Skip to content

[RISCV][TTI] Reduce cost of a <N x i1> build_vector pattern #109449

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 2 commits into from
Sep 23, 2024
Merged

[RISCV][TTI] Reduce cost of a <N x i1> build_vector pattern #109449

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
bdd4441
[RISCV][TTI] Reduce cost of a <N x i1> build_vector pattern
preames Sep 20, 2024
b75bde1
clang-format and style fixup afterwards
preames Sep 20, 2024
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
13 changes: 11 additions & 2 deletions llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -635,8 +635,17 @@ InstructionCost RISCVTTIImpl::getScalarizationOverhead(
InstructionCost Cost = BaseT::getScalarizationOverhead(
Ty, DemandedElts, Insert, Extract, CostKind);
std::pair<InstructionCost, MVT> LT = getTypeLegalizationCost(Ty);
if (Insert && !Extract && LT.first.isValid() && LT.second.isVector() &&
Ty->getScalarSizeInBits() != 1) {
if (Insert && !Extract && LT.first.isValid() && LT.second.isVector()) {
if (Ty->getScalarSizeInBits() == 1) {
auto *WideVecTy = cast<VectorType>(Ty->getWithNewBitWidth(8));
// Note: Implicit scalar anyextend is assumed to be free since the i1
// must be stored in a GPR.
return getScalarizationOverhead(WideVecTy, DemandedElts, Insert, Extract,
CostKind) +
getCastInstrCost(Instruction::Trunc, Ty, WideVecTy,
TTI::CastContextHint::None, CostKind, nullptr);
}

assert(LT.second.isFixedLengthVector());
MVT ContainerVT = TLI->getContainerForFixedLengthVector(LT.second);
if (isM1OrSmaller(ContainerVT)) {
Expand Down
16 changes: 8 additions & 8 deletions llvm/test/Analysis/CostModel/RISCV/rvv-intrinsics.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1029,10 +1029,10 @@ define void @store() {

