[llvm-branch-commits] [llvm] [LV] Reduce register usage for scaled reductions (PR #133090)
Sam Tebbs via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Wed Apr 2 09:03:38 PDT 2025
https://github.com/SamTebbs33 updated https://github.com/llvm/llvm-project/pull/133090
>From 9a9164fce2a7fe1d602fd24cf9a9026b06190f31 Mon Sep 17 00:00:00 2001
From: Samuel Tebbs <samuel.tebbs at arm.com>
Date: Wed, 26 Mar 2025 14:01:59 +0000
Subject: [PATCH 1/5] [LV] Reduce register usage for scaled reductions
---
.../Transforms/Vectorize/LoopVectorize.cpp | 24 +-
.../Transforms/Vectorize/VPRecipeBuilder.h | 3 +-
llvm/lib/Transforms/Vectorize/VPlan.h | 14 +-
.../partial-reduce-dot-product-neon.ll | 118 ++++--
.../AArch64/partial-reduce-dot-product.ll | 344 +++++++++---------
5 files changed, 282 insertions(+), 221 deletions(-)
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index 1dbcbdbe083fe..400a510be308b 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -5019,10 +5019,23 @@ calculateRegisterUsage(VPlan &Plan, ArrayRef<ElementCount> VFs,
// even in the scalar case.
RegUsage[ClassID] += 1;
} else {
+ // The output from scaled phis and scaled reductions actually have
+ // fewer lanes than the VF.
+ auto VF = VFs[J];
+ if (auto *ReductionR = dyn_cast<VPReductionPHIRecipe>(R))
+ VF = VF.divideCoefficientBy(ReductionR->getVFScaleFactor());
+ else if (auto *PartialReductionR =
+ dyn_cast<VPPartialReductionRecipe>(R))
+ VF = VF.divideCoefficientBy(PartialReductionR->getScaleFactor());
+ if (VF != VFs[J])
+ LLVM_DEBUG(dbgs() << "LV(REG): Scaled down VF from " << VFs[J]
+ << " to " << VF << " for ";
+ R->dump(););
+
for (VPValue *DefV : R->definedValues()) {
Type *ScalarTy = TypeInfo.inferScalarType(DefV);
unsigned ClassID = TTI.getRegisterClassForType(true, ScalarTy);
- RegUsage[ClassID] += GetRegUsage(ScalarTy, VFs[J]);
+ RegUsage[ClassID] += GetRegUsage(ScalarTy, VF);
}
}
}
@@ -8951,8 +8964,8 @@ VPRecipeBase *VPRecipeBuilder::tryToCreateWidenRecipe(
if (isa<LoadInst>(Instr) || isa<StoreInst>(Instr))
return tryToWidenMemory(Instr, Operands, Range);
- if (getScalingForReduction(Instr))
- return tryToCreatePartialReduction(Instr, Operands);
+ if (auto ScaleFactor = getScalingForReduction(Instr))
+ return tryToCreatePartialReduction(Instr, Operands, ScaleFactor.value());
if (!shouldWiden(Instr, Range))
return nullptr;
@@ -8976,7 +8989,8 @@ VPRecipeBase *VPRecipeBuilder::tryToCreateWidenRecipe(
VPRecipeBase *
VPRecipeBuilder::tryToCreatePartialReduction(Instruction *Reduction,
- ArrayRef<VPValue *> Operands) {
+ ArrayRef<VPValue *> Operands,
+ unsigned ScaleFactor) {
assert(Operands.size() == 2 &&
"Unexpected number of operands for partial reduction");
@@ -9009,7 +9023,7 @@ VPRecipeBuilder::tryToCreatePartialReduction(Instruction *Reduction,
BinOp = Builder.createSelect(Mask, BinOp, Zero, Reduction->getDebugLoc());
}
return new VPPartialReductionRecipe(ReductionOpcode, BinOp, Accumulator,
- Reduction);
+ ScaleFactor, Reduction);
}
void LoopVectorizationPlanner::buildVPlansWithVPRecipes(ElementCount MinVF,
diff --git a/llvm/lib/Transforms/Vectorize/VPRecipeBuilder.h b/llvm/lib/Transforms/Vectorize/VPRecipeBuilder.h
index 334cfbad8bd7c..fd0064a34c4c9 100644
--- a/llvm/lib/Transforms/Vectorize/VPRecipeBuilder.h
+++ b/llvm/lib/Transforms/Vectorize/VPRecipeBuilder.h
@@ -178,7 +178,8 @@ class VPRecipeBuilder {
/// Create and return a partial reduction recipe for a reduction instruction
/// along with binary operation and reduction phi operands.
VPRecipeBase *tryToCreatePartialReduction(Instruction *Reduction,
- ArrayRef<VPValue *> Operands);
+ ArrayRef<VPValue *> Operands,
+ unsigned ScaleFactor);
/// Set the recipe created for given ingredient.
void setRecipe(Instruction *I, VPRecipeBase *R) {
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
index 37e0a176ab1cc..376526e804b4b 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.h
+++ b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -2033,6 +2033,8 @@ class VPReductionPHIRecipe : public VPHeaderPHIRecipe,
/// Generate the phi/select nodes.
void execute(VPTransformState &State) override;
+ unsigned getVFScaleFactor() const { return VFScaleFactor; }
+
#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
/// Print the recipe.
void print(raw_ostream &O, const Twine &Indent,
@@ -2063,17 +2065,19 @@ class VPReductionPHIRecipe : public VPHeaderPHIRecipe,
/// scalar value.
class VPPartialReductionRecipe : public VPSingleDefRecipe {
unsigned Opcode;
+ unsigned ScaleFactor;
public:
VPPartialReductionRecipe(Instruction *ReductionInst, VPValue *Op0,
- VPValue *Op1)
+ VPValue *Op1, unsigned ScaleFactor)
: VPPartialReductionRecipe(ReductionInst->getOpcode(), Op0, Op1,
- ReductionInst) {}
+ ScaleFactor, ReductionInst) {}
VPPartialReductionRecipe(unsigned Opcode, VPValue *Op0, VPValue *Op1,
+ unsigned ScaleFactor,
Instruction *ReductionInst = nullptr)
: VPSingleDefRecipe(VPDef::VPPartialReductionSC,
ArrayRef<VPValue *>({Op0, Op1}), ReductionInst),
- Opcode(Opcode) {
+ Opcode(Opcode), ScaleFactor(ScaleFactor) {
[[maybe_unused]] auto *AccumulatorRecipe =
getOperand(1)->getDefiningRecipe();
assert((isa<VPReductionPHIRecipe>(AccumulatorRecipe) ||
@@ -2084,7 +2088,7 @@ class VPPartialReductionRecipe : public VPSingleDefRecipe {
VPPartialReductionRecipe *clone() override {
return new VPPartialReductionRecipe(Opcode, getOperand(0), getOperand(1),
- getUnderlyingInstr());
+ ScaleFactor, getUnderlyingInstr());
}
VP_CLASSOF_IMPL(VPDef::VPPartialReductionSC)
@@ -2099,6 +2103,8 @@ class VPPartialReductionRecipe : public VPSingleDefRecipe {
/// Get the binary op's opcode.
unsigned getOpcode() const { return Opcode; }
+ unsigned getScaleFactor() const { return ScaleFactor; }
+
#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
/// Print the recipe.
