Skip to content

[CostModel][X86] Treat lrint/llrint as fptosi calls #90883

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
May 3, 2024
Merged

[CostModel][X86] Treat lrint/llrint as fptosi calls #90883

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
10 changes: 10 additions & 0 deletions llvm/lib/Target/X86/X86TargetTransformInfo.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4153,6 +4153,16 @@ X86TTIImpl::getIntrinsicInstrCost(const IntrinsicCostAttributes &ICA,
}
}
break;
case Intrinsic::lrint:
case Intrinsic::llrint:
// X86 can use the CVTP2SI instructions to lower lrint/llrint calls, which
// have the same costs as the CVTTP2SI (fptosi) instructions
if (!ICA.isTypeBasedOnly()) {
const SmallVectorImpl<Type *> &ArgTys = ICA.getArgTypes();
return getCastInstrCost(Instruction::FPToSI, RetTy, ArgTys[0],
TTI::CastContextHint::None, CostKind);
}
break;
case Intrinsic::maxnum:
case Intrinsic::minnum:
// FMINNUM has same costs so don't duplicate.
Expand Down
89 changes: 39 additions & 50 deletions llvm/test/Analysis/CostModel/X86/arith-fp-codesize.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -666,35 +666,46 @@ define i32 @rint(i32 %arg) {
define i32 @lrint(i32 %arg) {
; SSE1-LABEL: 'lrint'
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call i32 @llvm.lrint.i32.f32(float undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x i32> @llvm.lrint.v4i32.v4f32(<4 x float> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x i32> @llvm.lrint.v8i32.v8f32(<8 x float> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16F32 = call <16 x i32> @llvm.lrint.v16i32.v16f32(<16 x float> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x i32> @llvm.lrint.v4i32.v4f32(<4 x float> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x i32> @llvm.lrint.v8i32.v8f32(<8 x float> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x i32> @llvm.lrint.v16i32.v16f32(<16 x float> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call i32 @llvm.lrint.i32.f64(double undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x i32> @llvm.lrint.v2i32.v2f64(<2 x double> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x i32> @llvm.lrint.v4i32.v4f64(<4 x double> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x i32> @llvm.lrint.v8i32.v8f64(<8 x double> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x i32> @llvm.lrint.v2i32.v2f64(<2 x double> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x i32> @llvm.lrint.v4i32.v4f64(<4 x double> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x i32> @llvm.lrint.v8i32.v8f64(<8 x double> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
; SSE2-LABEL: 'lrint'
; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call i32 @llvm.lrint.i32.f32(float undef)
; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x i32> @llvm.lrint.v4i32.v4f32(<4 x float> undef)
; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V8F32 = call <8 x i32> @llvm.lrint.v8i32.v8f32(<8 x float> undef)
; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V16F32 = call <16 x i32> @llvm.lrint.v16i32.v16f32(<16 x float> undef)
; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call i32 @llvm.lrint.i32.f64(double undef)
; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x i32> @llvm.lrint.v2i32.v2f64(<2 x double> undef)
; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V4F64 = call <4 x i32> @llvm.lrint.v4i32.v4f64(<4 x double> undef)
; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V8F64 = call <8 x i32> @llvm.lrint.v8i32.v8f64(<8 x double> undef)
; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
; AVX-LABEL: 'lrint'
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call i32 @llvm.lrint.i32.f32(float undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x i32> @llvm.lrint.v4i32.v4f32(<4 x float> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V8F32 = call <8 x i32> @llvm.lrint.v8i32.v8f32(<8 x float> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %V16F32 = call <16 x i32> @llvm.lrint.v16i32.v16f32(<16 x float> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x i32> @llvm.lrint.v4i32.v4f32(<4 x float> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x i32> @llvm.lrint.v8i32.v8f32(<8 x float> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %V16F32 = call <16 x i32> @llvm.lrint.v16i32.v16f32(<16 x float> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call i32 @llvm.lrint.i32.f64(double undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x i32> @llvm.lrint.v2i32.v2f64(<2 x double> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x i32> @llvm.lrint.v4i32.v4f64(<4 x double> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V8F64 = call <8 x i32> @llvm.lrint.v8i32.v8f64(<8 x double> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x i32> @llvm.lrint.v2i32.v2f64(<2 x double> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x i32> @llvm.lrint.v4i32.v4f64(<4 x double> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x i32> @llvm.lrint.v8i32.v8f64(<8 x double> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
; AVX512-LABEL: 'lrint'
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call i32 @llvm.lrint.i32.f32(float undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x i32> @llvm.lrint.v4i32.v4f32(<4 x float> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V8F32 = call <8 x i32> @llvm.lrint.v8i32.v8f32(<8 x float> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %V16F32 = call <16 x i32> @llvm.lrint.v16i32.v16f32(<16 x float> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x i32> @llvm.lrint.v4i32.v4f32(<4 x float> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x i32> @llvm.lrint.v8i32.v8f32(<8 x float> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x i32> @llvm.lrint.v16i32.v16f32(<16 x float> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call i32 @llvm.lrint.i32.f64(double undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x i32> @llvm.lrint.v2i32.v2f64(<2 x double> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x i32> @llvm.lrint.v4i32.v4f64(<4 x double> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %V8F64 = call <8 x i32> @llvm.lrint.v8i32.v8f64(<8 x double> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x i32> @llvm.lrint.v2i32.v2f64(<2 x double> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x i32> @llvm.lrint.v4i32.v4f64(<4 x double> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x i32> @llvm.lrint.v8i32.v8f64(<8 x double> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
%F32 = call i32 @llvm.lrint.i32.f32(float undef)
Expand All @@ -711,38 +722,16 @@ define i32 @lrint(i32 %arg) {
}