define void @strided_load() {
; CHECK-LABEL: 'strided_load'
; CHECK-NEXT: Cost Model: Found an estimated cost of 16 for instruction: %ti1_2 = call <2 x i1> @llvm.experimental.vp.strided.load.v2i1.p0.i64(ptr undef, i64 undef, <2 x i1> undef, i32 undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 34 for instruction: %ti1_4 = call <4 x i1> @llvm.experimental.vp.strided.load.v4i1.p0.i64(ptr undef, i64 undef, <4 x i1> undef, i32 undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 70 for instruction: %ti1_8 = call <8 x i1> @llvm.experimental.vp.strided.load.v8i1.p0.i64(ptr undef, i64 undef, <8 x i1> undef, i32 undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 142 for instruction: %ti1_16 = call <16 x i1> @llvm.experimental.vp.strided.load.v16i1.p0.i64(ptr undef, i64 undef, <16 x i1> undef, i32 undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %ti1_2 = call <2 x i1> @llvm.experimental.vp.strided.load.v2i1.p0.i64(ptr undef, i64 undef, <2 x i1> undef, i32 undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 17 for instruction: %ti1_4 = call <4 x i1> @llvm.experimental.vp.strided.load.v4i1.p0.i64(ptr undef, i64 undef, <4 x i1> undef, i32 undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 33 for instruction: %ti1_8 = call <8 x i1> @llvm.experimental.vp.strided.load.v8i1.p0.i64(ptr undef, i64 undef, <8 x i1> undef, i32 undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 65 for instruction: %ti1_16 = call <16 x i1> @llvm.experimental.vp.strided.load.v16i1.p0.i64(ptr undef, i64 undef, <16 x i1> undef, i32 undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %t0 = call <2 x i8> @llvm.experimental.vp.strided.load.v2i8.p0.i64(ptr undef, i64 undef, <2 x i1> undef, i32 undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %t2 = call <4 x i8> @llvm.experimental.vp.strided.load.v4i8.p0.i64(ptr undef, i64 undef, <4 x i1> undef, i32 undef)
; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %t4 = call <8 x i8> @llvm.experimental.vp.strided.load.v8i8.p0.i64(ptr undef, i64 undef, <8 x i1> undef, i32 undef)
Expand All @@ -1056,10 +1056,10 @@ define void @strided_load() {
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; TYPEBASED-LABEL: 'strided_load'
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %ti1_2 = call <2 x i1> @llvm.experimental.vp.strided.load.v2i1.p0.i64(ptr undef, i64 undef, <2 x i1> undef, i32 undef)
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 42 for instruction: %ti1_4 = call <4 x i1> @llvm.experimental.vp.strided.load.v4i1.p0.i64(ptr undef, i64 undef, <4 x i1> undef, i32 undef)
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 86 for instruction: %ti1_8 = call <8 x i1> @llvm.experimental.vp.strided.load.v8i1.p0.i64(ptr undef, i64 undef, <8 x i1> undef, i32 undef)
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 174 for instruction: %ti1_16 = call <16 x i1> @llvm.experimental.vp.strided.load.v16i1.p0.i64(ptr undef, i64 undef, <16 x i1> undef, i32 undef)
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 13 for instruction: %ti1_2 = call <2 x i1> @llvm.experimental.vp.strided.load.v2i1.p0.i64(ptr undef, i64 undef, <2 x i1> undef, i32 undef)
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 25 for instruction: %ti1_4 = call <4 x i1> @llvm.experimental.vp.strided.load.v4i1.p0.i64(ptr undef, i64 undef, <4 x i1> undef, i32 undef)
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 49 for instruction: %ti1_8 = call <8 x i1> @llvm.experimental.vp.strided.load.v8i1.p0.i64(ptr undef, i64 undef, <8 x i1> undef, i32 undef)
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 97 for instruction: %ti1_16 = call <16 x i1> @llvm.experimental.vp.strided.load.v16i1.p0.i64(ptr undef, i64 undef, <16 x i1> undef, i32 undef)
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %t0 = call <2 x i8> @llvm.experimental.vp.strided.load.v2i8.p0.i64(ptr undef, i64 undef, <2 x i1> undef, i32 undef)
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 23 for instruction: %t2 = call <4 x i8> @llvm.experimental.vp.strided.load.v4i8.p0.i64(ptr undef, i64 undef, <4 x i1> undef, i32 undef)
; TYPEBASED-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %t4 = call <8 x i8> @llvm.experimental.vp.strided.load.v8i8.p0.i64(ptr undef, i64 undef, <8 x i1> undef, i32 undef)
Expand Down
7 changes: 3 additions & 4 deletions llvm/test/Transforms/VectorCombine/RISCV/shuffle-of-intrinsics.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,10 +48,9 @@ entry:
define <8 x i1> @test4(<4 x float> %0, <4 x float> %1) {
; CHECK-LABEL: @test4(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP2:%.*]] = call <4 x i1> @llvm.is.fpclass.v4f32(<4 x float> [[TMP0:%.*]], i32 0)
; CHECK-NEXT: [[TMP3:%.*]] = call <4 x i1> @llvm.is.fpclass.v4f32(<4 x float> [[TMP1:%.*]], i32 0)
; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <4 x i1> [[TMP2]], <4 x i1> [[TMP3]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
; CHECK-NEXT: ret <8 x i1> [[TMP4]]
; CHECK-NEXT: [[TMP2:%.*]] = shufflevector <4 x float> [[TMP0:%.*]], <4 x float> [[TMP1:%.*]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
; CHECK-NEXT: [[TMP3:%.*]] = call <8 x i1> @llvm.is.fpclass.v8f32(<8 x float> [[TMP2]], i32 0)
; CHECK-NEXT: ret <8 x i1> [[TMP3]]
;
entry:
%2 = call <4 x i1> @llvm.is.fpclass.v4f32(<4 x float> %0, i32 0)
Expand Down
Loading