[Matrix] Add special case dot product lowering

[virnarula]

Summary

Add special case to matrix lowering for dot products. Normal matrix lowering if optimized for either row-major or column-major, which results in many shufflevector instructions being generated for one vector. We work around this in our special case. We can also use vector-reduce adds instead of sequential adds to sum the result of the element-wise multiplication, which takes advantage of SIMD instructions.

[virnarula]

Fix formatting with clang-format

[fhahn]

llvm/lib/Transforms/Scalar/LowerMatrixIntrinsics.cpp

885

I don't think this is necessary and it looks a bit odd to overwrite the earlier Changed status. It should be sufficient to remove setting Changed on line 877.

1214

Please add a comment describing what this function does.

1217

I think we should be able to just look up the shape of operands 0 and 1 in ShapeMap instead?

1220

nit: coding style recommends omitting {} for single-line blocks.

1229

nit: Coding style uses UpperCase for variables.

1249

might be good to explain what costs we are comparing.

1252

nit: the style guide recommends avoiding auto in most cases where the type is not entirely obvious from the context (e.g. auto for dyn_cast is fine).

1262

nit: use UpperCase for variables.

1264

To check for a specific intrinsic, you could either use something like match(&I, m_Intrinsic<Intrinsic::matrix_transpose>(m_Value(TA))) or cast to IntrinsicInst and checking getIntrinsicID().

1292

I am not sure if that's correct. Here we need to load a vector with 1 row and N columns. llvm.matrix.column.load may have a stride > 1 between columns, which is ignored. We should probably limit this to .llvm.matrix.column.load with a stride of 1 here for now. Needs a test case.

1317

nit: according to the style guide, comments are meant to be full sentences, with the first letter capitalized and a period at the end. Applies to most comments added in the patch.

1323

nit: no need for {}

1326

nit: no need for {}

1330

nit: no need to return at the end of a void function.

[thegameg]

llvm/lib/Transforms/Scalar/LowerMatrixIntrinsics.cpp

1240

You can probably remove this and use IRBuilder::Create(F)AddReduce directly below?

1244

Maybe we can do all the FMF and cost checks before everything?

1272

You might want to specify what you mean by "differently" (not using LowerLoad because we would end up with scalar loads for the row vector).

1274

To add on top of what @fhahn said above, you can combine matchers like this:

match(&I, m_CombineOr(m_Load(), m_Intrinsic<Intrinsic::matrix_column_major_load>())

llvm/test/Transforms/LowerMatrixIntrinsics/dot-product.ll

165

Very nice!

[virnarula]

Address first round of comments.

[virnarula]

llvm/lib/Transforms/Scalar/LowerMatrixIntrinsics.cpp

1217

Using ShapeMap works when we load matrix inside the function. However, it doesn't if the matrix is passed in as an argument. As such, I have not implemented this.

1292

Most of our test cases didn't account for this (they used strides greater than 1). I am submitting an additional review that will update the test cases to use strides of 1 but keep some that don't as test cases.

1292

https://reviews.llvm.org/D131444

[thegameg]

LGTM, thank you!

llvm/lib/Transforms/Scalar/LowerMatrixIntrinsics.cpp

1237

Not sure what's going on here. Intrinsic with Instruction::Add? It also doesn't seem to be really used.

1253

You cold match the stride directly here:

ConstantInt *LoadStride = nullptr;
if (match(Val, m_Intrinsic<Intrinsic::matrix_column_major_load>(
                       m_Value(), m_ConstantInt(LoadStride), m_Value(), m_Value()) &&
    LoadStride->getZExtValue() == Stride) {

and pass Stride as an argument of GetBuiltinLoad.

[virnarula]

llvm/lib/Transforms/Scalar/LowerMatrixIntrinsics.cpp

1237

I use it to get the function type when estimating sequential add cost. line 1243. Do you suggest doing it a different way?

1253

I tried this but I realized that I would have to have a control flow to check if the load has size of n anyways. This is because if we return null, it could mean there's a LoadInst (instead of a builtin load) or the stride in incorrect, which require different behaviors.

[thegameg]

llvm/lib/Transforms/Scalar/LowerMatrixIntrinsics.cpp

1237

It looks like you only use AddOpCode and Add is only used for the type, right?

[fhahn]

Rebased. I also extended int coverage, so we have enough coverage to land this and build in it.

[fhahn]

LGTM, thanks!

llvm/lib/Transforms/Scalar/LowerMatrixIntrinsics.cpp

 //===- LowerMatrixIntrinsics.cpp -  Lower matrix intrinsics -----*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
for (auto *BB : RPOT)
       }
 
     // Second, try to fuse candidates.
     SmallPtrSet<Instruction *, 16> FusedInsts;
     for (CallInst *CI : MaybeFusableInsts)
       LowerMatrixMultiplyFused(CI, FusedInsts);
     Changed = !FusedInsts.empty();
 
-    // Third, lower remaining instructions with shape information.
+    // Third, try to lower any dot products
+    for (CallInst *CI : MaybeFusableInsts) {
+      if (FusedInsts.find(CI) != FusedInsts.end()) // skip if already fused
+        continue;
+      lowerDotProduct(CI, FusedInsts, getFastMathFlags(CI));
+    }
+    Changed = !FusedInsts.empty();
+
+    // Fourth, lower remaining instructions with shape information.
     for (Instruction *Inst : MatrixInsts) {
       if (FusedInsts.count(Inst))
         continue;
 
       IRBuilder<> Builder(Inst);
 
       if (CallInst *CInst = dyn_cast<CallInst>(Inst))
         Changed |= VisitCallInst(CInst);
for (Use &U : llvm::make_early_inc_range(Inst->uses())) {
       if (ShapeMap.find(U.getUser()) == ShapeMap.end()) {
         if (!Flattened)
           Flattened = Matrix.embedInVector(Builder);
         U.set(Flattened);
       }
     }
   }
 
+  void lowerDotProduct(CallInst *MatMul,
+                       SmallPtrSet<Instruction *, 16> &FusedInsts,
+                       FastMathFlags FMF) {
+    ShapeInfo LShape(MatMul->getArgOperand(2), MatMul->getArgOperand(3));
+    ShapeInfo RShape(MatMul->getArgOperand(3), MatMul->getArgOperand(4));
+
+    if (LShape.NumRows != 1 || RShape.NumColumns != 1) { // not a dot product
+      return;
+    }
+
+    Value *LHS = MatMul->getArgOperand(0);
+    Value *RHS = MatMul->getArgOperand(1);
+
+    // Separate functions for floating point and integer computations
+    Type *ElementType = cast<VectorType>(LHS->getType())->getElementType();
+    bool integerOperands = ElementType->isIntegerTy();
+    Function *Reduce, *Add;
+    int AddOpCode;
+    if (integerOperands) {
+      Reduce = Intrinsic::getDeclaration(
+          Func.getParent(), Intrinsic::vector_reduce_add, LHS->getType());
+      Add = Intrinsic::getDeclaration(Func.getParent(), Instruction::Add,
+                                      ElementType);
+      AddOpCode = Instruction::Add;
+    } else {
+      Reduce = Intrinsic::getDeclaration(
+          Func.getParent(), Intrinsic::vector_reduce_fadd, LHS->getType());
+      Add = Intrinsic::getDeclaration(Func.getParent(), Instruction::FAdd,
+                                      ElementType);
+      AddOpCode = Instruction::FAdd;
+      if (!FMF.allowReassoc()) {
+        return;
+      } // reassociation required for vector.reduce.fadd
+    }
+
+    // Check that dot product lowering is profitable
+    FastMathFlags FMFReassoc;
+    FMFReassoc.setAllowReassoc();
+    auto ReductionCost = TTI.getArithmeticReductionCost(
+        AddOpCode, cast<VectorType>(LHS->getType()), FMFReassoc);
+    auto SequentialAddCost =
+        TTI.getArithmeticInstrCost(AddOpCode, Add->getType()) *
+        (LShape.NumColumns - 1);
+    if (ReductionCost >= SequentialAddCost) {
+      return;
+    }
+
+    // lambda which functions as dyn_cast<BuiltinLoad>
+    auto getBuiltinLoad = [](Value *Val) -> CallInst * {
+      CallInst *CI = dyn_cast<CallInst>(Val);
+      if (CI && CI->getCalledFunction()->getName().startswith(
+                    "llvm.matrix.column.major.load")) {
+        return CI;
+      }
+      return nullptr;
+    };
+
+    // Since row vectors loads have to be lowered differently, matmul must be
+    // the only user
+    CallInst *LHSBuiltinLoad = getBuiltinLoad(LHS);
+    if (LHSBuiltinLoad || isa<LoadInst>(LHS)) {
+      if (LHS->hasOneUse())
+        FusedInsts.insert(cast<Instruction>(LHS));
+      else
+        return;
+    }
+
+    CallInst *RHSBuiltinLoad = getBuiltinLoad(RHS);
+    if (RHSBuiltinLoad || isa<LoadInst>(RHS)) {
+      if (RHS->hasOneUse())
+        FusedInsts.insert(cast<Instruction>(RHS));
+      else
+        return;
+    }
+    FusedInsts.insert(MatMul);
+    IRBuilder<> Builder(MatMul);
+    // If vector uses the builtin load, lower to a LoadInst
+    if (LHSBuiltinLoad) {
+      LHS =
+          Builder.CreateLoad(LHS->getType(), LHSBuiltinLoad->getArgOperand(0));
+    }
+
+    if (RHSBuiltinLoad) {
+      RHS =
+          Builder.CreateLoad(RHS->getType(), RHSBuiltinLoad->getArgOperand(0));
+    }
+
+    // Insert mul/fmul and llvm.vector.reduce.fadd
+    Value *Mul = integerOperands ? Builder.CreateMul(LHS, RHS)
+                                 : Builder.CreateFMul(LHS, RHS);
+
+    Value *Result;
+    if (integerOperands) {
+      Result = Builder.CreateCall(Reduce, {Mul});
+    } else {
+      Result = Builder.CreateCall(
+          Reduce, {ConstantFP::get(
+                       cast<VectorType>(LHS->getType())->getElementType(), 0.0),
+                   Mul});
+      cast<Instruction>(Result)->setFastMathFlags(FMF);
+    }
+
+    // pack scalar back into a matrix and then replace matmul inst
+    Result = Builder.CreateInsertElement(PoisonValue::get(MatMul->getType()),
+                                         Result, uint64_t(0));
+    MatMul->replaceAllUsesWith(Result);
+    MatMul->eraseFromParent();
+
+    // Remove BuiltinLoad if we already generated a LoadInst for it
+    if (LHSBuiltinLoad) {
+      LHSBuiltinLoad->eraseFromParent();
+    }
+    if (RHSBuiltinLoad) {
+      RHSBuiltinLoad->eraseFromParent();
+    }
+
+    return;
+  }
+
   /// Compute \p Result += \p A * \p B for input matrices with left-associating
   /// addition.
   ///
   /// We can fold a transpose into the operand that is used to extract scalars.
   /// This is the first operands with row-major and the second with
   /// column-major.  If \p IsScalarMatrixTransposed we assume the appropriate
   /// operand is transposed.
   void emitMatrixMultiply(MatrixTy &Result, const MatrixTy &A,

