[llvm] [AArch64] Add costs for LD3/LD4 shuffles. (PR #89268)

via llvm-commits llvm-commits at lists.llvm.org
Thu Apr 18 10:26:06 PDT 2024


llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-llvm-analysis

Author: David Green (davemgreen)

<details>
<summary>Changes</summary>

Similar to #<!-- -->87934, this adds costs to the shuffles in a canonical LD3/LD4 pattern, which are represented in LLVM as deinterleaving-shuffle(load). This likely has less effect at the moment than the ST3/ST4 costs as instcombine will perform certain transforms without considering the cost.

---

Patch is 66.26 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/89268.diff


3 Files Affected:

- (modified) llvm/include/llvm/Analysis/TargetTransformInfoImpl.h (+2-2) 
- (modified) llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp (+31-1) 
- (modified) llvm/test/Analysis/CostModel/AArch64/shuffle-load.ll (+116-116) 


``````````diff
diff --git a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
index 5b40e49714069f..4d5cd963e09267 100644
--- a/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
+++ b/llvm/include/llvm/Analysis/TargetTransformInfoImpl.h
@@ -1376,7 +1376,7 @@ class TargetTransformInfoImplCRTPBase : public TargetTransformInfoImplBase {
 
           return TargetTTI->getShuffleCost(
               IsUnary ? TTI::SK_PermuteSingleSrc : TTI::SK_PermuteTwoSrc, VecTy,
-              AdjustMask, CostKind, 0, nullptr, {}, Shuffle);
+              AdjustMask, CostKind, 0, nullptr, Operands, Shuffle);
         }
 
         // Narrowing shuffle - perform shuffle at original wider width and
@@ -1385,7 +1385,7 @@ class TargetTransformInfoImplCRTPBase : public TargetTransformInfoImplBase {
 
         InstructionCost ShuffleCost = TargetTTI->getShuffleCost(
             IsUnary ? TTI::SK_PermuteSingleSrc : TTI::SK_PermuteTwoSrc,
-            VecSrcTy, AdjustMask, CostKind, 0, nullptr, {}, Shuffle);
+            VecSrcTy, AdjustMask, CostKind, 0, nullptr, Operands, Shuffle);
 
         SmallVector<int, 16> ExtractMask(Mask.size());
         std::iota(ExtractMask.begin(), ExtractMask.end(), 0);
diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
index e80931a03f30b6..b4d8e2b5fe4997 100644
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -3815,6 +3815,27 @@ InstructionCost AArch64TTIImpl::getSpliceCost(VectorType *Tp, int Index) {
   return LegalizationCost * LT.first;
 }
 
+/// Check if the mask is a DE-interleave mask of the given factor
+/// \p Factor like:
+///     <Index, Index+Factor, ..., Index+(NumElts-1)*Factor>
+static bool isDeInterleaveMaskOfFactor(ArrayRef<int> Mask, unsigned Factor) {
+  // Check all potential start indices from 0 to (Factor - 1).
+  for (unsigned Index = 0; Index < Factor; Index++) {
+    unsigned i = 0;
+
+    // Check that elements are in ascending order by Factor. Ignore undef
+    // elements.
+    for (; i < Mask.size(); i++)
+      if (Mask[i] >= 0 && static_cast<unsigned>(Mask[i]) != Index + i * Factor)
+        break;
+
+    if (i == Mask.size())
+      return true;
+  }
+
+  return false;
+}
+
 InstructionCost AArch64TTIImpl::getShuffleCost(
     TTI::ShuffleKind Kind, VectorType *Tp, ArrayRef<int> Mask,
     TTI::TargetCostKind CostKind, int Index, VectorType *SubTp,
@@ -3827,9 +3848,18 @@ InstructionCost AArch64TTIImpl::getShuffleCost(
       Tp->getScalarSizeInBits() == LT.second.getScalarSizeInBits() &&
       Mask.size() > LT.second.getVectorNumElements() && !Index && !SubTp) {
 
+    // Check for LD3/LD4 instructions, which are represented in llvm IR as
+    // deinterleaving-shuffle(load). The shuffle cost could potentially be free,
+    // but we model it with a cost of LT.first so that LD3/LD4 have a higher
+    // cost than just the load.
+    if (Args.size() >= 1 && isa<LoadInst>(Args[0]) &&
+        (isDeInterleaveMaskOfFactor(Mask, 3) ||
+         isDeInterleaveMaskOfFactor(Mask, 4)))
+      return std::max<InstructionCost>(1, LT.first / 4);
+
     // Check for ST3/ST4 instructions, which are represented in llvm IR as
     // store(interleaving-shuffle). The shuffle cost could potentially be free,
-    // but we model it with a cost of LT.first so that LD3/LD3 have a higher
+    // but we model it with a cost of LT.first so that ST3/ST4 have a higher
     // cost than just the store.