void print(raw_ostream &O, const Twine &Indent,
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product-neon.ll b/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product-neon.ll
index 341c84cebaa9a..a77447500d598 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product-neon.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product-neon.ll
@@ -770,10 +770,10 @@ define i32 @dotp_unrolled(i32 %num_out, i64 %num_in, ptr %a, ptr %b) {
; CHECK-INTERLEAVED-LABEL: define i32 @dotp_unrolled(
; CHECK-INTERLEAVED-SAME: i32 [[NUM_OUT:%.*]], i64 [[NUM_IN:%.*]], ptr [[A:%.*]], ptr [[B:%.*]]) #[[ATTR0]] {
; CHECK-INTERLEAVED-NEXT: entry:
-; CHECK-INTERLEAVED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[NUM_IN]], 16
+; CHECK-INTERLEAVED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[NUM_IN]], 32
; CHECK-INTERLEAVED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-INTERLEAVED: vector.ph:
-; CHECK-INTERLEAVED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[NUM_IN]], 16
+; CHECK-INTERLEAVED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[NUM_IN]], 32
; CHECK-INTERLEAVED-NEXT: [[N_VEC:%.*]] = sub i64 [[NUM_IN]], [[N_MOD_VF]]
; CHECK-INTERLEAVED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-INTERLEAVED: vector.body:
@@ -782,6 +782,10 @@ define i32 @dotp_unrolled(i32 %num_out, i64 %num_in, ptr %a, ptr %b) {
; CHECK-INTERLEAVED-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE10:%.*]], [[VECTOR_BODY]] ]
; CHECK-INTERLEAVED-NEXT: [[VEC_PHI2:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE7:%.*]], [[VECTOR_BODY]] ]
; CHECK-INTERLEAVED-NEXT: [[VEC_PHI3:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI4:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE16:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI5:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE17:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI6:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE1:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI7:%.*]] = phi <4 x i32> [ zeroinitializer, [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE11:%.*]], [[VECTOR_BODY]] ]
; CHECK-INTERLEAVED-NEXT: [[TMP1:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[INDEX]]
; CHECK-INTERLEAVED-NEXT: [[TMP2:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[INDEX]]
; CHECK-INTERLEAVED-NEXT: [[TMP3:%.*]] = or disjoint i64 [[INDEX]], 1
@@ -794,45 +798,81 @@ define i32 @dotp_unrolled(i32 %num_out, i64 %num_in, ptr %a, ptr %b) {
; CHECK-INTERLEAVED-NEXT: [[TMP10:%.*]] = getelementptr inbounds i8, ptr [[A]], i64 [[TMP9]]
; CHECK-INTERLEAVED-NEXT: [[TMP11:%.*]] = getelementptr inbounds i8, ptr [[B]], i64 [[TMP9]]
; CHECK-INTERLEAVED-NEXT: [[TMP12:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i32 0
+; CHECK-INTERLEAVED-NEXT: [[TMP38:%.*]] = getelementptr inbounds i8, ptr [[TMP1]], i32 16
; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP12]], align 1
-; CHECK-INTERLEAVED-NEXT: [[TMP36:%.*]] = sext <16 x i8> [[WIDE_LOAD]] to <16 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP13:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i32 0
-; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD4:%.*]] = load <16 x i8>, ptr [[TMP13]], align 1
-; CHECK-INTERLEAVED-NEXT: [[TMP14:%.*]] = sext <16 x i8> [[WIDE_LOAD4]] to <16 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP15:%.*]] = mul nsw <16 x i32> [[TMP14]], [[TMP36]]
-; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI3]], <16 x i32> [[TMP15]])
-; CHECK-INTERLEAVED-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i32 0
-; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD5:%.*]] = load <16 x i8>, ptr [[TMP16]], align 1
-; CHECK-INTERLEAVED-NEXT: [[TMP17:%.*]] = sext <16 x i8> [[WIDE_LOAD5]] to <16 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP18:%.*]] = getelementptr inbounds i8, ptr [[TMP5]], i32 0
-; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD6:%.*]] = load <16 x i8>, ptr [[TMP18]], align 1
-; CHECK-INTERLEAVED-NEXT: [[TMP19:%.*]] = sext <16 x i8> [[WIDE_LOAD6]] to <16 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP20:%.*]] = mul nsw <16 x i32> [[TMP17]], [[TMP19]]
-; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE7]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI2]], <16 x i32> [[TMP20]])
-; CHECK-INTERLEAVED-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[TMP7]], i32 0
-; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD8:%.*]] = load <16 x i8>, ptr [[TMP21]], align 1
-; CHECK-INTERLEAVED-NEXT: [[TMP22:%.*]] = sext <16 x i8> [[WIDE_LOAD8]] to <16 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP23:%.*]] = getelementptr inbounds i8, ptr [[TMP8]], i32 0
-; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD9:%.*]] = load <16 x i8>, ptr [[TMP23]], align 1
-; CHECK-INTERLEAVED-NEXT: [[TMP24:%.*]] = sext <16 x i8> [[WIDE_LOAD9]] to <16 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP25:%.*]] = mul nsw <16 x i32> [[TMP22]], [[TMP24]]
-; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE10]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI1]], <16 x i32> [[TMP25]])
-; CHECK-INTERLEAVED-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[TMP10]], i32 0
-; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD11:%.*]] = load <16 x i8>, ptr [[TMP26]], align 1
-; CHECK-INTERLEAVED-NEXT: [[TMP27:%.*]] = sext <16 x i8> [[WIDE_LOAD11]] to <16 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP28:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i32 0
-; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD12:%.*]] = load <16 x i8>, ptr [[TMP28]], align 1
-; CHECK-INTERLEAVED-NEXT: [[TMP29:%.*]] = sext <16 x i8> [[WIDE_LOAD12]] to <16 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP30:%.*]] = mul nsw <16 x i32> [[TMP27]], [[TMP29]]
-; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE13]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI]], <16 x i32> [[TMP30]])
-; CHECK-INTERLEAVED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
-; CHECK-INTERLEAVED-NEXT: [[TMP31:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-INTERLEAVED-NEXT: br i1 [[TMP31]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD8:%.*]] = load <16 x i8>, ptr [[TMP38]], align 1
+; CHECK-INTERLEAVED-NEXT: [[TMP13:%.*]] = sext <16 x i8> [[WIDE_LOAD]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP14:%.*]] = sext <16 x i8> [[WIDE_LOAD8]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP15:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i32 0
+; CHECK-INTERLEAVED-NEXT: [[TMP16:%.*]] = getelementptr inbounds i8, ptr [[TMP2]], i32 16
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD9:%.*]] = load <16 x i8>, ptr [[TMP15]], align 1
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD10:%.*]] = load <16 x i8>, ptr [[TMP16]], align 1
+; CHECK-INTERLEAVED-NEXT: [[TMP17:%.*]] = sext <16 x i8> [[WIDE_LOAD9]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP18:%.*]] = sext <16 x i8> [[WIDE_LOAD10]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP19:%.*]] = mul nsw <16 x i32> [[TMP17]], [[TMP13]]
+; CHECK-INTERLEAVED-NEXT: [[TMP20:%.*]] = mul nsw <16 x i32> [[TMP18]], [[TMP14]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE1]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI6]], <16 x i32> [[TMP19]])
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE11]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI7]], <16 x i32> [[TMP20]])
+; CHECK-INTERLEAVED-NEXT: [[TMP21:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i32 0
+; CHECK-INTERLEAVED-NEXT: [[TMP22:%.*]] = getelementptr inbounds i8, ptr [[TMP4]], i32 16
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD12:%.*]] = load <16 x i8>, ptr [[TMP21]], align 1
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD13:%.*]] = load <16 x i8>, ptr [[TMP22]], align 1
+; CHECK-INTERLEAVED-NEXT: [[TMP23:%.*]] = sext <16 x i8> [[WIDE_LOAD12]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP24:%.*]] = sext <16 x i8> [[WIDE_LOAD13]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP25:%.*]] = getelementptr inbounds i8, ptr [[TMP5]], i32 0
+; CHECK-INTERLEAVED-NEXT: [[TMP26:%.*]] = getelementptr inbounds i8, ptr [[TMP5]], i32 16
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD14:%.*]] = load <16 x i8>, ptr [[TMP25]], align 1
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD15:%.*]] = load <16 x i8>, ptr [[TMP26]], align 1
+; CHECK-INTERLEAVED-NEXT: [[TMP27:%.*]] = sext <16 x i8> [[WIDE_LOAD14]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP28:%.*]] = sext <16 x i8> [[WIDE_LOAD15]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP29:%.*]] = mul nsw <16 x i32> [[TMP23]], [[TMP27]]
+; CHECK-INTERLEAVED-NEXT: [[TMP30:%.*]] = mul nsw <16 x i32> [[TMP24]], [[TMP28]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE16]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI4]], <16 x i32> [[TMP29]])
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE17]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI5]], <16 x i32> [[TMP30]])
+; CHECK-INTERLEAVED-NEXT: [[TMP31:%.*]] = getelementptr inbounds i8, ptr [[TMP7]], i32 0
+; CHECK-INTERLEAVED-NEXT: [[TMP32:%.*]] = getelementptr inbounds i8, ptr [[TMP7]], i32 16
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD18:%.*]] = load <16 x i8>, ptr [[TMP31]], align 1
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD19:%.*]] = load <16 x i8>, ptr [[TMP32]], align 1
+; CHECK-INTERLEAVED-NEXT: [[TMP33:%.*]] = sext <16 x i8> [[WIDE_LOAD18]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP34:%.*]] = sext <16 x i8> [[WIDE_LOAD19]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP35:%.*]] = getelementptr inbounds i8, ptr [[TMP8]], i32 0