llvm/test/Transforms/LowerMatrixIntrinsics/dot-product.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; REQUIRES: aarch64-registered-target
 ; RUN: opt -passes='lower-matrix-intrinsics' -mtriple=arm64-apple-iphoneos -S < %s | FileCheck %s
 
 define <1 x float> @dotproduct_float_v6(<6 x float> %a, <6 x float> %b) {
 ; CHECK-LABEL: @dotproduct_float_v6(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[SPLIT:%.*]] = shufflevector <6 x float> [[A:%.*]], <6 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[SPLIT1:%.*]] = shufflevector <6 x float> [[A]], <6 x float> poison, <1 x i32> <i32 1>
-; CHECK-NEXT:    [[SPLIT2:%.*]] = shufflevector <6 x float> [[A]], <6 x float> poison, <1 x i32> <i32 2>
-; CHECK-NEXT:    [[SPLIT3:%.*]] = shufflevector <6 x float> [[A]], <6 x float> poison, <1 x i32> <i32 3>
-; CHECK-NEXT:    [[SPLIT4:%.*]] = shufflevector <6 x float> [[A]], <6 x float> poison, <1 x i32> <i32 4>
-; CHECK-NEXT:    [[SPLIT5:%.*]] = shufflevector <6 x float> [[A]], <6 x float> poison, <1 x i32> <i32 5>
-; CHECK-NEXT:    [[SPLIT6:%.*]] = shufflevector <6 x float> [[B:%.*]], <6 x float> poison, <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5>
-; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <1 x float> [[SPLIT]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP0:%.*]] = extractelement <6 x float> [[SPLIT6]], i64 0
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x float> poison, float [[TMP0]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP1:%.*]] = fmul fast <1 x float> [[BLOCK]], [[SPLAT_SPLAT]]
-; CHECK-NEXT:    [[BLOCK7:%.*]] = shufflevector <1 x float> [[SPLIT1]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <6 x float> [[SPLIT6]], i64 1
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT8:%.*]] = insertelement <1 x float> poison, float [[TMP2]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT9:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT8]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP3:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK7]], <1 x float> [[SPLAT_SPLAT9]], <1 x float> [[TMP1]])
-; CHECK-NEXT:    [[BLOCK10:%.*]] = shufflevector <1 x float> [[SPLIT2]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <6 x float> [[SPLIT6]], i64 2
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT11:%.*]] = insertelement <1 x float> poison, float [[TMP4]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT12:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT11]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP5:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK10]], <1 x float> [[SPLAT_SPLAT12]], <1 x float> [[TMP3]])
-; CHECK-NEXT:    [[BLOCK13:%.*]] = shufflevector <1 x float> [[SPLIT3]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <6 x float> [[SPLIT6]], i64 3
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT14:%.*]] = insertelement <1 x float> poison, float [[TMP6]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT15:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT14]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP7:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK13]], <1 x float> [[SPLAT_SPLAT15]], <1 x float> [[TMP5]])
-; CHECK-NEXT:    [[BLOCK16:%.*]] = shufflevector <1 x float> [[SPLIT4]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <6 x float> [[SPLIT6]], i64 4
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT17:%.*]] = insertelement <1 x float> poison, float [[TMP8]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT18:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT17]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP9:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK16]], <1 x float> [[SPLAT_SPLAT18]], <1 x float> [[TMP7]])
-; CHECK-NEXT:    [[BLOCK19:%.*]] = shufflevector <1 x float> [[SPLIT5]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <6 x float> [[SPLIT6]], i64 5
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <1 x float> poison, float [[TMP10]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT20]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP11:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK19]], <1 x float> [[SPLAT_SPLAT21]], <1 x float> [[TMP9]])
-; CHECK-NEXT:    [[TMP12:%.*]] = shufflevector <1 x float> [[TMP11]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP13:%.*]] = shufflevector <1 x float> undef, <1 x float> [[TMP12]], <1 x i32> <i32 1>
-; CHECK-NEXT:    ret <1 x float> [[TMP13]]
+; CHECK-NEXT:    [[TMP0:%.*]] = fmul <6 x float> [[A:%.*]], [[B:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast float @llvm.vector.reduce.fadd.v6f32(float 0.000000e+00, <6 x float> [[TMP0]])
+; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <1 x float> poison, float [[TMP1]], i64 0
+; CHECK-NEXT:    ret <1 x float> [[TMP2]]
 ;
 entry:
   %c = tail call fast <1 x float> @llvm.matrix.multiply.v1f32.v6f32.v6f32(<6 x float> %a, <6 x float> %b, i32 1, i32 6, i32 1)
   ret <1 x float> %c
 }
 
 declare <1 x float> @llvm.matrix.multiply.v1f32.v6f32.v6f32(<6 x float>, <6 x float>, i32, i32, i32)
 
entry:
   ret <1 x float> %c
 }
 
 declare <1 x float> @llvm.matrix.multiply.v1f32.v1f32.v1f32(<1 x float>, <1 x float>, i32, i32, i32)
 
 define <1 x float> @dotproduct_float_v3(<3 x float> %a, <3 x float> %b) {
 ; CHECK-LABEL: @dotproduct_float_v3(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[SPLIT:%.*]] = shufflevector <3 x float> [[A:%.*]], <3 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[SPLIT1:%.*]] = shufflevector <3 x float> [[A]], <3 x float> poison, <1 x i32> <i32 1>
-; CHECK-NEXT:    [[SPLIT2:%.*]] = shufflevector <3 x float> [[A]], <3 x float> poison, <1 x i32> <i32 2>
-; CHECK-NEXT:    [[SPLIT3:%.*]] = shufflevector <3 x float> [[B:%.*]], <3 x float> poison, <3 x i32> <i32 0, i32 1, i32 2>
-; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <1 x float> [[SPLIT]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP0:%.*]] = extractelement <3 x float> [[SPLIT3]], i64 0
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x float> poison, float [[TMP0]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP1:%.*]] = fmul fast <1 x float> [[BLOCK]], [[SPLAT_SPLAT]]
-; CHECK-NEXT:    [[BLOCK4:%.*]] = shufflevector <1 x float> [[SPLIT1]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <3 x float> [[SPLIT3]], i64 1
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT5:%.*]] = insertelement <1 x float> poison, float [[TMP2]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT6:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT5]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP3:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK4]], <1 x float> [[SPLAT_SPLAT6]], <1 x float> [[TMP1]])
-; CHECK-NEXT:    [[BLOCK7:%.*]] = shufflevector <1 x float> [[SPLIT2]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <3 x float> [[SPLIT3]], i64 2
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT8:%.*]] = insertelement <1 x float> poison, float [[TMP4]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT9:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT8]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP5:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK7]], <1 x float> [[SPLAT_SPLAT9]], <1 x float> [[TMP3]])
-; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <1 x float> [[TMP5]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP7:%.*]] = shufflevector <1 x float> undef, <1 x float> [[TMP6]], <1 x i32> <i32 1>
-; CHECK-NEXT:    ret <1 x float> [[TMP7]]
+; CHECK-NEXT:    [[TMP0:%.*]] = fmul <3 x float> [[A:%.*]], [[B:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast float @llvm.vector.reduce.fadd.v3f32(float 0.000000e+00, <3 x float> [[TMP0]])
+; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <1 x float> poison, float [[TMP1]], i64 0
+; CHECK-NEXT:    ret <1 x float> [[TMP2]]
 ;
 entry:
   %c = tail call fast <1 x float> @llvm.matrix.multiply.v1f32.v3f32.v3f32(<3 x float> %a, <3 x float> %b, i32 1, i32 3, i32 1)
   ret <1 x float> %c
 }
 
