[llvm] [RISCV][TTI] Adjust VLS shuffle costing to account for sub-mask reuse (PR #129793)
Philip Reames via llvm-commits
llvm-commits at lists.llvm.org
Wed Mar 19 13:48:34 PDT 2025
https://github.com/preames updated https://github.com/llvm/llvm-project/pull/129793
>From 6e19a9891d1968337a58ca5f5c3de98a9b3443f1 Mon Sep 17 00:00:00 2001
From: Philip Reames <preames at rivosinc.com>
Date: Tue, 4 Mar 2025 14:46:25 -0800
Subject: [PATCH 1/2] [RISCV][TTI] Adjust VLS shuffle costing to account for
sub-mask reuse
If we have a shuffle which can be split via VLA where two or more of
the destinations have exactly the same elements, then we only need
to account for them once in costing. The duplicate copes are are
(at worst) whole register moves.
Note that this change only handles the single source case. Doing the
multiple source case seemed a bit more complicated, and I didn't have
a motivating test case.
---
llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp | 3 +++
llvm/test/Analysis/CostModel/RISCV/shuffle-exact-vlen.ll | 8 ++++----
2 files changed, 7 insertions(+), 4 deletions(-)
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 11a658758a9cb..6871e4b843b31 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -430,11 +430,14 @@ costShuffleViaVRegSplitting(RISCVTTIImpl &TTI, MVT LegalVT,
copy(Mask, NormalizedMask.begin());
InstructionCost Cost = 0;
int NumShuffles = 0;
+ DenseSet<std::pair<ArrayRef<int>, unsigned>> ReusedSingleSrcShuffles;
processShuffleMasks(
NormalizedMask, NumOfSrcRegs, NumOfDestRegs, NumOfDestRegs, []() {},
[&](ArrayRef<int> RegMask, unsigned SrcReg, unsigned DestReg) {
if (ShuffleVectorInst::isIdentityMask(RegMask, RegMask.size()))
return;
+ if (!ReusedSingleSrcShuffles.insert(std::make_pair(RegMask, SrcReg)).second)
+ return;
++NumShuffles;
Cost += TTI.getShuffleCost(TTI::SK_PermuteSingleSrc, SingleOpTy,
RegMask, CostKind, 0, nullptr);
diff --git a/llvm/test/Analysis/CostModel/RISCV/shuffle-exact-vlen.ll b/llvm/test/Analysis/CostModel/RISCV/shuffle-exact-vlen.ll
index 06c709e4cc879..23d5999237e30 100644
--- a/llvm/test/Analysis/CostModel/RISCV/shuffle-exact-vlen.ll
+++ b/llvm/test/Analysis/CostModel/RISCV/shuffle-exact-vlen.ll
@@ -42,7 +42,7 @@ define void @shuffle() vscale_range(2,2) {
; CHECK-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v11 = shufflevector <2 x i16> poison, <2 x i16> poison, <2 x i32> <i32 1, i32 0>
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v12 = shufflevector <4 x i16> poison, <4 x i16> poison, <4 x i32> <i32 1, i32 3, i32 2, i32 0>
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v13 = shufflevector <8 x i16> poison, <8 x i16> poison, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
-; CHECK-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v10b = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
+; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v10b = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
; CHECK-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v14 = shufflevector <2 x i32> poison, <2 x i32> poison, <2 x i32> <i32 1, i32 0>
; CHECK-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v15 = shufflevector <4 x i32> poison, <4 x i32> poison, <4 x i32> <i32 1, i32 3, i32 2, i32 0>
; CHECK-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v16 = shufflevector <2 x float> poison, <2 x float> poison, <2 x i32> <i32 1, i32 0>
@@ -58,7 +58,7 @@ define void @shuffle() vscale_range(2,2) {
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v11 = shufflevector <2 x i16> poison, <2 x i16> poison, <2 x i32> <i32 1, i32 0>
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v12 = shufflevector <4 x i16> poison, <4 x i16> poison, <4 x i32> <i32 1, i32 3, i32 2, i32 0>
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v13 = shufflevector <8 x i16> poison, <8 x i16> poison, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
-; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 8 for instruction: %v10b = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
+; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v10b = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v14 = shufflevector <2 x i32> poison, <2 x i32> poison, <2 x i32> <i32 1, i32 0>
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 4 for instruction: %v15 = shufflevector <4 x i32> poison, <4 x i32> poison, <4 x i32> <i32 1, i32 3, i32 2, i32 0>