+; CHECK-INTERLEAVED-NEXT: [[TMP36:%.*]] = getelementptr inbounds i8, ptr [[TMP8]], i32 16
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD20:%.*]] = load <16 x i8>, ptr [[TMP35]], align 1
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD21:%.*]] = load <16 x i8>, ptr [[TMP36]], align 1
+; CHECK-INTERLEAVED-NEXT: [[TMP37:%.*]] = sext <16 x i8> [[WIDE_LOAD20]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP56:%.*]] = sext <16 x i8> [[WIDE_LOAD21]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP39:%.*]] = mul nsw <16 x i32> [[TMP33]], [[TMP37]]
+; CHECK-INTERLEAVED-NEXT: [[TMP40:%.*]] = mul nsw <16 x i32> [[TMP34]], [[TMP56]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE7]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI2]], <16 x i32> [[TMP39]])
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI3]], <16 x i32> [[TMP40]])
+; CHECK-INTERLEAVED-NEXT: [[TMP41:%.*]] = getelementptr inbounds i8, ptr [[TMP10]], i32 0
+; CHECK-INTERLEAVED-NEXT: [[TMP42:%.*]] = getelementptr inbounds i8, ptr [[TMP10]], i32 16
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD24:%.*]] = load <16 x i8>, ptr [[TMP41]], align 1
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD25:%.*]] = load <16 x i8>, ptr [[TMP42]], align 1
+; CHECK-INTERLEAVED-NEXT: [[TMP43:%.*]] = sext <16 x i8> [[WIDE_LOAD24]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP44:%.*]] = sext <16 x i8> [[WIDE_LOAD25]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP45:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i32 0
+; CHECK-INTERLEAVED-NEXT: [[TMP46:%.*]] = getelementptr inbounds i8, ptr [[TMP11]], i32 16
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD26:%.*]] = load <16 x i8>, ptr [[TMP45]], align 1
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD27:%.*]] = load <16 x i8>, ptr [[TMP46]], align 1
+; CHECK-INTERLEAVED-NEXT: [[TMP47:%.*]] = sext <16 x i8> [[WIDE_LOAD26]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP48:%.*]] = sext <16 x i8> [[WIDE_LOAD27]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP49:%.*]] = mul nsw <16 x i32> [[TMP43]], [[TMP47]]
+; CHECK-INTERLEAVED-NEXT: [[TMP50:%.*]] = mul nsw <16 x i32> [[TMP44]], [[TMP48]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE13]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI]], <16 x i32> [[TMP49]])
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE10]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI1]], <16 x i32> [[TMP50]])
+; CHECK-INTERLEAVED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 32
+; CHECK-INTERLEAVED-NEXT: [[TMP51:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-INTERLEAVED-NEXT: br i1 [[TMP51]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]]
; CHECK-INTERLEAVED: middle.block:
-; CHECK-INTERLEAVED-NEXT: [[TMP32:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE13]])
-; CHECK-INTERLEAVED-NEXT: [[TMP33:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE10]])
-; CHECK-INTERLEAVED-NEXT: [[TMP34:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE7]])
-; CHECK-INTERLEAVED-NEXT: [[TMP35:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE]])
+; CHECK-INTERLEAVED-NEXT: [[BIN_RDX:%.*]] = add <4 x i32> [[PARTIAL_REDUCE10]], [[PARTIAL_REDUCE13]]
+; CHECK-INTERLEAVED-NEXT: [[TMP52:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX]])
+; CHECK-INTERLEAVED-NEXT: [[BIN_RDX30:%.*]] = add <4 x i32> [[PARTIAL_REDUCE]], [[PARTIAL_REDUCE7]]
+; CHECK-INTERLEAVED-NEXT: [[TMP53:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX30]])
+; CHECK-INTERLEAVED-NEXT: [[BIN_RDX31:%.*]] = add <4 x i32> [[PARTIAL_REDUCE17]], [[PARTIAL_REDUCE16]]
+; CHECK-INTERLEAVED-NEXT: [[TMP54:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX31]])
+; CHECK-INTERLEAVED-NEXT: [[BIN_RDX32:%.*]] = add <4 x i32> [[PARTIAL_REDUCE11]], [[PARTIAL_REDUCE1]]
+; CHECK-INTERLEAVED-NEXT: [[TMP55:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[BIN_RDX32]])
; CHECK-INTERLEAVED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[NUM_IN]], [[N_VEC]]
; CHECK-INTERLEAVED-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
; CHECK-INTERLEAVED: scalar.ph:
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll b/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll
index 69133c78811a5..5eeeacbfe4980 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll
@@ -3116,8 +3116,8 @@ for.exit: ; preds = %for.body
ret i32 %add
}
-define dso_local void @not_dotp_high_register_pressure(ptr %a, ptr %b, ptr %sum, i32 %n) #1 {
-; CHECK-INTERLEAVE1-LABEL: define dso_local void @not_dotp_high_register_pressure(
+define dso_local void @dotp_high_register_pressure(ptr %a, ptr %b, ptr %sum, i32 %n) #1 {
+; CHECK-INTERLEAVE1-LABEL: define dso_local void @dotp_high_register_pressure(
; CHECK-INTERLEAVE1-SAME: ptr [[A:%.*]], ptr [[B:%.*]], ptr [[SUM:%.*]], i32 [[N:%.*]]) #[[ATTR1]] {
; CHECK-INTERLEAVE1-NEXT: entry:
; CHECK-INTERLEAVE1-NEXT: [[CMP100:%.*]] = icmp sgt i32 [[N]], 0
@@ -3139,10 +3139,10 @@ define dso_local void @not_dotp_high_register_pressure(ptr %a, ptr %b, ptr %sum,
; CHECK-INTERLEAVE1-NEXT: [[ARRAYIDX58_PROMOTED:%.*]] = load i32, ptr [[ARRAYIDX58]], align 4
; CHECK-INTERLEAVE1-NEXT: [[ARRAYIDX67_PROMOTED:%.*]] = load i32, ptr [[ARRAYIDX67]], align 4
; CHECK-INTERLEAVE1-NEXT: [[WIDE_TRIP_COUNT:%.*]] = zext nneg i32 [[N]] to i64
-; CHECK-INTERLEAVE1-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 4
+; CHECK-INTERLEAVE1-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 16
; CHECK-INTERLEAVE1-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-INTERLEAVE1: vector.ph:
-; CHECK-INTERLEAVE1-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 4
+; CHECK-INTERLEAVE1-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 16
; CHECK-INTERLEAVE1-NEXT: [[N_VEC:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_MOD_VF]]
; CHECK-INTERLEAVE1-NEXT: [[TMP0:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[ARRAYIDX67_PROMOTED]], i32 0
; CHECK-INTERLEAVE1-NEXT: [[TMP1:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[ARRAYIDX58_PROMOTED]], i32 0
@@ -3155,70 +3155,70 @@ define dso_local void @not_dotp_high_register_pressure(ptr %a, ptr %b, ptr %sum,
; CHECK-INTERLEAVE1-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-INTERLEAVE1: vector.body:
; CHECK-INTERLEAVE1-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP0]], [[VECTOR_PH]] ], [ [[TMP36:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i32> [ [[TMP1]], [[VECTOR_PH]] ], [ [[TMP33:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI2:%.*]] = phi <4 x i32> [ [[TMP2]], [[VECTOR_PH]] ], [ [[TMP30:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI3:%.*]] = phi <4 x i32> [ [[TMP3]], [[VECTOR_PH]] ], [ [[TMP27:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI4:%.*]] = phi <4 x i32> [ [[TMP4]], [[VECTOR_PH]] ], [ [[TMP24:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI5:%.*]] = phi <4 x i32> [ [[TMP5]], [[VECTOR_PH]] ], [ [[TMP21:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI6:%.*]] = phi <4 x i32> [ [[TMP6]], [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI7:%.*]] = phi <4 x i32> [ [[TMP7]], [[VECTOR_PH]] ], [ [[TMP15:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP0]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE21:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i32> [ [[TMP1]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE20:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI2:%.*]] = phi <4 x i32> [ [[TMP2]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE19:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI3:%.*]] = phi <4 x i32> [ [[TMP3]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE18:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI4:%.*]] = phi <4 x i32> [ [[TMP4]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE17:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI5:%.*]] = phi <4 x i32> [ [[TMP5]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE16:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI6:%.*]] = phi <4 x i32> [ [[TMP6]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE15:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVE1-NEXT: [[VEC_PHI7:%.*]] = phi <4 x i32> [ [[TMP7]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE:%.*]], [[VECTOR_BODY]] ]
; CHECK-INTERLEAVE1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[INDEX]]
; CHECK-INTERLEAVE1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP8]], i32 0
-; CHECK-INTERLEAVE1-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP9]], align 1
-; CHECK-INTERLEAVE1-NEXT: [[TMP10:%.*]] = zext <4 x i8> [[WIDE_LOAD]] to <4 x i32>
+; CHECK-INTERLEAVE1-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP9]], align 1
+; CHECK-INTERLEAVE1-NEXT: [[TMP10:%.*]] = zext <16 x i8> [[WIDE_LOAD]] to <16 x i32>
; CHECK-INTERLEAVE1-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDEX]], 3
; CHECK-INTERLEAVE1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP11]]
-; CHECK-INTERLEAVE1-NEXT: [[WIDE_VEC:%.*]] = load <32 x i8>, ptr [[TMP12]], align 1
-; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 0, i32 8, i32 16, i32 24>
-; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC8:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 1, i32 9, i32 17, i32 25>
-; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC9:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 2, i32 10, i32 18, i32 26>
-; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC10:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 3, i32 11, i32 19, i32 27>
-; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC11:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 4, i32 12, i32 20, i32 28>
-; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC12:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 5, i32 13, i32 21, i32 29>
-; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC13:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 6, i32 14, i32 22, i32 30>
-; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC14:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 7, i32 15, i32 23, i32 31>
-; CHECK-INTERLEAVE1-NEXT: [[TMP13:%.*]] = sext <4 x i8> [[STRIDED_VEC]] to <4 x i32>
-; CHECK-INTERLEAVE1-NEXT: [[TMP14:%.*]] = mul nsw <4 x i32> [[TMP13]], [[TMP10]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP15]] = add <4 x i32> [[TMP14]], [[VEC_PHI7]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP16:%.*]] = sext <4 x i8> [[STRIDED_VEC8]] to <4 x i32>
-; CHECK-INTERLEAVE1-NEXT: [[TMP17:%.*]] = mul nsw <4 x i32> [[TMP16]], [[TMP10]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP18]] = add <4 x i32> [[TMP17]], [[VEC_PHI6]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP19:%.*]] = sext <4 x i8> [[STRIDED_VEC9]] to <4 x i32>
-; CHECK-INTERLEAVE1-NEXT: [[TMP20:%.*]] = mul nsw <4 x i32> [[TMP19]], [[TMP10]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP21]] = add <4 x i32> [[TMP20]], [[VEC_PHI5]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP22:%.*]] = sext <4 x i8> [[STRIDED_VEC10]] to <4 x i32>
-; CHECK-INTERLEAVE1-NEXT: [[TMP23:%.*]] = mul nsw <4 x i32> [[TMP22]], [[TMP10]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP24]] = add <4 x i32> [[TMP23]], [[VEC_PHI4]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP25:%.*]] = sext <4 x i8> [[STRIDED_VEC11]] to <4 x i32>
-; CHECK-INTERLEAVE1-NEXT: [[TMP26:%.*]] = mul nsw <4 x i32> [[TMP25]], [[TMP10]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP27]] = add <4 x i32> [[TMP26]], [[VEC_PHI3]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP28:%.*]] = sext <4 x i8> [[STRIDED_VEC12]] to <4 x i32>
-; CHECK-INTERLEAVE1-NEXT: [[TMP29:%.*]] = mul nsw <4 x i32> [[TMP28]], [[TMP10]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP30]] = add <4 x i32> [[TMP29]], [[VEC_PHI2]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP31:%.*]] = sext <4 x i8> [[STRIDED_VEC13]] to <4 x i32>
-; CHECK-INTERLEAVE1-NEXT: [[TMP32:%.*]] = mul nsw <4 x i32> [[TMP31]], [[TMP10]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP33]] = add <4 x i32> [[TMP32]], [[VEC_PHI1]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP34:%.*]] = sext <4 x i8> [[STRIDED_VEC14]] to <4 x i32>
-; CHECK-INTERLEAVE1-NEXT: [[TMP35:%.*]] = mul nsw <4 x i32> [[TMP34]], [[TMP10]]
-; CHECK-INTERLEAVE1-NEXT: [[TMP36]] = add <4 x i32> [[TMP35]], [[VEC_PHI]]
-; CHECK-INTERLEAVE1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-INTERLEAVE1-NEXT: [[TMP37:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-INTERLEAVE1-NEXT: br i1 [[TMP37]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP39:![0-9]+]]
+; CHECK-INTERLEAVE1-NEXT: [[WIDE_VEC:%.*]] = load <128 x i8>, ptr [[TMP12]], align 1
+; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 0, i32 8, i32 16, i32 24, i32 32, i32 40, i32 48, i32 56, i32 64, i32 72, i32 80, i32 88, i32 96, i32 104, i32 112, i32 120>
+; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC8:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 1, i32 9, i32 17, i32 25, i32 33, i32 41, i32 49, i32 57, i32 65, i32 73, i32 81, i32 89, i32 97, i32 105, i32 113, i32 121>
+; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC9:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 2, i32 10, i32 18, i32 26, i32 34, i32 42, i32 50, i32 58, i32 66, i32 74, i32 82, i32 90, i32 98, i32 106, i32 114, i32 122>
+; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC10:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 3, i32 11, i32 19, i32 27, i32 35, i32 43, i32 51, i32 59, i32 67, i32 75, i32 83, i32 91, i32 99, i32 107, i32 115, i32 123>
+; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC11:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 4, i32 12, i32 20, i32 28, i32 36, i32 44, i32 52, i32 60, i32 68, i32 76, i32 84, i32 92, i32 100, i32 108, i32 116, i32 124>
+; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC12:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 5, i32 13, i32 21, i32 29, i32 37, i32 45, i32 53, i32 61, i32 69, i32 77, i32 85, i32 93, i32 101, i32 109, i32 117, i32 125>
+; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC13:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 6, i32 14, i32 22, i32 30, i32 38, i32 46, i32 54, i32 62, i32 70, i32 78, i32 86, i32 94, i32 102, i32 110, i32 118, i32 126>
+; CHECK-INTERLEAVE1-NEXT: [[STRIDED_VEC14:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 7, i32 15, i32 23, i32 31, i32 39, i32 47, i32 55, i32 63, i32 71, i32 79, i32 87, i32 95, i32 103, i32 111, i32 119, i32 127>
+; CHECK-INTERLEAVE1-NEXT: [[TMP13:%.*]] = sext <16 x i8> [[STRIDED_VEC]] to <16 x i32>
+; CHECK-INTERLEAVE1-NEXT: [[TMP14:%.*]] = mul nsw <16 x i32> [[TMP13]], [[TMP10]]
+; CHECK-INTERLEAVE1-NEXT: [[PARTIAL_REDUCE]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI7]], <16 x i32> [[TMP14]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP15:%.*]] = sext <16 x i8> [[STRIDED_VEC8]] to <16 x i32>
+; CHECK-INTERLEAVE1-NEXT: [[TMP16:%.*]] = mul nsw <16 x i32> [[TMP15]], [[TMP10]]
+; CHECK-INTERLEAVE1-NEXT: [[PARTIAL_REDUCE15]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI6]], <16 x i32> [[TMP16]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP17:%.*]] = sext <16 x i8> [[STRIDED_VEC9]] to <16 x i32>
+; CHECK-INTERLEAVE1-NEXT: [[TMP18:%.*]] = mul nsw <16 x i32> [[TMP17]], [[TMP10]]
+; CHECK-INTERLEAVE1-NEXT: [[PARTIAL_REDUCE16]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI5]], <16 x i32> [[TMP18]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP19:%.*]] = sext <16 x i8> [[STRIDED_VEC10]] to <16 x i32>
+; CHECK-INTERLEAVE1-NEXT: [[TMP20:%.*]] = mul nsw <16 x i32> [[TMP19]], [[TMP10]]
+; CHECK-INTERLEAVE1-NEXT: [[PARTIAL_REDUCE17]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI4]], <16 x i32> [[TMP20]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP21:%.*]] = sext <16 x i8> [[STRIDED_VEC11]] to <16 x i32>
+; CHECK-INTERLEAVE1-NEXT: [[TMP22:%.*]] = mul nsw <16 x i32> [[TMP21]], [[TMP10]]
+; CHECK-INTERLEAVE1-NEXT: [[PARTIAL_REDUCE18]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI3]], <16 x i32> [[TMP22]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP23:%.*]] = sext <16 x i8> [[STRIDED_VEC12]] to <16 x i32>
+; CHECK-INTERLEAVE1-NEXT: [[TMP24:%.*]] = mul nsw <16 x i32> [[TMP23]], [[TMP10]]
+; CHECK-INTERLEAVE1-NEXT: [[PARTIAL_REDUCE19]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI2]], <16 x i32> [[TMP24]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP25:%.*]] = sext <16 x i8> [[STRIDED_VEC13]] to <16 x i32>
+; CHECK-INTERLEAVE1-NEXT: [[TMP26:%.*]] = mul nsw <16 x i32> [[TMP25]], [[TMP10]]
+; CHECK-INTERLEAVE1-NEXT: [[PARTIAL_REDUCE20]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI1]], <16 x i32> [[TMP26]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP27:%.*]] = sext <16 x i8> [[STRIDED_VEC14]] to <16 x i32>
+; CHECK-INTERLEAVE1-NEXT: [[TMP28:%.*]] = mul nsw <16 x i32> [[TMP27]], [[TMP10]]
+; CHECK-INTERLEAVE1-NEXT: [[PARTIAL_REDUCE21]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI]], <16 x i32> [[TMP28]])
+; CHECK-INTERLEAVE1-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
+; CHECK-INTERLEAVE1-NEXT: [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-INTERLEAVE1-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP39:![0-9]+]]
; CHECK-INTERLEAVE1: middle.block:
-; CHECK-INTERLEAVE1-NEXT: [[TMP38:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP36]])
-; CHECK-INTERLEAVE1-NEXT: [[TMP39:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP33]])
-; CHECK-INTERLEAVE1-NEXT: [[TMP40:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP30]])
-; CHECK-INTERLEAVE1-NEXT: [[TMP41:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP27]])
-; CHECK-INTERLEAVE1-NEXT: [[TMP42:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP24]])
-; CHECK-INTERLEAVE1-NEXT: [[TMP43:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP21]])
-; CHECK-INTERLEAVE1-NEXT: [[TMP44:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP18]])
-; CHECK-INTERLEAVE1-NEXT: [[TMP45:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP15]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP30:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE21]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP31:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE20]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP32:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE19]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP33:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE18]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP34:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE17]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP35:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE16]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP36:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE15]])
+; CHECK-INTERLEAVE1-NEXT: [[TMP37:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE]])
; CHECK-INTERLEAVE1-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[WIDE_TRIP_COUNT]], [[N_VEC]]
; CHECK-INTERLEAVE1-NEXT: br i1 [[CMP_N]], label [[FOR_COND_FOR_COND_CLEANUP_CRIT_EDGE:%.*]], label [[SCALAR_PH]]
; CHECK-INTERLEAVE1: scalar.ph:
;
-; CHECK-INTERLEAVED-LABEL: define dso_local void @not_dotp_high_register_pressure(
+; CHECK-INTERLEAVED-LABEL: define dso_local void @dotp_high_register_pressure(
; CHECK-INTERLEAVED-SAME: ptr [[A:%.*]], ptr [[B:%.*]], ptr [[SUM:%.*]], i32 [[N:%.*]]) #[[ATTR1]] {