 declare <1 x float> @llvm.matrix.multiply.v1f32.v3f32.v3f32(<3 x float>, <3 x float>, i32, i32, i32)
 
 define <1 x float> @intrinsic_column_major_load_dot_product_float_v6(ptr %lhs_address, ptr %rhs_address) {
 ; CHECK-LABEL: @intrinsic_column_major_load_dot_product_float_v6(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <6 x float>, ptr [[LHS_ADDRESS:%.*]], align 4
-; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <1 x float>, ptr [[RHS_ADDRESS:%.*]], align 4
-; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr float, ptr [[RHS_ADDRESS]], i64 6
-; CHECK-NEXT:    [[COL_LOAD2:%.*]] = load <1 x float>, ptr [[VEC_GEP]], align 4
-; CHECK-NEXT:    [[VEC_GEP3:%.*]] = getelementptr float, ptr [[RHS_ADDRESS]], i64 12
-; CHECK-NEXT:    [[COL_LOAD4:%.*]] = load <1 x float>, ptr [[VEC_GEP3]], align 4
-; CHECK-NEXT:    [[VEC_GEP5:%.*]] = getelementptr float, ptr [[RHS_ADDRESS]], i64 18
-; CHECK-NEXT:    [[COL_LOAD6:%.*]] = load <1 x float>, ptr [[VEC_GEP5]], align 4
-; CHECK-NEXT:    [[VEC_GEP7:%.*]] = getelementptr float, ptr [[RHS_ADDRESS]], i64 24
-; CHECK-NEXT:    [[COL_LOAD8:%.*]] = load <1 x float>, ptr [[VEC_GEP7]], align 4
-; CHECK-NEXT:    [[VEC_GEP9:%.*]] = getelementptr float, ptr [[RHS_ADDRESS]], i64 30
-; CHECK-NEXT:    [[COL_LOAD10:%.*]] = load <1 x float>, ptr [[VEC_GEP9]], align 4
-; CHECK-NEXT:    [[SPLIT:%.*]] = shufflevector <6 x float> [[COL_LOAD]], <6 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[SPLIT11:%.*]] = shufflevector <6 x float> [[COL_LOAD]], <6 x float> poison, <1 x i32> <i32 1>
-; CHECK-NEXT:    [[SPLIT12:%.*]] = shufflevector <6 x float> [[COL_LOAD]], <6 x float> poison, <1 x i32> <i32 2>
-; CHECK-NEXT:    [[SPLIT13:%.*]] = shufflevector <6 x float> [[COL_LOAD]], <6 x float> poison, <1 x i32> <i32 3>
-; CHECK-NEXT:    [[SPLIT14:%.*]] = shufflevector <6 x float> [[COL_LOAD]], <6 x float> poison, <1 x i32> <i32 4>
-; CHECK-NEXT:    [[SPLIT15:%.*]] = shufflevector <6 x float> [[COL_LOAD]], <6 x float> poison, <1 x i32> <i32 5>
-; CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <1 x float> [[COL_LOAD1]], <1 x float> [[COL_LOAD2]], <2 x i32> <i32 0, i32 1>
-; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <1 x float> [[COL_LOAD4]], <1 x float> [[COL_LOAD6]], <2 x i32> <i32 0, i32 1>
-; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <1 x float> [[COL_LOAD8]], <1 x float> [[COL_LOAD10]], <2 x i32> <i32 0, i32 1>
-; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <2 x float> [[TMP0]], <2 x float> [[TMP1]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEXT:    [[TMP4:%.*]] = shufflevector <2 x float> [[TMP2]], <2 x float> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <4 x float> [[TMP3]], <4 x float> [[TMP4]], <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5>
-; CHECK-NEXT:    [[SPLIT16:%.*]] = shufflevector <6 x float> [[TMP5]], <6 x float> poison, <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5>
-; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <1 x float> [[SPLIT]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <6 x float> [[SPLIT16]], i64 0
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x float> poison, float [[TMP6]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP7:%.*]] = fmul fast <1 x float> [[BLOCK]], [[SPLAT_SPLAT]]
-; CHECK-NEXT:    [[BLOCK17:%.*]] = shufflevector <1 x float> [[SPLIT11]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <6 x float> [[SPLIT16]], i64 1
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT18:%.*]] = insertelement <1 x float> poison, float [[TMP8]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT19:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT18]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP9:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK17]], <1 x float> [[SPLAT_SPLAT19]], <1 x float> [[TMP7]])
-; CHECK-NEXT:    [[BLOCK20:%.*]] = shufflevector <1 x float> [[SPLIT12]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <6 x float> [[SPLIT16]], i64 2
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT21:%.*]] = insertelement <1 x float> poison, float [[TMP10]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT22:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT21]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP11:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK20]], <1 x float> [[SPLAT_SPLAT22]], <1 x float> [[TMP9]])
-; CHECK-NEXT:    [[BLOCK23:%.*]] = shufflevector <1 x float> [[SPLIT13]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <6 x float> [[SPLIT16]], i64 3
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT24:%.*]] = insertelement <1 x float> poison, float [[TMP12]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT25:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT24]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP13:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK23]], <1 x float> [[SPLAT_SPLAT25]], <1 x float> [[TMP11]])
-; CHECK-NEXT:    [[BLOCK26:%.*]] = shufflevector <1 x float> [[SPLIT14]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <6 x float> [[SPLIT16]], i64 4
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT27:%.*]] = insertelement <1 x float> poison, float [[TMP14]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT28:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT27]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP15:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK26]], <1 x float> [[SPLAT_SPLAT28]], <1 x float> [[TMP13]])
-; CHECK-NEXT:    [[BLOCK29:%.*]] = shufflevector <1 x float> [[SPLIT15]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <6 x float> [[SPLIT16]], i64 5
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT30:%.*]] = insertelement <1 x float> poison, float [[TMP16]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT31:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT30]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP17:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK29]], <1 x float> [[SPLAT_SPLAT31]], <1 x float> [[TMP15]])
-; CHECK-NEXT:    [[TMP18:%.*]] = shufflevector <1 x float> [[TMP17]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP19:%.*]] = shufflevector <1 x float> undef, <1 x float> [[TMP18]], <1 x i32> <i32 1>
-; CHECK-NEXT:    ret <1 x float> [[TMP19]]
+; CHECK-NEXT:    [[TMP0:%.*]] = load <6 x float>, ptr [[LHS_ADDRESS:%.*]], align 32
+; CHECK-NEXT:    [[TMP1:%.*]] = load <6 x float>, ptr [[RHS_ADDRESS:%.*]], align 32
+; CHECK-NEXT:    [[TMP2:%.*]] = fmul <6 x float> [[TMP0]], [[TMP1]]
+; CHECK-NEXT:    [[TMP3:%.*]] = call fast float @llvm.vector.reduce.fadd.v6f32(float 0.000000e+00, <6 x float> [[TMP2]])
+; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <1 x float> poison, float [[TMP3]], i64 0
+; CHECK-NEXT:    ret <1 x float> [[TMP4]]
 ;
 entry:
   %lhs = tail call fast <6 x float> @llvm.matrix.column.major.load.v6f32.i64(ptr nonnull align 4 %lhs_address, i64 6, i1 false, i32 6, i32 1)
   %rhs = tail call fast <6 x float> @llvm.matrix.column.major.load.v6f32.i64(ptr nonnull align 4 %rhs_address, i64 6, i1 false, i32 1, i32 6)
   %result = tail call fast <1 x float> @llvm.matrix.multiply.v1f32.v6f32.v6f32(<6 x float> %lhs, <6 x float> %rhs, i32 1, i32 6, i32 1)
   ret <1 x float> %result
 }
 
 declare <6 x float> @llvm.matrix.column.major.load.v6f32.i64(ptr nonnull align 4, i64, i1, i32, i32)
 
