,
PrefetchOpLowering, PrefetchOpLowering,
RemFOpLowering, RemFOpLowering,
▲ Show 20 Lines • Show All 193 Lines • Show Last 20 Lines

mlir/test/Conversion/StandardToLLVM/convert-to-llvmir.mlir

Show First 20 Lines • Show All 546 Lines • ▼ Show 20 Lines// CHECK-NEXT: = llvm.fpext {{.*}} : !llvm<"<2 x half>"> to !llvm<"<2 x float>">
%0 = fpext %arg0: vector<2xf16> to vector<2xf32> %0 = fpext %arg0: vector<2xf16> to vector<2xf32>
// CHECK-NEXT: = llvm.fpext {{.*}} : !llvm<"<2 x half>"> to !llvm<"<2 x double>"> // CHECK-NEXT: = llvm.fpext {{.*}} : !llvm<"<2 x half>"> to !llvm<"<2 x double>">
%1 = fpext %arg0: vector<2xf16> to vector<2xf64> %1 = fpext %arg0: vector<2xf16> to vector<2xf64>
// CHECK-NEXT: = llvm.fpext {{.*}} : !llvm<"<2 x float>"> to !llvm<"<2 x double>"> // CHECK-NEXT: = llvm.fpext {{.*}} : !llvm<"<2 x float>"> to !llvm<"<2 x double>">
%2 = fpext %arg1: vector<2xf32> to vector<2xf64> %2 = fpext %arg1: vector<2xf32> to vector<2xf64>
return return
} }
// Checking conversion of floating point to integer types.
// CHECK-LABEL: @fptosi
func @fptosi(%arg0 : f32, %arg1 : f64) {
// CHECK-NEXT: = llvm.fptosi {{.*}} : !llvm.float to !llvm.i32
%0 = fptosi %arg0: f32 to i32
// CHECK-NEXT: = llvm.fptosi {{.*}} : !llvm.float to !llvm.i64
%1 = fptosi %arg0: f32 to i64
// CHECK-NEXT: = llvm.fptosi {{.*}} : !llvm.double to !llvm.i32
%2 = fptosi %arg1: f64 to i32
// CHECK-NEXT: = llvm.fptosi {{.*}} : !llvm.double to !llvm.i64
%3 = fptosi %arg1: f64 to i64
return
}
// Checking conversion of integer types to floating point. // Checking conversion of integer types to floating point.
// CHECK-LABEL: @fptrunc // CHECK-LABEL: @fptrunc
func @fptrunc(%arg0 : f32, %arg1 : f64) { func @fptrunc(%arg0 : f32, %arg1 : f64) {
// CHECK-NEXT: = llvm.fptrunc {{.*}} : !llvm.float to !llvm.half // CHECK-NEXT: = llvm.fptrunc {{.*}} : !llvm.float to !llvm.half
%0 = fptrunc %arg0: f32 to f16 %0 = fptrunc %arg0: f32 to f16
// CHECK-NEXT: = llvm.fptrunc {{.*}} : !llvm.double to !llvm.half // CHECK-NEXT: = llvm.fptrunc {{.*}} : !llvm.double to !llvm.half
%1 = fptrunc %arg1: f64 to f16 %1 = fptrunc %arg1: f64 to f16
// CHECK-NEXT: = llvm.fptrunc {{.*}} : !llvm.double to !llvm.float // CHECK-NEXT: = llvm.fptrunc {{.*}} : !llvm.double to !llvm.float
▲ Show 20 Lines • Show All 620 Lines • Show Last 20 Lines