; CHECK-INTERLEAVED-NEXT: entry:
; CHECK-INTERLEAVED-NEXT: [[CMP100:%.*]] = icmp sgt i32 [[N]], 0
@@ -3240,10 +3240,10 @@ define dso_local void @not_dotp_high_register_pressure(ptr %a, ptr %b, ptr %sum,
; CHECK-INTERLEAVED-NEXT: [[ARRAYIDX58_PROMOTED:%.*]] = load i32, ptr [[ARRAYIDX58]], align 4
; CHECK-INTERLEAVED-NEXT: [[ARRAYIDX67_PROMOTED:%.*]] = load i32, ptr [[ARRAYIDX67]], align 4
; CHECK-INTERLEAVED-NEXT: [[WIDE_TRIP_COUNT:%.*]] = zext nneg i32 [[N]] to i64
-; CHECK-INTERLEAVED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 4
+; CHECK-INTERLEAVED-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 16
; CHECK-INTERLEAVED-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-INTERLEAVED: vector.ph:
-; CHECK-INTERLEAVED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 4
+; CHECK-INTERLEAVED-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 16
; CHECK-INTERLEAVED-NEXT: [[N_VEC:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_MOD_VF]]
; CHECK-INTERLEAVED-NEXT: [[TMP0:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[ARRAYIDX67_PROMOTED]], i32 0
; CHECK-INTERLEAVED-NEXT: [[TMP1:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[ARRAYIDX58_PROMOTED]], i32 0
@@ -3256,70 +3256,70 @@ define dso_local void @not_dotp_high_register_pressure(ptr %a, ptr %b, ptr %sum,
; CHECK-INTERLEAVED-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-INTERLEAVED: vector.body:
; CHECK-INTERLEAVED-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVED-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP0]], [[VECTOR_PH]] ], [ [[TMP36:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVED-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i32> [ [[TMP1]], [[VECTOR_PH]] ], [ [[TMP33:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVED-NEXT: [[VEC_PHI2:%.*]] = phi <4 x i32> [ [[TMP2]], [[VECTOR_PH]] ], [ [[TMP30:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVED-NEXT: [[VEC_PHI3:%.*]] = phi <4 x i32> [ [[TMP3]], [[VECTOR_PH]] ], [ [[TMP27:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVED-NEXT: [[VEC_PHI4:%.*]] = phi <4 x i32> [ [[TMP4]], [[VECTOR_PH]] ], [ [[TMP24:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVED-NEXT: [[VEC_PHI5:%.*]] = phi <4 x i32> [ [[TMP5]], [[VECTOR_PH]] ], [ [[TMP21:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVED-NEXT: [[VEC_PHI6:%.*]] = phi <4 x i32> [ [[TMP6]], [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
-; CHECK-INTERLEAVED-NEXT: [[VEC_PHI7:%.*]] = phi <4 x i32> [ [[TMP7]], [[VECTOR_PH]] ], [ [[TMP15:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP0]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE21:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i32> [ [[TMP1]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE20:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI2:%.*]] = phi <4 x i32> [ [[TMP2]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE19:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI3:%.*]] = phi <4 x i32> [ [[TMP3]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE18:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI4:%.*]] = phi <4 x i32> [ [[TMP4]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE17:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI5:%.*]] = phi <4 x i32> [ [[TMP5]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE16:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI6:%.*]] = phi <4 x i32> [ [[TMP6]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE15:%.*]], [[VECTOR_BODY]] ]
+; CHECK-INTERLEAVED-NEXT: [[VEC_PHI7:%.*]] = phi <4 x i32> [ [[TMP7]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE:%.*]], [[VECTOR_BODY]] ]
; CHECK-INTERLEAVED-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[INDEX]]
; CHECK-INTERLEAVED-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP8]], i32 0
-; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP9]], align 1
-; CHECK-INTERLEAVED-NEXT: [[TMP10:%.*]] = zext <4 x i8> [[WIDE_LOAD]] to <4 x i32>
+; CHECK-INTERLEAVED-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP9]], align 1
+; CHECK-INTERLEAVED-NEXT: [[TMP10:%.*]] = zext <16 x i8> [[WIDE_LOAD]] to <16 x i32>
; CHECK-INTERLEAVED-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDEX]], 3
; CHECK-INTERLEAVED-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP11]]
-; CHECK-INTERLEAVED-NEXT: [[WIDE_VEC:%.*]] = load <32 x i8>, ptr [[TMP12]], align 1
-; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 0, i32 8, i32 16, i32 24>
-; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC8:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 1, i32 9, i32 17, i32 25>
-; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC9:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 2, i32 10, i32 18, i32 26>
-; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC10:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 3, i32 11, i32 19, i32 27>
-; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC11:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 4, i32 12, i32 20, i32 28>
-; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC12:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 5, i32 13, i32 21, i32 29>
-; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC13:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 6, i32 14, i32 22, i32 30>
-; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC14:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 7, i32 15, i32 23, i32 31>
-; CHECK-INTERLEAVED-NEXT: [[TMP13:%.*]] = sext <4 x i8> [[STRIDED_VEC]] to <4 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP14:%.*]] = mul nsw <4 x i32> [[TMP13]], [[TMP10]]
-; CHECK-INTERLEAVED-NEXT: [[TMP15]] = add <4 x i32> [[TMP14]], [[VEC_PHI7]]
-; CHECK-INTERLEAVED-NEXT: [[TMP16:%.*]] = sext <4 x i8> [[STRIDED_VEC8]] to <4 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP17:%.*]] = mul nsw <4 x i32> [[TMP16]], [[TMP10]]
-; CHECK-INTERLEAVED-NEXT: [[TMP18]] = add <4 x i32> [[TMP17]], [[VEC_PHI6]]
-; CHECK-INTERLEAVED-NEXT: [[TMP19:%.*]] = sext <4 x i8> [[STRIDED_VEC9]] to <4 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP20:%.*]] = mul nsw <4 x i32> [[TMP19]], [[TMP10]]
-; CHECK-INTERLEAVED-NEXT: [[TMP21]] = add <4 x i32> [[TMP20]], [[VEC_PHI5]]
-; CHECK-INTERLEAVED-NEXT: [[TMP22:%.*]] = sext <4 x i8> [[STRIDED_VEC10]] to <4 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP23:%.*]] = mul nsw <4 x i32> [[TMP22]], [[TMP10]]
-; CHECK-INTERLEAVED-NEXT: [[TMP24]] = add <4 x i32> [[TMP23]], [[VEC_PHI4]]
-; CHECK-INTERLEAVED-NEXT: [[TMP25:%.*]] = sext <4 x i8> [[STRIDED_VEC11]] to <4 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP26:%.*]] = mul nsw <4 x i32> [[TMP25]], [[TMP10]]
-; CHECK-INTERLEAVED-NEXT: [[TMP27]] = add <4 x i32> [[TMP26]], [[VEC_PHI3]]
-; CHECK-INTERLEAVED-NEXT: [[TMP28:%.*]] = sext <4 x i8> [[STRIDED_VEC12]] to <4 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP29:%.*]] = mul nsw <4 x i32> [[TMP28]], [[TMP10]]
-; CHECK-INTERLEAVED-NEXT: [[TMP30]] = add <4 x i32> [[TMP29]], [[VEC_PHI2]]
-; CHECK-INTERLEAVED-NEXT: [[TMP31:%.*]] = sext <4 x i8> [[STRIDED_VEC13]] to <4 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP32:%.*]] = mul nsw <4 x i32> [[TMP31]], [[TMP10]]
-; CHECK-INTERLEAVED-NEXT: [[TMP33]] = add <4 x i32> [[TMP32]], [[VEC_PHI1]]
-; CHECK-INTERLEAVED-NEXT: [[TMP34:%.*]] = sext <4 x i8> [[STRIDED_VEC14]] to <4 x i32>
-; CHECK-INTERLEAVED-NEXT: [[TMP35:%.*]] = mul nsw <4 x i32> [[TMP34]], [[TMP10]]
-; CHECK-INTERLEAVED-NEXT: [[TMP36]] = add <4 x i32> [[TMP35]], [[VEC_PHI]]
-; CHECK-INTERLEAVED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-INTERLEAVED-NEXT: [[TMP37:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-INTERLEAVED-NEXT: br i1 [[TMP37]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP41:![0-9]+]]
+; CHECK-INTERLEAVED-NEXT: [[WIDE_VEC:%.*]] = load <128 x i8>, ptr [[TMP12]], align 1
+; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 0, i32 8, i32 16, i32 24, i32 32, i32 40, i32 48, i32 56, i32 64, i32 72, i32 80, i32 88, i32 96, i32 104, i32 112, i32 120>
+; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC8:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 1, i32 9, i32 17, i32 25, i32 33, i32 41, i32 49, i32 57, i32 65, i32 73, i32 81, i32 89, i32 97, i32 105, i32 113, i32 121>
+; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC9:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 2, i32 10, i32 18, i32 26, i32 34, i32 42, i32 50, i32 58, i32 66, i32 74, i32 82, i32 90, i32 98, i32 106, i32 114, i32 122>
+; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC10:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 3, i32 11, i32 19, i32 27, i32 35, i32 43, i32 51, i32 59, i32 67, i32 75, i32 83, i32 91, i32 99, i32 107, i32 115, i32 123>
+; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC11:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 4, i32 12, i32 20, i32 28, i32 36, i32 44, i32 52, i32 60, i32 68, i32 76, i32 84, i32 92, i32 100, i32 108, i32 116, i32 124>
+; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC12:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 5, i32 13, i32 21, i32 29, i32 37, i32 45, i32 53, i32 61, i32 69, i32 77, i32 85, i32 93, i32 101, i32 109, i32 117, i32 125>
+; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC13:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 6, i32 14, i32 22, i32 30, i32 38, i32 46, i32 54, i32 62, i32 70, i32 78, i32 86, i32 94, i32 102, i32 110, i32 118, i32 126>
+; CHECK-INTERLEAVED-NEXT: [[STRIDED_VEC14:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 7, i32 15, i32 23, i32 31, i32 39, i32 47, i32 55, i32 63, i32 71, i32 79, i32 87, i32 95, i32 103, i32 111, i32 119, i32 127>