 define <1 x float> @LoadInst_dot_product_float_v7(ptr %lhs_address, ptr %rhs_address) {
 ; CHECK-LABEL: @LoadInst_dot_product_float_v7(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <1 x float>, ptr [[LHS_ADDRESS:%.*]], align 32
-; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr float, ptr [[LHS_ADDRESS]], i64 1
-; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <1 x float>, ptr [[VEC_GEP]], align 4
-; CHECK-NEXT:    [[VEC_GEP2:%.*]] = getelementptr float, ptr [[LHS_ADDRESS]], i64 2
-; CHECK-NEXT:    [[COL_LOAD3:%.*]] = load <1 x float>, ptr [[VEC_GEP2]], align 8
-; CHECK-NEXT:    [[VEC_GEP4:%.*]] = getelementptr float, ptr [[LHS_ADDRESS]], i64 3
-; CHECK-NEXT:    [[COL_LOAD5:%.*]] = load <1 x float>, ptr [[VEC_GEP4]], align 4
-; CHECK-NEXT:    [[VEC_GEP6:%.*]] = getelementptr float, ptr [[LHS_ADDRESS]], i64 4
-; CHECK-NEXT:    [[COL_LOAD7:%.*]] = load <1 x float>, ptr [[VEC_GEP6]], align 16
-; CHECK-NEXT:    [[VEC_GEP8:%.*]] = getelementptr float, ptr [[LHS_ADDRESS]], i64 5
-; CHECK-NEXT:    [[COL_LOAD9:%.*]] = load <1 x float>, ptr [[VEC_GEP8]], align 4
-; CHECK-NEXT:    [[VEC_GEP10:%.*]] = getelementptr float, ptr [[LHS_ADDRESS]], i64 6
-; CHECK-NEXT:    [[COL_LOAD11:%.*]] = load <1 x float>, ptr [[VEC_GEP10]], align 8
-; CHECK-NEXT:    [[COL_LOAD12:%.*]] = load <7 x float>, ptr [[RHS_ADDRESS:%.*]], align 32
-; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <1 x float> [[COL_LOAD]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP0:%.*]] = extractelement <7 x float> [[COL_LOAD12]], i64 0
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x float> poison, float [[TMP0]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP1:%.*]] = fmul fast <1 x float> [[BLOCK]], [[SPLAT_SPLAT]]
-; CHECK-NEXT:    [[BLOCK13:%.*]] = shufflevector <1 x float> [[COL_LOAD1]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <7 x float> [[COL_LOAD12]], i64 1
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT14:%.*]] = insertelement <1 x float> poison, float [[TMP2]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT15:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT14]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP3:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK13]], <1 x float> [[SPLAT_SPLAT15]], <1 x float> [[TMP1]])
-; CHECK-NEXT:    [[BLOCK16:%.*]] = shufflevector <1 x float> [[COL_LOAD3]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <7 x float> [[COL_LOAD12]], i64 2
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT17:%.*]] = insertelement <1 x float> poison, float [[TMP4]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT18:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT17]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP5:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK16]], <1 x float> [[SPLAT_SPLAT18]], <1 x float> [[TMP3]])
-; CHECK-NEXT:    [[BLOCK19:%.*]] = shufflevector <1 x float> [[COL_LOAD5]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <7 x float> [[COL_LOAD12]], i64 3
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <1 x float> poison, float [[TMP6]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT20]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP7:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK19]], <1 x float> [[SPLAT_SPLAT21]], <1 x float> [[TMP5]])
-; CHECK-NEXT:    [[BLOCK22:%.*]] = shufflevector <1 x float> [[COL_LOAD7]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <7 x float> [[COL_LOAD12]], i64 4
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT23:%.*]] = insertelement <1 x float> poison, float [[TMP8]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT24:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT23]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP9:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK22]], <1 x float> [[SPLAT_SPLAT24]], <1 x float> [[TMP7]])
-; CHECK-NEXT:    [[BLOCK25:%.*]] = shufflevector <1 x float> [[COL_LOAD9]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <7 x float> [[COL_LOAD12]], i64 5
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT26:%.*]] = insertelement <1 x float> poison, float [[TMP10]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT27:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT26]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP11:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK25]], <1 x float> [[SPLAT_SPLAT27]], <1 x float> [[TMP9]])
-; CHECK-NEXT:    [[BLOCK28:%.*]] = shufflevector <1 x float> [[COL_LOAD11]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <7 x float> [[COL_LOAD12]], i64 6
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT29:%.*]] = insertelement <1 x float> poison, float [[TMP12]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT30:%.*]] = shufflevector <1 x float> [[SPLAT_SPLATINSERT29]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP13:%.*]] = call fast <1 x float> @llvm.fmuladd.v1f32(<1 x float> [[BLOCK28]], <1 x float> [[SPLAT_SPLAT30]], <1 x float> [[TMP11]])
-; CHECK-NEXT:    [[TMP14:%.*]] = shufflevector <1 x float> [[TMP13]], <1 x float> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP15:%.*]] = shufflevector <1 x float> undef, <1 x float> [[TMP14]], <1 x i32> <i32 1>
-; CHECK-NEXT:    ret <1 x float> [[TMP15]]
+; CHECK-NEXT:    [[LHS:%.*]] = load <7 x float>, ptr [[LHS_ADDRESS:%.*]], align 32
+; CHECK-NEXT:    [[RHS:%.*]] = load <7 x float>, ptr [[RHS_ADDRESS:%.*]], align 32
+; CHECK-NEXT:    [[TMP0:%.*]] = fmul <7 x float> [[LHS]], [[RHS]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast float @llvm.vector.reduce.fadd.v7f32(float 0.000000e+00, <7 x float> [[TMP0]])
+; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <1 x float> poison, float [[TMP1]], i64 0
+; CHECK-NEXT:    ret <1 x float> [[TMP2]]
 ;
 entry:
   %lhs = load <7 x float>, ptr %lhs_address
   %rhs = load <7 x float>, ptr %rhs_address
   %c = tail call fast <1 x float> @llvm.matrix.multiply.v1f32.v7f32.v7f32(<7 x float> %lhs, <7 x float> %rhs, i32 1, i32 7, i32 1)
   ret <1 x float> %c
 }
 
 declare <1 x float> @llvm.matrix.multiply.v1f32.v7f32.v7f32(<7 x float>, <7 x float>, i32, i32, i32)
 
 define <1 x double> @dotproduct_double_v6(<6 x double> %a, <6 x double> %b) {
 ; CHECK-LABEL: @dotproduct_double_v6(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[SPLIT:%.*]] = shufflevector <6 x double> [[A:%.*]], <6 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[SPLIT1:%.*]] = shufflevector <6 x double> [[A]], <6 x double> poison, <1 x i32> <i32 1>
-; CHECK-NEXT:    [[SPLIT2:%.*]] = shufflevector <6 x double> [[A]], <6 x double> poison, <1 x i32> <i32 2>
-; CHECK-NEXT:    [[SPLIT3:%.*]] = shufflevector <6 x double> [[A]], <6 x double> poison, <1 x i32> <i32 3>
-; CHECK-NEXT:    [[SPLIT4:%.*]] = shufflevector <6 x double> [[A]], <6 x double> poison, <1 x i32> <i32 4>
-; CHECK-NEXT:    [[SPLIT5:%.*]] = shufflevector <6 x double> [[A]], <6 x double> poison, <1 x i32> <i32 5>
-; CHECK-NEXT:    [[SPLIT6:%.*]] = shufflevector <6 x double> [[B:%.*]], <6 x double> poison, <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5>
-; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <1 x double> [[SPLIT]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP0:%.*]] = extractelement <6 x double> [[SPLIT6]], i64 0
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x double> poison, double [[TMP0]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP1:%.*]] = fmul fast <1 x double> [[BLOCK]], [[SPLAT_SPLAT]]
-; CHECK-NEXT:    [[BLOCK7:%.*]] = shufflevector <1 x double> [[SPLIT1]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <6 x double> [[SPLIT6]], i64 1
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT8:%.*]] = insertelement <1 x double> poison, double [[TMP2]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT9:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT8]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP3:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK7]], <1 x double> [[SPLAT_SPLAT9]], <1 x double> [[TMP1]])
-; CHECK-NEXT:    [[BLOCK10:%.*]] = shufflevector <1 x double> [[SPLIT2]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <6 x double> [[SPLIT6]], i64 2
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT11:%.*]] = insertelement <1 x double> poison, double [[TMP4]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT12:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT11]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP5:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK10]], <1 x double> [[SPLAT_SPLAT12]], <1 x double> [[TMP3]])
-; CHECK-NEXT:    [[BLOCK13:%.*]] = shufflevector <1 x double> [[SPLIT3]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <6 x double> [[SPLIT6]], i64 3
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT14:%.*]] = insertelement <1 x double> poison, double [[TMP6]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT15:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT14]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP7:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK13]], <1 x double> [[SPLAT_SPLAT15]], <1 x double> [[TMP5]])
-; CHECK-NEXT:    [[BLOCK16:%.*]] = shufflevector <1 x double> [[SPLIT4]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <6 x double> [[SPLIT6]], i64 4
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT17:%.*]] = insertelement <1 x double> poison, double [[TMP8]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT18:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT17]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP9:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK16]], <1 x double> [[SPLAT_SPLAT18]], <1 x double> [[TMP7]])
-; CHECK-NEXT:    [[BLOCK19:%.*]] = shufflevector <1 x double> [[SPLIT5]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <6 x double> [[SPLIT6]], i64 5
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <1 x double> poison, double [[TMP10]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT20]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP11:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK19]], <1 x double> [[SPLAT_SPLAT21]], <1 x double> [[TMP9]])
-; CHECK-NEXT:    [[TMP12:%.*]] = shufflevector <1 x double> [[TMP11]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP13:%.*]] = shufflevector <1 x double> undef, <1 x double> [[TMP12]], <1 x i32> <i32 1>
-; CHECK-NEXT:    ret <1 x double> [[TMP13]]
+; CHECK-NEXT:    [[TMP0:%.*]] = fmul <6 x double> [[A:%.*]], [[B:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast double @llvm.vector.reduce.fadd.v6f64(double 0.000000e+00, <6 x double> [[TMP0]])
+; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <1 x double> poison, double [[TMP1]], i64 0
+; CHECK-NEXT:    ret <1 x double> [[TMP2]]
 ;
 entry:
   %c = tail call fast <1 x double> @llvm.matrix.multiply.v1f64.v6f64.v6f64(<6 x double> %a, <6 x double> %b, i32 1, i32 6, i32 1)
   ret <1 x double> %c
 }
 