define i32 @llrint(i32 %arg) {
; SSE1-LABEL: 'llrint'
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call i64 @llvm.llrint.i64.f32(float undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F32 = call <4 x i64> @llvm.llrint.v4i64.v4f32(<4 x float> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F32 = call <8 x i64> @llvm.llrint.v8i64.v8f32(<8 x float> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 32 for instruction: %V16F32 = call <16 x i64> @llvm.llrint.v16i64.v16f32(<16 x float> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call i64 @llvm.llrint.i64.f64(double undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x i64> @llvm.llrint.v2i64.v2f64(<2 x double> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %V4F64 = call <4 x i64> @llvm.llrint.v4i64.v4f64(<4 x double> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %V8F64 = call <8 x i64> @llvm.llrint.v8i64.v8f64(<8 x double> undef)
; SSE1-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
; AVX-LABEL: 'llrint'
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call i64 @llvm.llrint.i64.f32(float undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V4F32 = call <4 x i64> @llvm.llrint.v4i64.v4f32(<4 x float> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V8F32 = call <8 x i64> @llvm.llrint.v8i64.v8f32(<8 x float> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 36 for instruction: %V16F32 = call <16 x i64> @llvm.llrint.v16i64.v16f32(<16 x float> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call i64 @llvm.llrint.i64.f64(double undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x i64> @llvm.llrint.v2i64.v2f64(<2 x double> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V4F64 = call <4 x i64> @llvm.llrint.v4i64.v4f64(<4 x double> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 18 for instruction: %V8F64 = call <8 x i64> @llvm.llrint.v8i64.v8f64(<8 x double> undef)
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
; AVX512-LABEL: 'llrint'
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call i64 @llvm.llrint.i64.f32(float undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V4F32 = call <4 x i64> @llvm.llrint.v4i64.v4f32(<4 x float> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %V8F32 = call <8 x i64> @llvm.llrint.v8i64.v8f32(<8 x float> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 38 for instruction: %V16F32 = call <16 x i64> @llvm.llrint.v16i64.v16f32(<16 x float> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call i64 @llvm.llrint.i64.f64(double undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %V2F64 = call <2 x i64> @llvm.llrint.v2i64.v2f64(<2 x double> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %V4F64 = call <4 x i64> @llvm.llrint.v4i64.v4f64(<4 x double> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %V8F64 = call <8 x i64> @llvm.llrint.v8i64.v8f64(<8 x double> undef)
; AVX512-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
; CHECK-LABEL: 'llrint'
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F32 = call i64 @llvm.llrint.i64.f32(float undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F32 = call <4 x i64> @llvm.llrint.v4i64.v4f32(<4 x float> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F32 = call <8 x i64> @llvm.llrint.v8i64.v8f32(<8 x float> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V16F32 = call <16 x i64> @llvm.llrint.v16i64.v16f32(<16 x float> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %F64 = call i64 @llvm.llrint.i64.f64(double undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V2F64 = call <2 x i64> @llvm.llrint.v2i64.v2f64(<2 x double> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V4F64 = call <4 x i64> @llvm.llrint.v4i64.v4f64(<4 x double> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %V8F64 = call <8 x i64> @llvm.llrint.v8i64.v8f64(<8 x double> undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
;
%F32 = call i64 @llvm.llrint.i64.f32(float undef)
%V4F32 = call <4 x i64> @llvm.llrint.v4i64.v4f32(<4 x float> undef)
Expand Down
Loading