+; CHECK-INTERLEAVED-NEXT: [[TMP13:%.*]] = sext <16 x i8> [[STRIDED_VEC]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP14:%.*]] = mul nsw <16 x i32> [[TMP13]], [[TMP10]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI7]], <16 x i32> [[TMP14]])
+; CHECK-INTERLEAVED-NEXT: [[TMP15:%.*]] = sext <16 x i8> [[STRIDED_VEC8]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP16:%.*]] = mul nsw <16 x i32> [[TMP15]], [[TMP10]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE15]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI6]], <16 x i32> [[TMP16]])
+; CHECK-INTERLEAVED-NEXT: [[TMP17:%.*]] = sext <16 x i8> [[STRIDED_VEC9]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP18:%.*]] = mul nsw <16 x i32> [[TMP17]], [[TMP10]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE16]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI5]], <16 x i32> [[TMP18]])
+; CHECK-INTERLEAVED-NEXT: [[TMP19:%.*]] = sext <16 x i8> [[STRIDED_VEC10]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP20:%.*]] = mul nsw <16 x i32> [[TMP19]], [[TMP10]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE17]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI4]], <16 x i32> [[TMP20]])
+; CHECK-INTERLEAVED-NEXT: [[TMP21:%.*]] = sext <16 x i8> [[STRIDED_VEC11]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP22:%.*]] = mul nsw <16 x i32> [[TMP21]], [[TMP10]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE18]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI3]], <16 x i32> [[TMP22]])
+; CHECK-INTERLEAVED-NEXT: [[TMP23:%.*]] = sext <16 x i8> [[STRIDED_VEC12]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP24:%.*]] = mul nsw <16 x i32> [[TMP23]], [[TMP10]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE19]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI2]], <16 x i32> [[TMP24]])
+; CHECK-INTERLEAVED-NEXT: [[TMP25:%.*]] = sext <16 x i8> [[STRIDED_VEC13]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP26:%.*]] = mul nsw <16 x i32> [[TMP25]], [[TMP10]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE20]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI1]], <16 x i32> [[TMP26]])
+; CHECK-INTERLEAVED-NEXT: [[TMP27:%.*]] = sext <16 x i8> [[STRIDED_VEC14]] to <16 x i32>
+; CHECK-INTERLEAVED-NEXT: [[TMP28:%.*]] = mul nsw <16 x i32> [[TMP27]], [[TMP10]]
+; CHECK-INTERLEAVED-NEXT: [[PARTIAL_REDUCE21]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI]], <16 x i32> [[TMP28]])
+; CHECK-INTERLEAVED-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
+; CHECK-INTERLEAVED-NEXT: [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-INTERLEAVED-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP41:![0-9]+]]
; CHECK-INTERLEAVED: middle.block:
-; CHECK-INTERLEAVED-NEXT: [[TMP38:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP36]])
-; CHECK-INTERLEAVED-NEXT: [[TMP39:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP33]])
-; CHECK-INTERLEAVED-NEXT: [[TMP40:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP30]])
-; CHECK-INTERLEAVED-NEXT: [[TMP41:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP27]])
-; CHECK-INTERLEAVED-NEXT: [[TMP42:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP24]])
-; CHECK-INTERLEAVED-NEXT: [[TMP43:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP21]])
-; CHECK-INTERLEAVED-NEXT: [[TMP44:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP18]])
-; CHECK-INTERLEAVED-NEXT: [[TMP45:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP15]])
+; CHECK-INTERLEAVED-NEXT: [[TMP30:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE21]])
+; CHECK-INTERLEAVED-NEXT: [[TMP31:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE20]])
+; CHECK-INTERLEAVED-NEXT: [[TMP32:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE19]])
+; CHECK-INTERLEAVED-NEXT: [[TMP33:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE18]])
+; CHECK-INTERLEAVED-NEXT: [[TMP34:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE17]])
+; CHECK-INTERLEAVED-NEXT: [[TMP35:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE16]])
+; CHECK-INTERLEAVED-NEXT: [[TMP36:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE15]])
+; CHECK-INTERLEAVED-NEXT: [[TMP37:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE]])
; CHECK-INTERLEAVED-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[WIDE_TRIP_COUNT]], [[N_VEC]]
; CHECK-INTERLEAVED-NEXT: br i1 [[CMP_N]], label [[FOR_COND_FOR_COND_CLEANUP_CRIT_EDGE:%.*]], label [[SCALAR_PH]]
; CHECK-INTERLEAVED: scalar.ph:
;
-; CHECK-MAXBW-LABEL: define dso_local void @not_dotp_high_register_pressure(
+; CHECK-MAXBW-LABEL: define dso_local void @dotp_high_register_pressure(
; CHECK-MAXBW-SAME: ptr [[A:%.*]], ptr [[B:%.*]], ptr [[SUM:%.*]], i32 [[N:%.*]]) #[[ATTR1]] {
; CHECK-MAXBW-NEXT: entry:
; CHECK-MAXBW-NEXT: [[CMP100:%.*]] = icmp sgt i32 [[N]], 0
@@ -3341,10 +3341,10 @@ define dso_local void @not_dotp_high_register_pressure(ptr %a, ptr %b, ptr %sum,
; CHECK-MAXBW-NEXT: [[ARRAYIDX58_PROMOTED:%.*]] = load i32, ptr [[ARRAYIDX58]], align 4
; CHECK-MAXBW-NEXT: [[ARRAYIDX67_PROMOTED:%.*]] = load i32, ptr [[ARRAYIDX67]], align 4
; CHECK-MAXBW-NEXT: [[WIDE_TRIP_COUNT:%.*]] = zext nneg i32 [[N]] to i64
-; CHECK-MAXBW-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 4
+; CHECK-MAXBW-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[WIDE_TRIP_COUNT]], 16
; CHECK-MAXBW-NEXT: br i1 [[MIN_ITERS_CHECK]], label [[SCALAR_PH:%.*]], label [[VECTOR_PH:%.*]]
; CHECK-MAXBW: vector.ph:
-; CHECK-MAXBW-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 4
+; CHECK-MAXBW-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[WIDE_TRIP_COUNT]], 16
; CHECK-MAXBW-NEXT: [[N_VEC:%.*]] = sub i64 [[WIDE_TRIP_COUNT]], [[N_MOD_VF]]
; CHECK-MAXBW-NEXT: [[TMP0:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[ARRAYIDX67_PROMOTED]], i32 0
; CHECK-MAXBW-NEXT: [[TMP1:%.*]] = insertelement <4 x i32> zeroinitializer, i32 [[ARRAYIDX58_PROMOTED]], i32 0
@@ -3357,65 +3357,65 @@ define dso_local void @not_dotp_high_register_pressure(ptr %a, ptr %b, ptr %sum,
; CHECK-MAXBW-NEXT: br label [[VECTOR_BODY:%.*]]
; CHECK-MAXBW: vector.body:
; CHECK-MAXBW-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ]
-; CHECK-MAXBW-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP0]], [[VECTOR_PH]] ], [ [[TMP36:%.*]], [[VECTOR_BODY]] ]
-; CHECK-MAXBW-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i32> [ [[TMP1]], [[VECTOR_PH]] ], [ [[TMP33:%.*]], [[VECTOR_BODY]] ]
-; CHECK-MAXBW-NEXT: [[VEC_PHI2:%.*]] = phi <4 x i32> [ [[TMP2]], [[VECTOR_PH]] ], [ [[TMP30:%.*]], [[VECTOR_BODY]] ]
-; CHECK-MAXBW-NEXT: [[VEC_PHI3:%.*]] = phi <4 x i32> [ [[TMP3]], [[VECTOR_PH]] ], [ [[TMP27:%.*]], [[VECTOR_BODY]] ]
-; CHECK-MAXBW-NEXT: [[VEC_PHI4:%.*]] = phi <4 x i32> [ [[TMP4]], [[VECTOR_PH]] ], [ [[TMP24:%.*]], [[VECTOR_BODY]] ]
-; CHECK-MAXBW-NEXT: [[VEC_PHI5:%.*]] = phi <4 x i32> [ [[TMP5]], [[VECTOR_PH]] ], [ [[TMP21:%.*]], [[VECTOR_BODY]] ]
-; CHECK-MAXBW-NEXT: [[VEC_PHI6:%.*]] = phi <4 x i32> [ [[TMP6]], [[VECTOR_PH]] ], [ [[TMP18:%.*]], [[VECTOR_BODY]] ]
-; CHECK-MAXBW-NEXT: [[VEC_PHI7:%.*]] = phi <4 x i32> [ [[TMP7]], [[VECTOR_PH]] ], [ [[TMP15:%.*]], [[VECTOR_BODY]] ]
+; CHECK-MAXBW-NEXT: [[VEC_PHI:%.*]] = phi <4 x i32> [ [[TMP0]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE21:%.*]], [[VECTOR_BODY]] ]
+; CHECK-MAXBW-NEXT: [[VEC_PHI1:%.*]] = phi <4 x i32> [ [[TMP1]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE20:%.*]], [[VECTOR_BODY]] ]
+; CHECK-MAXBW-NEXT: [[VEC_PHI2:%.*]] = phi <4 x i32> [ [[TMP2]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE19:%.*]], [[VECTOR_BODY]] ]
+; CHECK-MAXBW-NEXT: [[VEC_PHI3:%.*]] = phi <4 x i32> [ [[TMP3]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE18:%.*]], [[VECTOR_BODY]] ]
+; CHECK-MAXBW-NEXT: [[VEC_PHI4:%.*]] = phi <4 x i32> [ [[TMP4]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE17:%.*]], [[VECTOR_BODY]] ]
+; CHECK-MAXBW-NEXT: [[VEC_PHI5:%.*]] = phi <4 x i32> [ [[TMP5]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE16:%.*]], [[VECTOR_BODY]] ]
+; CHECK-MAXBW-NEXT: [[VEC_PHI6:%.*]] = phi <4 x i32> [ [[TMP6]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE15:%.*]], [[VECTOR_BODY]] ]
+; CHECK-MAXBW-NEXT: [[VEC_PHI7:%.*]] = phi <4 x i32> [ [[TMP7]], [[VECTOR_PH]] ], [ [[PARTIAL_REDUCE:%.*]], [[VECTOR_BODY]] ]
; CHECK-MAXBW-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw i8, ptr [[A]], i64 [[INDEX]]
; CHECK-MAXBW-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw i8, ptr [[TMP8]], i32 0
-; CHECK-MAXBW-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i8>, ptr [[TMP9]], align 1
-; CHECK-MAXBW-NEXT: [[TMP10:%.*]] = zext <4 x i8> [[WIDE_LOAD]] to <4 x i32>
+; CHECK-MAXBW-NEXT: [[WIDE_LOAD:%.*]] = load <16 x i8>, ptr [[TMP9]], align 1
+; CHECK-MAXBW-NEXT: [[TMP10:%.*]] = zext <16 x i8> [[WIDE_LOAD]] to <16 x i32>
; CHECK-MAXBW-NEXT: [[TMP11:%.*]] = shl nsw i64 [[INDEX]], 3
; CHECK-MAXBW-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw i8, ptr [[B]], i64 [[TMP11]]
-; CHECK-MAXBW-NEXT: [[WIDE_VEC:%.*]] = load <32 x i8>, ptr [[TMP12]], align 1
-; CHECK-MAXBW-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 0, i32 8, i32 16, i32 24>
-; CHECK-MAXBW-NEXT: [[STRIDED_VEC8:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 1, i32 9, i32 17, i32 25>
-; CHECK-MAXBW-NEXT: [[STRIDED_VEC9:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 2, i32 10, i32 18, i32 26>
-; CHECK-MAXBW-NEXT: [[STRIDED_VEC10:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 3, i32 11, i32 19, i32 27>
-; CHECK-MAXBW-NEXT: [[STRIDED_VEC11:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 4, i32 12, i32 20, i32 28>
-; CHECK-MAXBW-NEXT: [[STRIDED_VEC12:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 5, i32 13, i32 21, i32 29>