 declare <1 x double> @llvm.matrix.multiply.v1f64.v6f64.v6f64(<6 x double>, <6 x double>, i32, i32, i32)
 
 define <1 x double> @intrinsic_column_major_load_dot_product_double_v6(ptr %lhs_address, ptr %rhs_address) {
 ; CHECK-LABEL: @intrinsic_column_major_load_dot_product_double_v6(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <6 x double>, ptr [[LHS_ADDRESS:%.*]], align 4
-; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <1 x double>, ptr [[RHS_ADDRESS:%.*]], align 4
-; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr double, ptr [[RHS_ADDRESS]], i64 6
-; CHECK-NEXT:    [[COL_LOAD2:%.*]] = load <1 x double>, ptr [[VEC_GEP]], align 4
-; CHECK-NEXT:    [[VEC_GEP3:%.*]] = getelementptr double, ptr [[RHS_ADDRESS]], i64 12
-; CHECK-NEXT:    [[COL_LOAD4:%.*]] = load <1 x double>, ptr [[VEC_GEP3]], align 4
-; CHECK-NEXT:    [[VEC_GEP5:%.*]] = getelementptr double, ptr [[RHS_ADDRESS]], i64 18
-; CHECK-NEXT:    [[COL_LOAD6:%.*]] = load <1 x double>, ptr [[VEC_GEP5]], align 4
-; CHECK-NEXT:    [[VEC_GEP7:%.*]] = getelementptr double, ptr [[RHS_ADDRESS]], i64 24
-; CHECK-NEXT:    [[COL_LOAD8:%.*]] = load <1 x double>, ptr [[VEC_GEP7]], align 4
-; CHECK-NEXT:    [[VEC_GEP9:%.*]] = getelementptr double, ptr [[RHS_ADDRESS]], i64 30
-; CHECK-NEXT:    [[COL_LOAD10:%.*]] = load <1 x double>, ptr [[VEC_GEP9]], align 4
-; CHECK-NEXT:    [[SPLIT:%.*]] = shufflevector <6 x double> [[COL_LOAD]], <6 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[SPLIT11:%.*]] = shufflevector <6 x double> [[COL_LOAD]], <6 x double> poison, <1 x i32> <i32 1>
-; CHECK-NEXT:    [[SPLIT12:%.*]] = shufflevector <6 x double> [[COL_LOAD]], <6 x double> poison, <1 x i32> <i32 2>
-; CHECK-NEXT:    [[SPLIT13:%.*]] = shufflevector <6 x double> [[COL_LOAD]], <6 x double> poison, <1 x i32> <i32 3>
-; CHECK-NEXT:    [[SPLIT14:%.*]] = shufflevector <6 x double> [[COL_LOAD]], <6 x double> poison, <1 x i32> <i32 4>
-; CHECK-NEXT:    [[SPLIT15:%.*]] = shufflevector <6 x double> [[COL_LOAD]], <6 x double> poison, <1 x i32> <i32 5>
-; CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <1 x double> [[COL_LOAD1]], <1 x double> [[COL_LOAD2]], <2 x i32> <i32 0, i32 1>
-; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <1 x double> [[COL_LOAD4]], <1 x double> [[COL_LOAD6]], <2 x i32> <i32 0, i32 1>
-; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <1 x double> [[COL_LOAD8]], <1 x double> [[COL_LOAD10]], <2 x i32> <i32 0, i32 1>
-; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <2 x double> [[TMP0]], <2 x double> [[TMP1]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEXT:    [[TMP4:%.*]] = shufflevector <2 x double> [[TMP2]], <2 x double> poison, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
-; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <4 x double> [[TMP3]], <4 x double> [[TMP4]], <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5>
-; CHECK-NEXT:    [[SPLIT16:%.*]] = shufflevector <6 x double> [[TMP5]], <6 x double> poison, <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5>
-; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <1 x double> [[SPLIT]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <6 x double> [[SPLIT16]], i64 0
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x double> poison, double [[TMP6]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP7:%.*]] = fmul fast <1 x double> [[BLOCK]], [[SPLAT_SPLAT]]
-; CHECK-NEXT:    [[BLOCK17:%.*]] = shufflevector <1 x double> [[SPLIT11]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <6 x double> [[SPLIT16]], i64 1
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT18:%.*]] = insertelement <1 x double> poison, double [[TMP8]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT19:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT18]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP9:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK17]], <1 x double> [[SPLAT_SPLAT19]], <1 x double> [[TMP7]])
-; CHECK-NEXT:    [[BLOCK20:%.*]] = shufflevector <1 x double> [[SPLIT12]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <6 x double> [[SPLIT16]], i64 2
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT21:%.*]] = insertelement <1 x double> poison, double [[TMP10]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT22:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT21]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP11:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK20]], <1 x double> [[SPLAT_SPLAT22]], <1 x double> [[TMP9]])
-; CHECK-NEXT:    [[BLOCK23:%.*]] = shufflevector <1 x double> [[SPLIT13]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <6 x double> [[SPLIT16]], i64 3
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT24:%.*]] = insertelement <1 x double> poison, double [[TMP12]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT25:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT24]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP13:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK23]], <1 x double> [[SPLAT_SPLAT25]], <1 x double> [[TMP11]])
-; CHECK-NEXT:    [[BLOCK26:%.*]] = shufflevector <1 x double> [[SPLIT14]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <6 x double> [[SPLIT16]], i64 4
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT27:%.*]] = insertelement <1 x double> poison, double [[TMP14]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT28:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT27]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP15:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK26]], <1 x double> [[SPLAT_SPLAT28]], <1 x double> [[TMP13]])
-; CHECK-NEXT:    [[BLOCK29:%.*]] = shufflevector <1 x double> [[SPLIT15]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP16:%.*]] = extractelement <6 x double> [[SPLIT16]], i64 5
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT30:%.*]] = insertelement <1 x double> poison, double [[TMP16]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT31:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT30]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP17:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK29]], <1 x double> [[SPLAT_SPLAT31]], <1 x double> [[TMP15]])
-; CHECK-NEXT:    [[TMP18:%.*]] = shufflevector <1 x double> [[TMP17]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP19:%.*]] = shufflevector <1 x double> undef, <1 x double> [[TMP18]], <1 x i32> <i32 1>
-; CHECK-NEXT:    ret <1 x double> [[TMP19]]
+; CHECK-NEXT:    [[TMP0:%.*]] = load <6 x double>, ptr [[LHS_ADDRESS:%.*]], align 64
+; CHECK-NEXT:    [[TMP1:%.*]] = load <6 x double>, ptr [[RHS_ADDRESS:%.*]], align 64
+; CHECK-NEXT:    [[TMP2:%.*]] = fmul <6 x double> [[TMP0]], [[TMP1]]
+; CHECK-NEXT:    [[TMP3:%.*]] = call fast double @llvm.vector.reduce.fadd.v6f64(double 0.000000e+00, <6 x double> [[TMP2]])
+; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <1 x double> poison, double [[TMP3]], i64 0
+; CHECK-NEXT:    ret <1 x double> [[TMP4]]
 ;
 entry:
   %lhs = tail call fast <6 x double> @llvm.matrix.column.major.load.v6f64.i64(ptr nonnull align 4 %lhs_address, i64 6, i1 false, i32 6, i32 1)
   %rhs = tail call fast <6 x double> @llvm.matrix.column.major.load.v6f64.i64(ptr nonnull align 4 %rhs_address, i64 6, i1 false, i32 1, i32 6)
   %result = tail call fast <1 x double> @llvm.matrix.multiply.v1f64.v6f64.v6f64(<6 x double> %lhs, <6 x double> %rhs, i32 1, i32 6, i32 1)
   ret <1 x double> %result
 }
 
 declare <6 x double> @llvm.matrix.column.major.load.v6f64.i64(ptr nonnull align 4, i64, i1, i32, i32)
 
