Skip to content

[CostModel][X86] Attempt to match cheap v4f32 shuffles that map to SHUFPS instruction #121778

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
Jan 6, 2025
Merged

[CostModel][X86] Attempt to match cheap v4f32 shuffles that map to SHUFPS instruction #121778

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
11 changes: 10 additions & 1 deletion llvm/lib/Target/X86/X86TargetTransformInfo.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2226,9 +2226,18 @@ InstructionCost X86TTIImpl::getShuffleCost(
{ TTI::SK_PermuteTwoSrc, MVT::v4f32, 2 }, // 2*shufps
};

if (ST->hasSSE1())
if (ST->hasSSE1()) {
if (LT.first == 1 && LT.second == MVT::v4f32 && Mask.size() == 4) {
Copy link
Member

@alexey-bataev alexey-bataev Jan 6, 2025 •
edited
Loading

Choose a reason for hiding this comment

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

What if BaseTp is not <4 x f32> and it is just promoted to MVT::v4f32, can it happen?

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

Not for any element type other than f32, which is why I added the Mask.size check - I did consider supporting v2f32/v3f32 handling by widening the mask with PoisonMaskElem but didn't find much need for it - wdyt?

// SHUFPS: both pairs must come from the same source register.
auto MatchSHUFPS = [](int X, int Y) {
return X < 0 || Y < 0 || ((X & 4) == (Y & 4));
};
if (MatchSHUFPS(Mask[0], Mask[1]) && MatchSHUFPS(Mask[2], Mask[3]))
return 1;
}
if (const auto *Entry = CostTableLookup(SSE1ShuffleTbl, Kind, LT.second))
return LT.first * Entry->Cost;
}

return BaseT::getShuffleCost(Kind, BaseTp, Mask, CostKind, Index, SubTp);
}
Expand Down
18 changes: 3 additions & 15 deletions llvm/test/Analysis/CostModel/X86/alternate-shuffle-cost.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -247,21 +247,9 @@ define <4 x float> @test_v4f32_2(<4 x float> %a, <4 x float> %b) {
}

define <4 x float> @test_v4f32_3(<4 x float> %a, <4 x float> %b) {
; SSE2-LABEL: 'test_v4f32_3'
; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %1
;
; SSSE3-LABEL: 'test_v4f32_3'
; SSSE3-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
; SSSE3-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %1
;
; SSE42-LABEL: 'test_v4f32_3'
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
; SSE42-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %1
;
; AVX-LABEL: 'test_v4f32_3'
; AVX-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
; AVX-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %1
; CHECK-LABEL: 'test_v4f32_3'
; CHECK-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
; CHECK-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret <4 x float> %1
;
%1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
ret <4 x float> %1
Expand Down
16 changes: 8 additions & 8 deletions llvm/test/Analysis/CostModel/X86/reduction.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -926,8 +926,8 @@ define fastcc double @pairwise_reduction4double(<4 x double> %rdx, double %f1) {

define fastcc float @pairwise_reduction8float(<8 x float> %rdx, float %f1) {
; SSE2-LABEL: 'pairwise_reduction8float'
; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.0.0 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 poison, i32 poison, i32 poison>
; SSE2-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.0.1 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 poison, i32 poison, i32 poison>
; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.0.0 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 poison, i32 poison, i32 poison>
; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.0.1 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 poison, i32 poison, i32 poison>
; SSE2-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx = fadd <8 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.0 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
; SSE2-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
Expand All @@ -939,8 +939,8 @@ define fastcc float @pairwise_reduction8float(<8 x float> %rdx, float %f1) {
; SSE2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret float %r
;
; SSSE3-LABEL: 'pairwise_reduction8float'
; SSSE3-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.0.0 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 poison, i32 poison, i32 poison>
; SSSE3-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.0.1 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 poison, i32 poison, i32 poison>
; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.0.0 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 poison, i32 poison, i32 poison>
; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.0.1 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 poison, i32 poison, i32 poison>
; SSSE3-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %bin.rdx = fadd <8 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.0 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
; SSSE3-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
Expand All @@ -952,8 +952,8 @@ define fastcc float @pairwise_reduction8float(<8 x float> %rdx, float %f1) {
; SSSE3-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret float %r
;
; SSE42-LABEL: 'pairwise_reduction8float'
; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.0.0 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 poison, i32 poison, i32 poison>
; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.0.1 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 poison, i32 poison, i32 poison>
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.0.0 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 poison, i32 poison, i32 poison>
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.0.1 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 poison, i32 poison, i32 poison>
; SSE42-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = fadd <8 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.0 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
; SSE42-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
Expand Down Expand Up @@ -991,8 +991,8 @@ define fastcc float @pairwise_reduction8float(<8 x float> %rdx, float %f1) {
; AVX2-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret float %r
;
; SLM-LABEL: 'pairwise_reduction8float'
; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.0.0 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 poison, i32 poison, i32 poison>
; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %rdx.shuf.0.1 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 poison, i32 poison, i32 poison>
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.0.0 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 4, i32 6, i32 poison, i32 poison, i32 poison, i32 poison>
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.0.1 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 5, i32 7, i32 poison, i32 poison, i32 poison, i32 poison>
; SLM-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %bin.rdx = fadd <8 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.0 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
; SLM-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %rdx.shuf.1.1 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison, i32 poison>
Expand Down
Loading
Loading