     if (CxtI && CxtI->hasOneUse() && isa<StoreInst>(*CxtI->user_begin()) &&
         (ShuffleVectorInst::isInterleaveMask(
diff --git a/llvm/test/Analysis/CostModel/AArch64/shuffle-load.ll b/llvm/test/Analysis/CostModel/AArch64/shuffle-load.ll
index 106f2f9edc2e43..bf81d7aa5e6893 100644
--- a/llvm/test/Analysis/CostModel/AArch64/shuffle-load.ll
+++ b/llvm/test/Analysis/CostModel/AArch64/shuffle-load.ll
@@ -572,45 +572,45 @@ define void @vld3(ptr %p) {
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v4i8_1 = shufflevector <12 x i8> %v4i8, <12 x i8> undef, <4 x i32> <i32 1, i32 4, i32 7, i32 10>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v4i8_2 = shufflevector <12 x i8> %v4i8, <12 x i8> undef, <4 x i32> <i32 2, i32 5, i32 8, i32 11>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8 = load <24 x i8>, ptr %p, align 32
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %v8i8_0 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %v8i8_1 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %v8i8_2 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_1 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_2 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i8 = load <48 x i8>, ptr %p, align 64
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v16i8_0 = shufflevector <48 x i8> %v16i8, <48 x i8> undef, <16 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21, i32 24, i32 27, i32 30, i32 33, i32 36, i32 39, i32 42, i32 45>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v16i8_1 = shufflevector <48 x i8> %v16i8, <48 x i8> undef, <16 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22, i32 25, i32 28, i32 31, i32 34, i32 37, i32 40, i32 43, i32 46>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v16i8_2 = shufflevector <48 x i8> %v16i8, <48 x i8> undef, <16 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23, i32 26, i32 29, i32 32, i32 35, i32 38, i32 41, i32 44, i32 47>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_0 = shufflevector <48 x i8> %v16i8, <48 x i8> undef, <16 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21, i32 24, i32 27, i32 30, i32 33, i32 36, i32 39, i32 42, i32 45>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_1 = shufflevector <48 x i8> %v16i8, <48 x i8> undef, <16 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22, i32 25, i32 28, i32 31, i32 34, i32 37, i32 40, i32 43, i32 46>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16i8_2 = shufflevector <48 x i8> %v16i8, <48 x i8> undef, <16 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23, i32 26, i32 29, i32 32, i32 35, i32 38, i32 41, i32 44, i32 47>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i16 = load <6 x i16>, ptr %p, align 16
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v2i16_0 = shufflevector <6 x i16> %v2i16, <6 x i16> undef, <2 x i32> <i32 0, i32 3>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v2i16_1 = shufflevector <6 x i16> %v2i16, <6 x i16> undef, <2 x i32> <i32 1, i32 4>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v2i16_2 = shufflevector <6 x i16> %v2i16, <6 x i16> undef, <2 x i32> <i32 2, i32 5>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4i16 = load <12 x i16>, ptr %p, align 32
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v4i16_0 = shufflevector <12 x i16> %v4i16, <12 x i16> undef, <4 x i32> <i32 0, i32 3, i32 6, i32 9>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v4i16_1 = shufflevector <12 x i16> %v4i16, <12 x i16> undef, <4 x i32> <i32 1, i32 4, i32 7, i32 10>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 28 for instruction: %v4i16_2 = shufflevector <12 x i16> %v4i16, <12 x i16> undef, <4 x i32> <i32 2, i32 5, i32 8, i32 11>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_0 = shufflevector <12 x i16> %v4i16, <12 x i16> undef, <4 x i32> <i32 0, i32 3, i32 6, i32 9>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_1 = shufflevector <12 x i16> %v4i16, <12 x i16> undef, <4 x i32> <i32 1, i32 4, i32 7, i32 10>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i16_2 = shufflevector <12 x i16> %v4i16, <12 x i16> undef, <4 x i32> <i32 2, i32 5, i32 8, i32 11>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v8i16 = load <24 x i16>, ptr %p, align 64