 define <1 x double> @LoadInst_dot_product_double_v7(ptr %lhs_address, ptr %rhs_address) {
 ; CHECK-LABEL: @LoadInst_dot_product_double_v7(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <1 x double>, ptr [[LHS_ADDRESS:%.*]], align 64
-; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr double, ptr [[LHS_ADDRESS]], i64 1
-; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <1 x double>, ptr [[VEC_GEP]], align 8
-; CHECK-NEXT:    [[VEC_GEP2:%.*]] = getelementptr double, ptr [[LHS_ADDRESS]], i64 2
-; CHECK-NEXT:    [[COL_LOAD3:%.*]] = load <1 x double>, ptr [[VEC_GEP2]], align 16
-; CHECK-NEXT:    [[VEC_GEP4:%.*]] = getelementptr double, ptr [[LHS_ADDRESS]], i64 3
-; CHECK-NEXT:    [[COL_LOAD5:%.*]] = load <1 x double>, ptr [[VEC_GEP4]], align 8
-; CHECK-NEXT:    [[VEC_GEP6:%.*]] = getelementptr double, ptr [[LHS_ADDRESS]], i64 4
-; CHECK-NEXT:    [[COL_LOAD7:%.*]] = load <1 x double>, ptr [[VEC_GEP6]], align 32
-; CHECK-NEXT:    [[VEC_GEP8:%.*]] = getelementptr double, ptr [[LHS_ADDRESS]], i64 5
-; CHECK-NEXT:    [[COL_LOAD9:%.*]] = load <1 x double>, ptr [[VEC_GEP8]], align 8
-; CHECK-NEXT:    [[VEC_GEP10:%.*]] = getelementptr double, ptr [[LHS_ADDRESS]], i64 6
-; CHECK-NEXT:    [[COL_LOAD11:%.*]] = load <1 x double>, ptr [[VEC_GEP10]], align 16
-; CHECK-NEXT:    [[COL_LOAD12:%.*]] = load <7 x double>, ptr [[RHS_ADDRESS:%.*]], align 64
-; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <1 x double> [[COL_LOAD]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP0:%.*]] = extractelement <7 x double> [[COL_LOAD12]], i64 0
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x double> poison, double [[TMP0]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP1:%.*]] = fmul fast <1 x double> [[BLOCK]], [[SPLAT_SPLAT]]
-; CHECK-NEXT:    [[BLOCK13:%.*]] = shufflevector <1 x double> [[COL_LOAD1]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <7 x double> [[COL_LOAD12]], i64 1
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT14:%.*]] = insertelement <1 x double> poison, double [[TMP2]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT15:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT14]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP3:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK13]], <1 x double> [[SPLAT_SPLAT15]], <1 x double> [[TMP1]])
-; CHECK-NEXT:    [[BLOCK16:%.*]] = shufflevector <1 x double> [[COL_LOAD3]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP4:%.*]] = extractelement <7 x double> [[COL_LOAD12]], i64 2
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT17:%.*]] = insertelement <1 x double> poison, double [[TMP4]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT18:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT17]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP5:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK16]], <1 x double> [[SPLAT_SPLAT18]], <1 x double> [[TMP3]])
-; CHECK-NEXT:    [[BLOCK19:%.*]] = shufflevector <1 x double> [[COL_LOAD5]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <7 x double> [[COL_LOAD12]], i64 3
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <1 x double> poison, double [[TMP6]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT20]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP7:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK19]], <1 x double> [[SPLAT_SPLAT21]], <1 x double> [[TMP5]])
-; CHECK-NEXT:    [[BLOCK22:%.*]] = shufflevector <1 x double> [[COL_LOAD7]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <7 x double> [[COL_LOAD12]], i64 4
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT23:%.*]] = insertelement <1 x double> poison, double [[TMP8]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT24:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT23]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP9:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK22]], <1 x double> [[SPLAT_SPLAT24]], <1 x double> [[TMP7]])
-; CHECK-NEXT:    [[BLOCK25:%.*]] = shufflevector <1 x double> [[COL_LOAD9]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP10:%.*]] = extractelement <7 x double> [[COL_LOAD12]], i64 5
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT26:%.*]] = insertelement <1 x double> poison, double [[TMP10]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT27:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT26]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP11:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK25]], <1 x double> [[SPLAT_SPLAT27]], <1 x double> [[TMP9]])
-; CHECK-NEXT:    [[BLOCK28:%.*]] = shufflevector <1 x double> [[COL_LOAD11]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <7 x double> [[COL_LOAD12]], i64 6
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT29:%.*]] = insertelement <1 x double> poison, double [[TMP12]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT30:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT29]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP13:%.*]] = call fast <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK28]], <1 x double> [[SPLAT_SPLAT30]], <1 x double> [[TMP11]])
-; CHECK-NEXT:    [[TMP14:%.*]] = shufflevector <1 x double> [[TMP13]], <1 x double> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP15:%.*]] = shufflevector <1 x double> undef, <1 x double> [[TMP14]], <1 x i32> <i32 1>
-; CHECK-NEXT:    ret <1 x double> [[TMP15]]
+; CHECK-NEXT:    [[LHS:%.*]] = load <7 x double>, ptr [[LHS_ADDRESS:%.*]], align 64
+; CHECK-NEXT:    [[RHS:%.*]] = load <7 x double>, ptr [[RHS_ADDRESS:%.*]], align 64
+; CHECK-NEXT:    [[TMP0:%.*]] = fmul <7 x double> [[LHS]], [[RHS]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call fast double @llvm.vector.reduce.fadd.v7f64(double 0.000000e+00, <7 x double> [[TMP0]])
+; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <1 x double> poison, double [[TMP1]], i64 0
+; CHECK-NEXT:    ret <1 x double> [[TMP2]]
 ;
 entry:
   %lhs = load <7 x double>, ptr %lhs_address
   %rhs = load <7 x double>, ptr %rhs_address
   %c = tail call fast <1 x double> @llvm.matrix.multiply.v1f64.v7f64.v7f64(<7 x double> %lhs, <7 x double> %rhs, i32 1, i32 7, i32 1)
   ret <1 x double> %c
 }
 
 declare <1 x double> @llvm.matrix.multiply.v1f64.v7f64.v7f64(<7 x double>, <7 x double>, i32, i32, i32)
 
 define <1 x i64> @dotproduct_i64_v8(<8 x i64> %a, <8 x i64> %b) {
 ; CHECK-LABEL: @dotproduct_i64_v8(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[SPLIT:%.*]] = shufflevector <8 x i64> [[A:%.*]], <8 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[SPLIT1:%.*]] = shufflevector <8 x i64> [[A]], <8 x i64> poison, <1 x i32> <i32 1>
-; CHECK-NEXT:    [[SPLIT2:%.*]] = shufflevector <8 x i64> [[A]], <8 x i64> poison, <1 x i32> <i32 2>
-; CHECK-NEXT:    [[SPLIT3:%.*]] = shufflevector <8 x i64> [[A]], <8 x i64> poison, <1 x i32> <i32 3>
-; CHECK-NEXT:    [[SPLIT4:%.*]] = shufflevector <8 x i64> [[A]], <8 x i64> poison, <1 x i32> <i32 4>
-; CHECK-NEXT:    [[SPLIT5:%.*]] = shufflevector <8 x i64> [[A]], <8 x i64> poison, <1 x i32> <i32 5>
-; CHECK-NEXT:    [[SPLIT6:%.*]] = shufflevector <8 x i64> [[A]], <8 x i64> poison, <1 x i32> <i32 6>
-; CHECK-NEXT:    [[SPLIT7:%.*]] = shufflevector <8 x i64> [[A]], <8 x i64> poison, <1 x i32> <i32 7>
-; CHECK-NEXT:    [[SPLIT8:%.*]] = shufflevector <8 x i64> [[B:%.*]], <8 x i64> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <1 x i64> [[SPLIT]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP0:%.*]] = extractelement <8 x i64> [[SPLIT8]], i64 0
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x i64> poison, i64 [[TMP0]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x i64> [[SPLAT_SPLATINSERT]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP1:%.*]] = mul <1 x i64> [[BLOCK]], [[SPLAT_SPLAT]]
-; CHECK-NEXT:    [[BLOCK9:%.*]] = shufflevector <1 x i64> [[SPLIT1]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <8 x i64> [[SPLIT8]], i64 1
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT10:%.*]] = insertelement <1 x i64> poison, i64 [[TMP2]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT11:%.*]] = shufflevector <1 x i64> [[SPLAT_SPLATINSERT10]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP3:%.*]] = mul <1 x i64> [[BLOCK9]], [[SPLAT_SPLAT11]]
-; CHECK-NEXT:    [[TMP4:%.*]] = add <1 x i64> [[TMP1]], [[TMP3]]
-; CHECK-NEXT:    [[BLOCK12:%.*]] = shufflevector <1 x i64> [[SPLIT2]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <8 x i64> [[SPLIT8]], i64 2
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT13:%.*]] = insertelement <1 x i64> poison, i64 [[TMP5]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT14:%.*]] = shufflevector <1 x i64> [[SPLAT_SPLATINSERT13]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP6:%.*]] = mul <1 x i64> [[BLOCK12]], [[SPLAT_SPLAT14]]
-; CHECK-NEXT:    [[TMP7:%.*]] = add <1 x i64> [[TMP4]], [[TMP6]]
-; CHECK-NEXT:    [[BLOCK15:%.*]] = shufflevector <1 x i64> [[SPLIT3]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <8 x i64> [[SPLIT8]], i64 3
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT16:%.*]] = insertelement <1 x i64> poison, i64 [[TMP8]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT17:%.*]] = shufflevector <1 x i64> [[SPLAT_SPLATINSERT16]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP9:%.*]] = mul <1 x i64> [[BLOCK15]], [[SPLAT_SPLAT17]]
-; CHECK-NEXT:    [[TMP10:%.*]] = add <1 x i64> [[TMP7]], [[TMP9]]
-; CHECK-NEXT:    [[BLOCK18:%.*]] = shufflevector <1 x i64> [[SPLIT4]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP11:%.*]] = extractelement <8 x i64> [[SPLIT8]], i64 4
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT19:%.*]] = insertelement <1 x i64> poison, i64 [[TMP11]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT20:%.*]] = shufflevector <1 x i64> [[SPLAT_SPLATINSERT19]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP12:%.*]] = mul <1 x i64> [[BLOCK18]], [[SPLAT_SPLAT20]]
-; CHECK-NEXT:    [[TMP13:%.*]] = add <1 x i64> [[TMP10]], [[TMP12]]
-; CHECK-NEXT:    [[BLOCK21:%.*]] = shufflevector <1 x i64> [[SPLIT5]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <8 x i64> [[SPLIT8]], i64 5
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT22:%.*]] = insertelement <1 x i64> poison, i64 [[TMP14]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT23:%.*]] = shufflevector <1 x i64> [[SPLAT_SPLATINSERT22]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP15:%.*]] = mul <1 x i64> [[BLOCK21]], [[SPLAT_SPLAT23]]
-; CHECK-NEXT:    [[TMP16:%.*]] = add <1 x i64> [[TMP13]], [[TMP15]]
-; CHECK-NEXT:    [[BLOCK24:%.*]] = shufflevector <1 x i64> [[SPLIT6]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP17:%.*]] = extractelement <8 x i64> [[SPLIT8]], i64 6
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT25:%.*]] = insertelement <1 x i64> poison, i64 [[TMP17]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT26:%.*]] = shufflevector <1 x i64> [[SPLAT_SPLATINSERT25]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP18:%.*]] = mul <1 x i64> [[BLOCK24]], [[SPLAT_SPLAT26]]
-; CHECK-NEXT:    [[TMP19:%.*]] = add <1 x i64> [[TMP16]], [[TMP18]]
-; CHECK-NEXT:    [[BLOCK27:%.*]] = shufflevector <1 x i64> [[SPLIT7]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP20:%.*]] = extractelement <8 x i64> [[SPLIT8]], i64 7
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT28:%.*]] = insertelement <1 x i64> poison, i64 [[TMP20]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT29:%.*]] = shufflevector <1 x i64> [[SPLAT_SPLATINSERT28]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP21:%.*]] = mul <1 x i64> [[BLOCK27]], [[SPLAT_SPLAT29]]
-; CHECK-NEXT:    [[TMP22:%.*]] = add <1 x i64> [[TMP19]], [[TMP21]]
-; CHECK-NEXT:    [[TMP23:%.*]] = shufflevector <1 x i64> [[TMP22]], <1 x i64> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP24:%.*]] = shufflevector <1 x i64> undef, <1 x i64> [[TMP23]], <1 x i32> <i32 1>
-; CHECK-NEXT:    ret <1 x i64> [[TMP24]]
+; CHECK-NEXT:    [[TMP0:%.*]] = mul <8 x i64> [[A:%.*]], [[B:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call i64 @llvm.vector.reduce.add.v8i64(<8 x i64> [[TMP0]])
+; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <1 x i64> poison, i64 [[TMP1]], i64 0
+; CHECK-NEXT:    ret <1 x i64> [[TMP2]]
 ;
 entry:
   %c = tail call <1 x i64> @llvm.matrix.multiply.v1i64.v8i64.v8i64(<8 x i64> %a, <8 x i64> %b, i32 1, i32 8, i32 1)
   ret <1 x i64> %c
 }
 