-; CHECK-MAXBW-NEXT: [[STRIDED_VEC13:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 6, i32 14, i32 22, i32 30>
-; CHECK-MAXBW-NEXT: [[STRIDED_VEC14:%.*]] = shufflevector <32 x i8> [[WIDE_VEC]], <32 x i8> poison, <4 x i32> <i32 7, i32 15, i32 23, i32 31>
-; CHECK-MAXBW-NEXT: [[TMP13:%.*]] = sext <4 x i8> [[STRIDED_VEC]] to <4 x i32>
-; CHECK-MAXBW-NEXT: [[TMP14:%.*]] = mul nsw <4 x i32> [[TMP13]], [[TMP10]]
-; CHECK-MAXBW-NEXT: [[TMP15]] = add <4 x i32> [[TMP14]], [[VEC_PHI7]]
-; CHECK-MAXBW-NEXT: [[TMP16:%.*]] = sext <4 x i8> [[STRIDED_VEC8]] to <4 x i32>
-; CHECK-MAXBW-NEXT: [[TMP17:%.*]] = mul nsw <4 x i32> [[TMP16]], [[TMP10]]
-; CHECK-MAXBW-NEXT: [[TMP18]] = add <4 x i32> [[TMP17]], [[VEC_PHI6]]
-; CHECK-MAXBW-NEXT: [[TMP19:%.*]] = sext <4 x i8> [[STRIDED_VEC9]] to <4 x i32>
-; CHECK-MAXBW-NEXT: [[TMP20:%.*]] = mul nsw <4 x i32> [[TMP19]], [[TMP10]]
-; CHECK-MAXBW-NEXT: [[TMP21]] = add <4 x i32> [[TMP20]], [[VEC_PHI5]]
-; CHECK-MAXBW-NEXT: [[TMP22:%.*]] = sext <4 x i8> [[STRIDED_VEC10]] to <4 x i32>
-; CHECK-MAXBW-NEXT: [[TMP23:%.*]] = mul nsw <4 x i32> [[TMP22]], [[TMP10]]
-; CHECK-MAXBW-NEXT: [[TMP24]] = add <4 x i32> [[TMP23]], [[VEC_PHI4]]
-; CHECK-MAXBW-NEXT: [[TMP25:%.*]] = sext <4 x i8> [[STRIDED_VEC11]] to <4 x i32>
-; CHECK-MAXBW-NEXT: [[TMP26:%.*]] = mul nsw <4 x i32> [[TMP25]], [[TMP10]]
-; CHECK-MAXBW-NEXT: [[TMP27]] = add <4 x i32> [[TMP26]], [[VEC_PHI3]]
-; CHECK-MAXBW-NEXT: [[TMP28:%.*]] = sext <4 x i8> [[STRIDED_VEC12]] to <4 x i32>
-; CHECK-MAXBW-NEXT: [[TMP29:%.*]] = mul nsw <4 x i32> [[TMP28]], [[TMP10]]
-; CHECK-MAXBW-NEXT: [[TMP30]] = add <4 x i32> [[TMP29]], [[VEC_PHI2]]
-; CHECK-MAXBW-NEXT: [[TMP31:%.*]] = sext <4 x i8> [[STRIDED_VEC13]] to <4 x i32>
-; CHECK-MAXBW-NEXT: [[TMP32:%.*]] = mul nsw <4 x i32> [[TMP31]], [[TMP10]]
-; CHECK-MAXBW-NEXT: [[TMP33]] = add <4 x i32> [[TMP32]], [[VEC_PHI1]]
-; CHECK-MAXBW-NEXT: [[TMP34:%.*]] = sext <4 x i8> [[STRIDED_VEC14]] to <4 x i32>
-; CHECK-MAXBW-NEXT: [[TMP35:%.*]] = mul nsw <4 x i32> [[TMP34]], [[TMP10]]
-; CHECK-MAXBW-NEXT: [[TMP36]] = add <4 x i32> [[TMP35]], [[VEC_PHI]]
-; CHECK-MAXBW-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4
-; CHECK-MAXBW-NEXT: [[TMP37:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
-; CHECK-MAXBW-NEXT: br i1 [[TMP37]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP39:![0-9]+]]
+; CHECK-MAXBW-NEXT: [[WIDE_VEC:%.*]] = load <128 x i8>, ptr [[TMP12]], align 1
+; CHECK-MAXBW-NEXT: [[STRIDED_VEC:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 0, i32 8, i32 16, i32 24, i32 32, i32 40, i32 48, i32 56, i32 64, i32 72, i32 80, i32 88, i32 96, i32 104, i32 112, i32 120>
+; CHECK-MAXBW-NEXT: [[STRIDED_VEC8:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 1, i32 9, i32 17, i32 25, i32 33, i32 41, i32 49, i32 57, i32 65, i32 73, i32 81, i32 89, i32 97, i32 105, i32 113, i32 121>
+; CHECK-MAXBW-NEXT: [[STRIDED_VEC9:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 2, i32 10, i32 18, i32 26, i32 34, i32 42, i32 50, i32 58, i32 66, i32 74, i32 82, i32 90, i32 98, i32 106, i32 114, i32 122>
+; CHECK-MAXBW-NEXT: [[STRIDED_VEC10:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 3, i32 11, i32 19, i32 27, i32 35, i32 43, i32 51, i32 59, i32 67, i32 75, i32 83, i32 91, i32 99, i32 107, i32 115, i32 123>
+; CHECK-MAXBW-NEXT: [[STRIDED_VEC11:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 4, i32 12, i32 20, i32 28, i32 36, i32 44, i32 52, i32 60, i32 68, i32 76, i32 84, i32 92, i32 100, i32 108, i32 116, i32 124>
+; CHECK-MAXBW-NEXT: [[STRIDED_VEC12:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 5, i32 13, i32 21, i32 29, i32 37, i32 45, i32 53, i32 61, i32 69, i32 77, i32 85, i32 93, i32 101, i32 109, i32 117, i32 125>
+; CHECK-MAXBW-NEXT: [[STRIDED_VEC13:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 6, i32 14, i32 22, i32 30, i32 38, i32 46, i32 54, i32 62, i32 70, i32 78, i32 86, i32 94, i32 102, i32 110, i32 118, i32 126>
+; CHECK-MAXBW-NEXT: [[STRIDED_VEC14:%.*]] = shufflevector <128 x i8> [[WIDE_VEC]], <128 x i8> poison, <16 x i32> <i32 7, i32 15, i32 23, i32 31, i32 39, i32 47, i32 55, i32 63, i32 71, i32 79, i32 87, i32 95, i32 103, i32 111, i32 119, i32 127>
+; CHECK-MAXBW-NEXT: [[TMP13:%.*]] = sext <16 x i8> [[STRIDED_VEC]] to <16 x i32>
+; CHECK-MAXBW-NEXT: [[TMP14:%.*]] = mul nsw <16 x i32> [[TMP13]], [[TMP10]]
+; CHECK-MAXBW-NEXT: [[PARTIAL_REDUCE]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI7]], <16 x i32> [[TMP14]])
+; CHECK-MAXBW-NEXT: [[TMP15:%.*]] = sext <16 x i8> [[STRIDED_VEC8]] to <16 x i32>
+; CHECK-MAXBW-NEXT: [[TMP16:%.*]] = mul nsw <16 x i32> [[TMP15]], [[TMP10]]
+; CHECK-MAXBW-NEXT: [[PARTIAL_REDUCE15]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI6]], <16 x i32> [[TMP16]])
+; CHECK-MAXBW-NEXT: [[TMP17:%.*]] = sext <16 x i8> [[STRIDED_VEC9]] to <16 x i32>
+; CHECK-MAXBW-NEXT: [[TMP18:%.*]] = mul nsw <16 x i32> [[TMP17]], [[TMP10]]
+; CHECK-MAXBW-NEXT: [[PARTIAL_REDUCE16]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI5]], <16 x i32> [[TMP18]])
+; CHECK-MAXBW-NEXT: [[TMP19:%.*]] = sext <16 x i8> [[STRIDED_VEC10]] to <16 x i32>
+; CHECK-MAXBW-NEXT: [[TMP20:%.*]] = mul nsw <16 x i32> [[TMP19]], [[TMP10]]
+; CHECK-MAXBW-NEXT: [[PARTIAL_REDUCE17]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI4]], <16 x i32> [[TMP20]])
+; CHECK-MAXBW-NEXT: [[TMP21:%.*]] = sext <16 x i8> [[STRIDED_VEC11]] to <16 x i32>
+; CHECK-MAXBW-NEXT: [[TMP22:%.*]] = mul nsw <16 x i32> [[TMP21]], [[TMP10]]
+; CHECK-MAXBW-NEXT: [[PARTIAL_REDUCE18]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI3]], <16 x i32> [[TMP22]])
+; CHECK-MAXBW-NEXT: [[TMP23:%.*]] = sext <16 x i8> [[STRIDED_VEC12]] to <16 x i32>
+; CHECK-MAXBW-NEXT: [[TMP24:%.*]] = mul nsw <16 x i32> [[TMP23]], [[TMP10]]
+; CHECK-MAXBW-NEXT: [[PARTIAL_REDUCE19]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI2]], <16 x i32> [[TMP24]])
+; CHECK-MAXBW-NEXT: [[TMP25:%.*]] = sext <16 x i8> [[STRIDED_VEC13]] to <16 x i32>
+; CHECK-MAXBW-NEXT: [[TMP26:%.*]] = mul nsw <16 x i32> [[TMP25]], [[TMP10]]
+; CHECK-MAXBW-NEXT: [[PARTIAL_REDUCE20]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI1]], <16 x i32> [[TMP26]])
+; CHECK-MAXBW-NEXT: [[TMP27:%.*]] = sext <16 x i8> [[STRIDED_VEC14]] to <16 x i32>
+; CHECK-MAXBW-NEXT: [[TMP28:%.*]] = mul nsw <16 x i32> [[TMP27]], [[TMP10]]
+; CHECK-MAXBW-NEXT: [[PARTIAL_REDUCE21]] = call <4 x i32> @llvm.experimental.vector.partial.reduce.add.v4i32.v16i32(<4 x i32> [[VEC_PHI]], <16 x i32> [[TMP28]])
+; CHECK-MAXBW-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 16
+; CHECK-MAXBW-NEXT: [[TMP29:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
+; CHECK-MAXBW-NEXT: br i1 [[TMP29]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP39:![0-9]+]]
; CHECK-MAXBW: middle.block:
-; CHECK-MAXBW-NEXT: [[TMP38:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP36]])
-; CHECK-MAXBW-NEXT: [[TMP39:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP33]])
-; CHECK-MAXBW-NEXT: [[TMP40:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP30]])
-; CHECK-MAXBW-NEXT: [[TMP41:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP27]])
-; CHECK-MAXBW-NEXT: [[TMP42:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP24]])
-; CHECK-MAXBW-NEXT: [[TMP43:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP21]])
-; CHECK-MAXBW-NEXT: [[TMP44:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP18]])
-; CHECK-MAXBW-NEXT: [[TMP45:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[TMP15]])
+; CHECK-MAXBW-NEXT: [[TMP30:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE21]])
+; CHECK-MAXBW-NEXT: [[TMP31:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE20]])
+; CHECK-MAXBW-NEXT: [[TMP32:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE19]])
+; CHECK-MAXBW-NEXT: [[TMP33:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE18]])
+; CHECK-MAXBW-NEXT: [[TMP34:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE17]])
+; CHECK-MAXBW-NEXT: [[TMP35:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE16]])
+; CHECK-MAXBW-NEXT: [[TMP36:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE15]])
+; CHECK-MAXBW-NEXT: [[TMP37:%.*]] = call i32 @llvm.vector.reduce.add.v4i32(<4 x i32> [[PARTIAL_REDUCE]])
; CHECK-MAXBW-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[WIDE_TRIP_COUNT]], [[N_VEC]]
; CHECK-MAXBW-NEXT: br i1 [[CMP_N]], label [[FOR_COND_FOR_COND_CLEANUP_CRIT_EDGE:%.*]], label [[SCALAR_PH]]
; CHECK-MAXBW: scalar.ph:
>From 9c8ac1d421f7e623cfb802593099ef79244c5558 Mon Sep 17 00:00:00 2001
From: Samuel Tebbs <samuel.tebbs at arm.com>
Date: Mon, 31 Mar 2025 15:23:46 +0100
Subject: [PATCH 2/5] Rename to getVFScaleFactor
---
llvm/lib/Transforms/Vectorize/LoopVectorize.cpp | 2 +-
llvm/lib/Transforms/Vectorize/VPlan.h | 14 +++++++-------
2 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index 400a510be308b..c4b88e9fc567b 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -5026,7 +5026,7 @@ calculateRegisterUsage(VPlan &Plan, ArrayRef<ElementCount> VFs,
VF = VF.divideCoefficientBy(ReductionR->getVFScaleFactor());
else if (auto *PartialReductionR =
dyn_cast<VPPartialReductionRecipe>(R))
- VF = VF.divideCoefficientBy(PartialReductionR->getScaleFactor());
+ VF = VF.divideCoefficientBy(PartialReductionR->getVFScaleFactor());
if (VF != VFs[J])
LLVM_DEBUG(dbgs() << "LV(REG): Scaled down VF from " << VFs[J]
<< " to " << VF << " for ";
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.h b/llvm/lib/Transforms/Vectorize/VPlan.h
index 376526e804b4b..5b7bf8eb17594 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.h
+++ b/llvm/lib/Transforms/Vectorize/VPlan.h
@@ -2065,19 +2065,19 @@ class VPReductionPHIRecipe : public VPHeaderPHIRecipe,
/// scalar value.