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v8i16_0 = shufflevector <24 x i16> %v8i16, <24 x i16> undef, <8 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i16_1 = shufflevector <24 x i16> %v8i16, <24 x i16> undef, <8 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %v8i16_2 = shufflevector <24 x i16> %v8i16, <24 x i16> undef, <8 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_0 = shufflevector <24 x i16> %v8i16, <24 x i16> undef, <8 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_1 = shufflevector <24 x i16> %v8i16, <24 x i16> undef, <8 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i16_2 = shufflevector <24 x i16> %v8i16, <24 x i16> undef, <8 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v16i16 = load <48 x i16>, ptr %p, align 128
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i16_0 = shufflevector <48 x i16> %v16i16, <48 x i16> undef, <16 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21, i32 24, i32 27, i32 30, i32 33, i32 36, i32 39, i32 42, i32 45>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v16i16_1 = shufflevector <48 x i16> %v16i16, <48 x i16> undef, <16 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22, i32 25, i32 28, i32 31, i32 34, i32 37, i32 40, i32 43, i32 46>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %v16i16_2 = shufflevector <48 x i16> %v16i16, <48 x i16> undef, <16 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23, i32 26, i32 29, i32 32, i32 35, i32 38, i32 41, i32 44, i32 47>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_0 = shufflevector <48 x i16> %v16i16, <48 x i16> undef, <16 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21, i32 24, i32 27, i32 30, i32 33, i32 36, i32 39, i32 42, i32 45>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_1 = shufflevector <48 x i16> %v16i16, <48 x i16> undef, <16 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22, i32 25, i32 28, i32 31, i32 34, i32 37, i32 40, i32 43, i32 46>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16i16_2 = shufflevector <48 x i16> %v16i16, <48 x i16> undef, <16 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23, i32 26, i32 29, i32 32, i32 35, i32 38, i32 41, i32 44, i32 47>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32 = load <6 x i32>, ptr %p, align 32
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_0 = shufflevector <6 x i32> %v2i32, <6 x i32> undef, <2 x i32> <i32 0, i32 3>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2i32_1 = shufflevector <6 x i32> %v2i32, <6 x i32> undef, <2 x i32> <i32 1, i32 4>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_1 = shufflevector <6 x i32> %v2i32, <6 x i32> undef, <2 x i32> <i32 1, i32 4>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i32_2 = shufflevector <6 x i32> %v2i32, <6 x i32> undef, <2 x i32> <i32 2, i32 5>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v4i32 = load <12 x i32>, ptr %p, align 64
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_0 = shufflevector <12 x i32> %v4i32, <12 x i32> undef, <4 x i32> <i32 0, i32 3, i32 6, i32 9>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v4i32_1 = shufflevector <12 x i32> %v4i32, <12 x i32> undef, <4 x i32> <i32 1, i32 4, i32 7, i32 10>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4i32_2 = shufflevector <12 x i32> %v4i32, <12 x i32> undef, <4 x i32> <i32 2, i32 5, i32 8, i32 11>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_0 = shufflevector <12 x i32> %v4i32, <12 x i32> undef, <4 x i32> <i32 0, i32 3, i32 6, i32 9>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_1 = shufflevector <12 x i32> %v4i32, <12 x i32> undef, <4 x i32> <i32 1, i32 4, i32 7, i32 10>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4i32_2 = shufflevector <12 x i32> %v4i32, <12 x i32> undef, <4 x i32> <i32 2, i32 5, i32 8, i32 11>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i32 = load <24 x i32>, ptr %p, align 128
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v8i32_0 = shufflevector <24 x i32> %v8i32, <24 x i32> undef, <8 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8i32_1 = shufflevector <24 x i32> %v8i32, <24 x i32> undef, <8 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v8i32_2 = shufflevector <24 x i32> %v8i32, <24 x i32> undef, <8 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_0 = shufflevector <24 x i32> %v8i32, <24 x i32> undef, <8 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_1 = shufflevector <24 x i32> %v8i32, <24 x i32> undef, <8 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i32_2 = shufflevector <24 x i32> %v8i32, <24 x i32> undef, <8 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v16i32 = load <48 x i32>, ptr %p, align 256