 declare <1 x i64> @llvm.matrix.multiply.v1i64.v8i64.v8i64(<8 x i64>, <8 x i64>, i32, i32, i32)
 
 define <1 x i32> @intrinsic_column_major_load_dot_product_i32_v8(ptr %lhs_address, ptr %rhs_address) {
 ; CHECK-LABEL: @intrinsic_column_major_load_dot_product_i32_v8(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <8 x i32>, ptr [[LHS_ADDRESS:%.*]], align 4
-; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <1 x i32>, ptr [[RHS_ADDRESS:%.*]], align 4
-; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr i32, ptr [[RHS_ADDRESS]], i64 8
-; CHECK-NEXT:    [[COL_LOAD2:%.*]] = load <1 x i32>, ptr [[VEC_GEP]], align 4
-; CHECK-NEXT:    [[VEC_GEP3:%.*]] = getelementptr i32, ptr [[RHS_ADDRESS]], i64 16
-; CHECK-NEXT:    [[COL_LOAD4:%.*]] = load <1 x i32>, ptr [[VEC_GEP3]], align 4
-; CHECK-NEXT:    [[VEC_GEP5:%.*]] = getelementptr i32, ptr [[RHS_ADDRESS]], i64 24
-; CHECK-NEXT:    [[COL_LOAD6:%.*]] = load <1 x i32>, ptr [[VEC_GEP5]], align 4
-; CHECK-NEXT:    [[VEC_GEP7:%.*]] = getelementptr i32, ptr [[RHS_ADDRESS]], i64 32
-; CHECK-NEXT:    [[COL_LOAD8:%.*]] = load <1 x i32>, ptr [[VEC_GEP7]], align 4
-; CHECK-NEXT:    [[VEC_GEP9:%.*]] = getelementptr i32, ptr [[RHS_ADDRESS]], i64 40
-; CHECK-NEXT:    [[COL_LOAD10:%.*]] = load <1 x i32>, ptr [[VEC_GEP9]], align 4
-; CHECK-NEXT:    [[VEC_GEP11:%.*]] = getelementptr i32, ptr [[RHS_ADDRESS]], i64 48
-; CHECK-NEXT:    [[COL_LOAD12:%.*]] = load <1 x i32>, ptr [[VEC_GEP11]], align 4
-; CHECK-NEXT:    [[VEC_GEP13:%.*]] = getelementptr i32, ptr [[RHS_ADDRESS]], i64 56
-; CHECK-NEXT:    [[COL_LOAD14:%.*]] = load <1 x i32>, ptr [[VEC_GEP13]], align 4
-; CHECK-NEXT:    [[SPLIT:%.*]] = shufflevector <8 x i32> [[COL_LOAD]], <8 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[SPLIT15:%.*]] = shufflevector <8 x i32> [[COL_LOAD]], <8 x i32> poison, <1 x i32> <i32 1>
-; CHECK-NEXT:    [[SPLIT16:%.*]] = shufflevector <8 x i32> [[COL_LOAD]], <8 x i32> poison, <1 x i32> <i32 2>
-; CHECK-NEXT:    [[SPLIT17:%.*]] = shufflevector <8 x i32> [[COL_LOAD]], <8 x i32> poison, <1 x i32> <i32 3>
-; CHECK-NEXT:    [[SPLIT18:%.*]] = shufflevector <8 x i32> [[COL_LOAD]], <8 x i32> poison, <1 x i32> <i32 4>
-; CHECK-NEXT:    [[SPLIT19:%.*]] = shufflevector <8 x i32> [[COL_LOAD]], <8 x i32> poison, <1 x i32> <i32 5>
-; CHECK-NEXT:    [[SPLIT20:%.*]] = shufflevector <8 x i32> [[COL_LOAD]], <8 x i32> poison, <1 x i32> <i32 6>
-; CHECK-NEXT:    [[SPLIT21:%.*]] = shufflevector <8 x i32> [[COL_LOAD]], <8 x i32> poison, <1 x i32> <i32 7>
-; CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <1 x i32> [[COL_LOAD1]], <1 x i32> [[COL_LOAD2]], <2 x i32> <i32 0, i32 1>
-; CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <1 x i32> [[COL_LOAD4]], <1 x i32> [[COL_LOAD6]], <2 x i32> <i32 0, i32 1>
-; CHECK-NEXT:    [[TMP2:%.*]] = shufflevector <1 x i32> [[COL_LOAD8]], <1 x i32> [[COL_LOAD10]], <2 x i32> <i32 0, i32 1>
-; CHECK-NEXT:    [[TMP3:%.*]] = shufflevector <1 x i32> [[COL_LOAD12]], <1 x i32> [[COL_LOAD14]], <2 x i32> <i32 0, i32 1>
-; CHECK-NEXT:    [[TMP4:%.*]] = shufflevector <2 x i32> [[TMP0]], <2 x i32> [[TMP1]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEXT:    [[TMP5:%.*]] = shufflevector <2 x i32> [[TMP2]], <2 x i32> [[TMP3]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
-; CHECK-NEXT:    [[TMP6:%.*]] = shufflevector <4 x i32> [[TMP4]], <4 x i32> [[TMP5]], <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEXT:    [[SPLIT22:%.*]] = shufflevector <8 x i32> [[TMP6]], <8 x i32> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
-; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <1 x i32> [[SPLIT]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <8 x i32> [[SPLIT22]], i64 0
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x i32> poison, i32 [[TMP7]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP8:%.*]] = mul <1 x i32> [[BLOCK]], [[SPLAT_SPLAT]]
-; CHECK-NEXT:    [[BLOCK23:%.*]] = shufflevector <1 x i32> [[SPLIT15]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP9:%.*]] = extractelement <8 x i32> [[SPLIT22]], i64 1
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT24:%.*]] = insertelement <1 x i32> poison, i32 [[TMP9]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT25:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT24]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP10:%.*]] = mul <1 x i32> [[BLOCK23]], [[SPLAT_SPLAT25]]
-; CHECK-NEXT:    [[TMP11:%.*]] = add <1 x i32> [[TMP8]], [[TMP10]]
-; CHECK-NEXT:    [[BLOCK26:%.*]] = shufflevector <1 x i32> [[SPLIT16]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP12:%.*]] = extractelement <8 x i32> [[SPLIT22]], i64 2
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT27:%.*]] = insertelement <1 x i32> poison, i32 [[TMP12]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT28:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT27]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP13:%.*]] = mul <1 x i32> [[BLOCK26]], [[SPLAT_SPLAT28]]
-; CHECK-NEXT:    [[TMP14:%.*]] = add <1 x i32> [[TMP11]], [[TMP13]]
-; CHECK-NEXT:    [[BLOCK29:%.*]] = shufflevector <1 x i32> [[SPLIT17]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP15:%.*]] = extractelement <8 x i32> [[SPLIT22]], i64 3
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT30:%.*]] = insertelement <1 x i32> poison, i32 [[TMP15]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT31:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT30]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP16:%.*]] = mul <1 x i32> [[BLOCK29]], [[SPLAT_SPLAT31]]
-; CHECK-NEXT:    [[TMP17:%.*]] = add <1 x i32> [[TMP14]], [[TMP16]]
-; CHECK-NEXT:    [[BLOCK32:%.*]] = shufflevector <1 x i32> [[SPLIT18]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP18:%.*]] = extractelement <8 x i32> [[SPLIT22]], i64 4
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT33:%.*]] = insertelement <1 x i32> poison, i32 [[TMP18]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT34:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT33]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP19:%.*]] = mul <1 x i32> [[BLOCK32]], [[SPLAT_SPLAT34]]
-; CHECK-NEXT:    [[TMP20:%.*]] = add <1 x i32> [[TMP17]], [[TMP19]]
-; CHECK-NEXT:    [[BLOCK35:%.*]] = shufflevector <1 x i32> [[SPLIT19]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP21:%.*]] = extractelement <8 x i32> [[SPLIT22]], i64 5
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT36:%.*]] = insertelement <1 x i32> poison, i32 [[TMP21]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT37:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT36]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP22:%.*]] = mul <1 x i32> [[BLOCK35]], [[SPLAT_SPLAT37]]
-; CHECK-NEXT:    [[TMP23:%.*]] = add <1 x i32> [[TMP20]], [[TMP22]]
-; CHECK-NEXT:    [[BLOCK38:%.*]] = shufflevector <1 x i32> [[SPLIT20]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP24:%.*]] = extractelement <8 x i32> [[SPLIT22]], i64 6
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT39:%.*]] = insertelement <1 x i32> poison, i32 [[TMP24]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT40:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT39]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP25:%.*]] = mul <1 x i32> [[BLOCK38]], [[SPLAT_SPLAT40]]
-; CHECK-NEXT:    [[TMP26:%.*]] = add <1 x i32> [[TMP23]], [[TMP25]]
-; CHECK-NEXT:    [[BLOCK41:%.*]] = shufflevector <1 x i32> [[SPLIT21]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP27:%.*]] = extractelement <8 x i32> [[SPLIT22]], i64 7
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT42:%.*]] = insertelement <1 x i32> poison, i32 [[TMP27]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT43:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT42]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP28:%.*]] = mul <1 x i32> [[BLOCK41]], [[SPLAT_SPLAT43]]
-; CHECK-NEXT:    [[TMP29:%.*]] = add <1 x i32> [[TMP26]], [[TMP28]]
-; CHECK-NEXT:    [[TMP30:%.*]] = shufflevector <1 x i32> [[TMP29]], <1 x i32> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP31:%.*]] = shufflevector <1 x i32> undef, <1 x i32> [[TMP30]], <1 x i32> <i32 1>
-; CHECK-NEXT:    ret <1 x i32> [[TMP31]]
+; CHECK-NEXT:    [[TMP0:%.*]] = load <8 x i32>, ptr [[LHS_ADDRESS:%.*]], align 32
+; CHECK-NEXT:    [[TMP1:%.*]] = load <8 x i32>, ptr [[RHS_ADDRESS:%.*]], align 32
+; CHECK-NEXT:    [[TMP2:%.*]] = mul <8 x i32> [[TMP0]], [[TMP1]]
+; CHECK-NEXT:    [[TMP3:%.*]] = call i32 @llvm.vector.reduce.add.v8i32(<8 x i32> [[TMP2]])
+; CHECK-NEXT:    [[TMP4:%.*]] = insertelement <1 x i32> poison, i32 [[TMP3]], i64 0
+; CHECK-NEXT:    ret <1 x i32> [[TMP4]]
 ;
 entry:
   %lhs = tail call <8 x i32> @llvm.matrix.column.major.load.v8i32.i64(ptr nonnull align 4 %lhs_address, i64 8, i1 false, i32 8, i32 1)
   %rhs = tail call <8 x i32> @llvm.matrix.column.major.load.v8i32.i64(ptr nonnull align 4 %rhs_address, i64 8, i1 false, i32 1, i32 8)
   %result = tail call <1 x i32> @llvm.matrix.multiply.v1i32.v8i32.v8i32(<8 x i32> %lhs, <8 x i32> %rhs, i32 1, i32 8, i32 1)
   ret <1 x i32> %result
 }
 
