Skip to content

[FPEnv][AMDGPU] Correct AMDGPUSimplifyLibCalls handling of strictfp attribute. #86705

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Mar 27, 2024
Merged

[FPEnv][AMDGPU] Correct AMDGPUSimplifyLibCalls handling of strictfp attribute. #86705

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions llvm/lib/Target/AMDGPU/AMDGPULibCalls.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -657,6 +657,8 @@ bool AMDGPULibCalls::fold(CallInst *CI) {
return true;

IRBuilder<> B(CI);
if (CI->isStrictFP())
B.setIsFPConstrained(true);

if (FPMathOperator *FPOp = dyn_cast<FPMathOperator>(CI)) {
// Under unsafe-math, evaluate calls if possible.
Expand Down
12 changes: 6 additions & 6 deletions llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-pown.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -808,7 +808,7 @@ define float @test_pown_fast_f32_nobuiltin(float %x, i32 %y) {
; CHECK-LABEL: define float @test_pown_fast_f32_nobuiltin
; CHECK-SAME: (float [[X:%.*]], i32 [[Y:%.*]]) {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[CALL:%.*]] = tail call fast float @_Z4pownfi(float [[X]], i32 [[Y]]) #[[ATTR3:[0-9]+]]
; CHECK-NEXT: [[CALL:%.*]] = tail call fast float @_Z4pownfi(float [[X]], i32 [[Y]]) #[[ATTR4:[0-9]+]]
; CHECK-NEXT: ret float [[CALL]]
;
entry:
Expand All @@ -820,11 +820,11 @@ define float @test_pown_fast_f32_strictfp(float %x, i32 %y) #1 {
; CHECK-LABEL: define float @test_pown_fast_f32_strictfp
; CHECK-SAME: (float [[X:%.*]], i32 [[Y:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[__FABS:%.*]] = call fast float @llvm.fabs.f32(float [[X]])
; CHECK-NEXT: [[__LOG2:%.*]] = call fast float @llvm.log2.f32(float [[__FABS]])
; CHECK-NEXT: [[POWNI2F:%.*]] = sitofp i32 [[Y]] to float
; CHECK-NEXT: [[__YLOGX:%.*]] = fmul fast float [[__LOG2]], [[POWNI2F]]
; CHECK-NEXT: [[__EXP2:%.*]] = call fast float @llvm.exp2.f32(float [[__YLOGX]])
; CHECK-NEXT: [[__FABS:%.*]] = call fast float @llvm.fabs.f32(float [[X]]) #[[ATTR0]]
; CHECK-NEXT: [[__LOG2:%.*]] = call fast float @llvm.log2.f32(float [[__FABS]]) #[[ATTR0]]
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this is still emitting an unconstrained log2

; CHECK-NEXT: [[POWNI2F:%.*]] = call fast float @llvm.experimental.constrained.sitofp.f32.i32(i32 [[Y]], metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR0]]
; CHECK-NEXT: [[__YLOGX:%.*]] = call fast float @llvm.experimental.constrained.fmul.f32(float [[POWNI2F]], float [[__LOG2]], metadata !"round.dynamic", metadata !"fpexcept.strict") #[[ATTR0]]
; CHECK-NEXT: [[__EXP2:%.*]] = call fast float @llvm.exp2.f32(float [[__YLOGX]]) #[[ATTR0]]
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

unconstrained exp2

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The way to fix those would be to add code to the IRBuilder in CreateCall to check for strictfp mode and switch over to the constrained call. That code hasn't been written yet. Is that something I need to do before I can move forward here?

; CHECK-NEXT: [[__YEVEN:%.*]] = shl i32 [[Y]], 31
; CHECK-NEXT: [[TMP0:%.*]] = bitcast float [[X]] to i32
; CHECK-NEXT: [[__POW_SIGN:%.*]] = and i32 [[__YEVEN]], [[TMP0]]
Expand Down
2 changes: 1 addition & 1 deletion llvm/test/CodeGen/AMDGPU/amdgpu-simplify-libcall-rootn.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -896,7 +896,7 @@ define float @test_rootn_f32__y_neg2__strictfp(float %x) #1 {
; CHECK-LABEL: define float @test_rootn_f32__y_neg2__strictfp(
; CHECK-SAME: float [[X:%.*]]) #[[ATTR0]] {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[__ROOTN2RSQRT:%.*]] = call float @_Z5rsqrtf(float [[X]])
; CHECK-NEXT: [[__ROOTN2RSQRT:%.*]] = call float @_Z5rsqrtf(float [[X]]) #[[ATTR0]]
; CHECK-NEXT: ret float [[__ROOTN2RSQRT]]
;
entry:
Expand Down