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v16i32_0 = shufflevector <48 x i32> %v16i32, <48 x i32> undef, <16 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21, i32 24, i32 27, i32 30, i32 33, i32 36, i32 39, i32 42, i32 45>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v16i32_1 = shufflevector <48 x i32> %v16i32, <48 x i32> undef, <16 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22, i32 25, i32 28, i32 31, i32 34, i32 37, i32 40, i32 43, i32 46>
-; CHECK-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v16i32_2 = shufflevector <48 x i32> %v16i32, <48 x i32> undef, <16 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23, i32 26, i32 29, i32 32, i32 35, i32 38, i32 41, i32 44, i32 47>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i32_0 = shufflevector <48 x i32> %v16i32, <48 x i32> undef, <16 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21, i32 24, i32 27, i32 30, i32 33, i32 36, i32 39, i32 42, i32 45>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i32_1 = shufflevector <48 x i32> %v16i32, <48 x i32> undef, <16 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22, i32 25, i32 28, i32 31, i32 34, i32 37, i32 40, i32 43, i32 46>
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i32_2 = shufflevector <48 x i32> %v16i32, <48 x i32> undef, <16 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23, i32 26, i32 29, i32 32, i32 35, i32 38, i32 41, i32 44, i32 47>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v2i64 = load <6 x i64>, ptr %p, align 64
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_0 = shufflevector <6 x i64> %v2i64, <6 x i64> undef, <2 x i32> <i32 0, i32 3>
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2i64_1 = shufflevector <6 x i64> %v2i64, <6 x i64> undef, <2 x i32> <i32 1, i32 4>
@@ -639,45 +639,45 @@ define void @vld3(ptr %p) {
 ; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v4i8_1 = shufflevector <12 x i8> %v4i8, <12 x i8> undef, <4 x i32> <i32 1, i32 4, i32 7, i32 10>
 ; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v4i8_2 = shufflevector <12 x i8> %v4i8, <12 x i8> undef, <4 x i32> <i32 2, i32 5, i32 8, i32 11>
 ; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8i8 = load <24 x i8>, ptr %p, align 32
-; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %v8i8_0 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21>
-; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %v8i8_1 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22>
-; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %v8i8_2 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23>
+; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_0 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21>
+; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_1 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22>
+; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8i8_2 = shufflevector <24 x i8> %v8i8, <24 x i8> undef, <8 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20, i32 23>
 ; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v16i8 = load <48 x i8>, ptr %p, align 64
-; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v16i8_0 = shufflevector <48 x i8> %v16i8, <48 x i8> undef, <16 x i32> <i32 0, i32 3, i32 6, i32 9, i32 12, i32 15, i32 18, i32 21, i32 24, i32 27, i32 30, i32 33, i32 36, i32 39, i32 42, i32 45>
-; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v16i8_1 = shufflevector <48 x i8> %v16i8, <48 x i8> undef, <16 x i32> <i32 1, i32 4, i32 7, i32 10, i32 13, i32 16, i32 19, i32 22, i32 25, i32 28, i32 31, i32 34, i32 37, i32 40, i32 43, i32 46>
-; CODESIZE-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %v16i8_2 = shufflevector <48 x i8> %v16i8, <48 x i8> undef, <16 x i32> <i32 2, i32 5, i32 8, i32 11, i32 14, i32 17, i32 20...
[truncated]

``````````

</details>


https://github.com/llvm/llvm-project/pull/89268


More information about the llvm-commits mailing list