class VPPartialReductionRecipe : public VPSingleDefRecipe {
unsigned Opcode;
- unsigned ScaleFactor;
+ unsigned VFScaleFactor;
public:
VPPartialReductionRecipe(Instruction *ReductionInst, VPValue *Op0,
- VPValue *Op1, unsigned ScaleFactor)
+ VPValue *Op1, unsigned VFScaleFactor)
: VPPartialReductionRecipe(ReductionInst->getOpcode(), Op0, Op1,
- ScaleFactor, ReductionInst) {}
+ VFScaleFactor, ReductionInst) {}
VPPartialReductionRecipe(unsigned Opcode, VPValue *Op0, VPValue *Op1,
- unsigned ScaleFactor,
+ unsigned VFScaleFactor,
Instruction *ReductionInst = nullptr)
: VPSingleDefRecipe(VPDef::VPPartialReductionSC,
ArrayRef<VPValue *>({Op0, Op1}), ReductionInst),
- Opcode(Opcode), ScaleFactor(ScaleFactor) {
+ Opcode(Opcode), VFScaleFactor(VFScaleFactor) {
[[maybe_unused]] auto *AccumulatorRecipe =
getOperand(1)->getDefiningRecipe();
assert((isa<VPReductionPHIRecipe>(AccumulatorRecipe) ||
@@ -2088,7 +2088,7 @@ class VPPartialReductionRecipe : public VPSingleDefRecipe {
VPPartialReductionRecipe *clone() override {
return new VPPartialReductionRecipe(Opcode, getOperand(0), getOperand(1),
- ScaleFactor, getUnderlyingInstr());
+ VFScaleFactor, getUnderlyingInstr());
}
VP_CLASSOF_IMPL(VPDef::VPPartialReductionSC)
@@ -2103,7 +2103,7 @@ class VPPartialReductionRecipe : public VPSingleDefRecipe {
/// Get the binary op's opcode.
unsigned getOpcode() const { return Opcode; }
- unsigned getScaleFactor() const { return ScaleFactor; }
+ unsigned getVFScaleFactor() const { return VFScaleFactor; }
#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
/// Print the recipe.
>From c72adbedf02da4eb406da679c47b36f3bdf4a0da Mon Sep 17 00:00:00 2001
From: Samuel Tebbs <samuel.tebbs at arm.com>
Date: Mon, 31 Mar 2025 15:06:54 +0100
Subject: [PATCH 3/5] Put if statement inside DEBUG
---
llvm/lib/Transforms/Vectorize/LoopVectorize.cpp | 9 +++++----
1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index c4b88e9fc567b..89097450fc5a7 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -5027,10 +5027,11 @@ calculateRegisterUsage(VPlan &Plan, ArrayRef<ElementCount> VFs,
else if (auto *PartialReductionR =
dyn_cast<VPPartialReductionRecipe>(R))
VF = VF.divideCoefficientBy(PartialReductionR->getVFScaleFactor());
- if (VF != VFs[J])
- LLVM_DEBUG(dbgs() << "LV(REG): Scaled down VF from " << VFs[J]
- << " to " << VF << " for ";
- R->dump(););
+ LLVM_DEBUG(if (VF != VFs[J]) {
+ dbgs() << "LV(REG): Scaled down VF from " << VFs[J] << " to "
+ << VF << " for ";
+ R->dump();
+ });
for (VPValue *DefV : R->definedValues()) {
Type *ScalarTy = TypeInfo.inferScalarType(DefV);
>From dfeaea64c24a695bc4b4f40df4ad4c936d4dd982 Mon Sep 17 00:00:00 2001
From: Samuel Tebbs <samuel.tebbs at arm.com>
Date: Wed, 2 Apr 2025 16:54:27 +0100
Subject: [PATCH 4/5] auto -> ElementCount
---
llvm/lib/Transforms/Vectorize/LoopVectorize.cpp | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
index 89097450fc5a7..9d64ecc37770f 100644
--- a/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ b/llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -5021,12 +5021,13 @@ calculateRegisterUsage(VPlan &Plan, ArrayRef<ElementCount> VFs,
} else {
// The output from scaled phis and scaled reductions actually have
// fewer lanes than the VF.
- auto VF = VFs[J];
+ ElementCount VF = VFs[J];
if (auto *ReductionR = dyn_cast<VPReductionPHIRecipe>(R))
VF = VF.divideCoefficientBy(ReductionR->getVFScaleFactor());
else if (auto *PartialReductionR =
dyn_cast<VPPartialReductionRecipe>(R))
VF = VF.divideCoefficientBy(PartialReductionR->getVFScaleFactor());
+
LLVM_DEBUG(if (VF != VFs[J]) {
dbgs() << "LV(REG): Scaled down VF from " << VFs[J] << " to "
<< VF << " for ";
>From 317de8ef7e19774960cd852fa6c3d772474b3e16 Mon Sep 17 00:00:00 2001
From: Samuel Tebbs <samuel.tebbs at arm.com>
Date: Wed, 2 Apr 2025 16:54:43 +0100
Subject: [PATCH 5/5] Add register pressure debug output checks
---
.../AArch64/partial-reduce-dot-product-neon.ll | 7 +++++++
.../LoopVectorize/AArch64/partial-reduce-dot-product.ll | 7 +++++++
2 files changed, 14 insertions(+)
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product-neon.ll b/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product-neon.ll
index a77447500d598..8772ef5ffc0d1 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product-neon.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product-neon.ll
@@ -2,6 +2,7 @@
; RUN: opt -passes=loop-vectorize -enable-epilogue-vectorization=false -mattr=+neon,+dotprod -force-vector-interleave=1 -S < %s | FileCheck %s --check-prefixes=CHECK-INTERLEAVE1
; RUN: opt -passes=loop-vectorize -enable-epilogue-vectorization=false -mattr=+neon,+dotprod -S < %s | FileCheck %s --check-prefixes=CHECK-INTERLEAVED
; RUN: opt -passes=loop-vectorize -enable-epilogue-vectorization=false -mattr=+neon,+dotprod -force-vector-interleave=1 -vectorizer-maximize-bandwidth -S < %s | FileCheck %s --check-prefixes=CHECK-MAXBW
+; RUN: opt -passes=loop-vectorize -debug-only=loop-vectorize --disable-output -S < %s 2>&1 | FileCheck %s --check-prefix=CHECK-REGS
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
target triple = "aarch64-none-unknown-elf"
@@ -947,6 +948,12 @@ define i32 @dotp_unrolled(i32 %num_out, i64 %num_in, ptr %a, ptr %b) {
; CHECK-MAXBW-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]]
; CHECK-MAXBW: scalar.ph:
;
+; CHECK-REGS: LV: Checking a loop in 'dotp_unrolled' from <stdin>
+; CHECK-REGS: LV(REG): VF = 16
+; CHECK-REGS-NEXT: LV(REG): Found max usage: 2 item
+; CHECK-REGS-NEXT: LV(REG): RegisterClass: Generic::ScalarRC, 9 registers
+; CHECK-REGS-NEXT: LV(REG): RegisterClass: Generic::VectorRC, 24 registers
+; CHECK-REGS-NEXT: LV(REG): Found invariant usage: 1 item
entry:
br label %for.body
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll b/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll
index 5eeeacbfe4980..327029f28c3b1 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/partial-reduce-dot-product.ll
@@ -2,6 +2,7 @@
; RUN: opt -passes=loop-vectorize -force-vector-interleave=1 -enable-epilogue-vectorization=false -S < %s | FileCheck %s --check-prefixes=CHECK-INTERLEAVE1
; RUN: opt -passes=loop-vectorize -enable-epilogue-vectorization=false -S < %s | FileCheck %s --check-prefixes=CHECK-INTERLEAVED
; RUN: opt -passes=loop-vectorize -force-vector-interleave=1 -vectorizer-maximize-bandwidth -enable-epilogue-vectorization=false -S < %s | FileCheck %s --check-prefixes=CHECK-MAXBW
+; RUN: opt -passes=loop-vectorize -debug-only=loop-vectorize --disable-output -S < %s 2>&1 | FileCheck %s --check-prefix=CHECK-REGS
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
target triple = "aarch64-none-unknown-elf"
@@ -3420,6 +3421,12 @@ define dso_local void @dotp_high_register_pressure(ptr %a, ptr %b, ptr %sum, i32
; CHECK-MAXBW-NEXT: br i1 [[CMP_N]], label [[FOR_COND_FOR_COND_CLEANUP_CRIT_EDGE:%.*]], label [[SCALAR_PH]]
; CHECK-MAXBW: scalar.ph:
;
+; CHECK-REGS-LABEL: LV: Checking a loop in 'dotp_high_register_pressure' from <stdin>
+; CHECK-REGS: LV(REG): VF = 16
+; CHECK-REGS-NEXT: LV(REG): Found max usage: 2 item
+; CHECK-REGS-NEXT: LV(REG): RegisterClass: Generic::ScalarRC, 3 registers
+; CHECK-REGS-NEXT: LV(REG): RegisterClass: Generic::VectorRC, 24 registers
+; CHECK-REGS-NEXT: LV(REG): Found invariant usage: 1 item
entry:
%cmp100 = icmp sgt i32 %n, 0
br i1 %cmp100, label %for.body.lr.ph, label %for.cond.cleanup
More information about the llvm-branch-commits
mailing list