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %v16 = shufflevector <2 x float> poison, <2 x float> poison, <2 x i32> <i32 1, i32 0>
@@ -738,7 +738,7 @@ define void @multipart() vscale_range(2,2) {
; CHECK-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %v16c = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
; CHECK-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %v16d = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
; CHECK-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v32a = shufflevector <4 x i32> poison, <4 x i32> poison, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-; CHECK-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %v32a4 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
+; CHECK-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v32a4 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
; CHECK-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v32idrev = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 15, i32 14, i32 13, i32 12, i32 16, i32 17, i32 18, i32 19, i32 31, i32 30, i32 29, i32 28>
; CHECK-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %v32many = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28, i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 26, i32 30>
; CHECK-NEXT: Cost Model: Found an estimated cost of 47 for instruction: %v32many2 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 1, i32 4, i32 8, i32 12, i32 17, i32 20, i32 24, i32 28, i32 2, i32 6, i32 11, i32 14, i32 18, i32 22, i32 27, i32 30>
@@ -758,7 +758,7 @@ define void @multipart() vscale_range(2,2) {
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %v16c = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 12 for instruction: %v16d = shufflevector <16 x i16> poison, <16 x i16> poison, <16 x i32> <i32 0, i32 16, i32 1, i32 17, i32 2, i32 18, i32 3, i32 19, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v32a = shufflevector <4 x i32> poison, <4 x i32> poison, <4 x i32> <i32 2, i32 3, i32 0, i32 1>
-; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 20 for instruction: %v32a4 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
+; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %v32a4 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 1>
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 6 for instruction: %v32idrev = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 15, i32 14, i32 13, i32 12, i32 16, i32 17, i32 18, i32 19, i32 31, i32 30, i32 29, i32 28>
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %v32many = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 0, i32 4, i32 8, i32 12, i32 16, i32 20, i32 24, i32 28, i32 2, i32 6, i32 10, i32 14, i32 18, i32 22, i32 26, i32 30>
; CHECK-SIZE-NEXT: Cost Model: Found an estimated cost of 11 for instruction: %v32many2 = shufflevector <16 x i32> poison, <16 x i32> poison, <16 x i32> <i32 1, i32 4, i32 8, i32 12, i32 17, i32 20, i32 24, i32 28, i32 2, i32 6, i32 11, i32 14, i32 18, i32 22, i32 27, i32 30>
>From 446e1de4add9335adcb40ca00cf2042dde865ff7 Mon Sep 17 00:00:00 2001
From: Philip Reames <preames at rivosinc.com>
Date: Wed, 19 Mar 2025 13:38:44 -0700
Subject: [PATCH 2/2] Address review comments
---
llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
index 3c2efc5bfbd0c..8d1fce43d1ecd 100644
--- a/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
+++ b/llvm/lib/Target/RISCV/RISCVTargetTransformInfo.cpp
@@ -430,13 +430,14 @@ costShuffleViaVRegSplitting(RISCVTTIImpl &TTI, MVT LegalVT,
copy(Mask, NormalizedMask.begin());
InstructionCost Cost = 0;
int NumShuffles = 0;
- DenseSet<std::pair<ArrayRef<int>, unsigned>> ReusedSingleSrcShuffles;
+ SmallDenseSet<std::pair<ArrayRef<int>, unsigned>> ReusedSingleSrcShuffles;
processShuffleMasks(
NormalizedMask, NumOfSrcRegs, NumOfDestRegs, NumOfDestRegs, []() {},
[&](ArrayRef<int> RegMask, unsigned SrcReg, unsigned DestReg) {
if (ShuffleVectorInst::isIdentityMask(RegMask, RegMask.size()))
return;
- if (!ReusedSingleSrcShuffles.insert(std::make_pair(RegMask, SrcReg)).second)
+ if (!ReusedSingleSrcShuffles.insert(std::make_pair(RegMask, SrcReg))
+ .second)
return;
++NumShuffles;
Cost += TTI.getShuffleCost(TTI::SK_PermuteSingleSrc, SingleOpTy,
More information about the llvm-commits
mailing list