 declare <8 x i32> @llvm.matrix.column.major.load.v8i32.i64(ptr nonnull align 4, i64, i1, i32, i32)
 declare <1 x i32> @llvm.matrix.multiply.v1i32.v8i32.v8i32(<8 x i32>, <8 x i32>, i32, i32, i32)
 
 define <1 x i16> @LoadInst_dot_product_i16_v6(ptr %lhs_address, ptr %rhs_address) {
 ; CHECK-LABEL: @LoadInst_dot_product_i16_v6(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[COL_LOAD:%.*]] = load <1 x i16>, ptr [[LHS_ADDRESS:%.*]], align 16
-; CHECK-NEXT:    [[VEC_GEP:%.*]] = getelementptr i16, ptr [[LHS_ADDRESS]], i64 1
-; CHECK-NEXT:    [[COL_LOAD1:%.*]] = load <1 x i16>, ptr [[VEC_GEP]], align 2
-; CHECK-NEXT:    [[VEC_GEP2:%.*]] = getelementptr i16, ptr [[LHS_ADDRESS]], i64 2
-; CHECK-NEXT:    [[COL_LOAD3:%.*]] = load <1 x i16>, ptr [[VEC_GEP2]], align 4
-; CHECK-NEXT:    [[VEC_GEP4:%.*]] = getelementptr i16, ptr [[LHS_ADDRESS]], i64 3
-; CHECK-NEXT:    [[COL_LOAD5:%.*]] = load <1 x i16>, ptr [[VEC_GEP4]], align 2
-; CHECK-NEXT:    [[VEC_GEP6:%.*]] = getelementptr i16, ptr [[LHS_ADDRESS]], i64 4
-; CHECK-NEXT:    [[COL_LOAD7:%.*]] = load <1 x i16>, ptr [[VEC_GEP6]], align 8
-; CHECK-NEXT:    [[VEC_GEP8:%.*]] = getelementptr i16, ptr [[LHS_ADDRESS]], i64 5
-; CHECK-NEXT:    [[COL_LOAD9:%.*]] = load <1 x i16>, ptr [[VEC_GEP8]], align 2
-; CHECK-NEXT:    [[COL_LOAD10:%.*]] = load <6 x i16>, ptr [[RHS_ADDRESS:%.*]], align 16
-; CHECK-NEXT:    [[BLOCK:%.*]] = shufflevector <1 x i16> [[COL_LOAD]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP0:%.*]] = extractelement <6 x i16> [[COL_LOAD10]], i64 0
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x i16> poison, i16 [[TMP0]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x i16> [[SPLAT_SPLATINSERT]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP1:%.*]] = mul <1 x i16> [[BLOCK]], [[SPLAT_SPLAT]]
-; CHECK-NEXT:    [[BLOCK11:%.*]] = shufflevector <1 x i16> [[COL_LOAD1]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP2:%.*]] = extractelement <6 x i16> [[COL_LOAD10]], i64 1
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT12:%.*]] = insertelement <1 x i16> poison, i16 [[TMP2]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT13:%.*]] = shufflevector <1 x i16> [[SPLAT_SPLATINSERT12]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP3:%.*]] = mul <1 x i16> [[BLOCK11]], [[SPLAT_SPLAT13]]
-; CHECK-NEXT:    [[TMP4:%.*]] = add <1 x i16> [[TMP1]], [[TMP3]]
-; CHECK-NEXT:    [[BLOCK14:%.*]] = shufflevector <1 x i16> [[COL_LOAD3]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP5:%.*]] = extractelement <6 x i16> [[COL_LOAD10]], i64 2
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT15:%.*]] = insertelement <1 x i16> poison, i16 [[TMP5]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT16:%.*]] = shufflevector <1 x i16> [[SPLAT_SPLATINSERT15]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP6:%.*]] = mul <1 x i16> [[BLOCK14]], [[SPLAT_SPLAT16]]
-; CHECK-NEXT:    [[TMP7:%.*]] = add <1 x i16> [[TMP4]], [[TMP6]]
-; CHECK-NEXT:    [[BLOCK17:%.*]] = shufflevector <1 x i16> [[COL_LOAD5]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP8:%.*]] = extractelement <6 x i16> [[COL_LOAD10]], i64 3
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT18:%.*]] = insertelement <1 x i16> poison, i16 [[TMP8]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT19:%.*]] = shufflevector <1 x i16> [[SPLAT_SPLATINSERT18]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP9:%.*]] = mul <1 x i16> [[BLOCK17]], [[SPLAT_SPLAT19]]
-; CHECK-NEXT:    [[TMP10:%.*]] = add <1 x i16> [[TMP7]], [[TMP9]]
-; CHECK-NEXT:    [[BLOCK20:%.*]] = shufflevector <1 x i16> [[COL_LOAD7]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP11:%.*]] = extractelement <6 x i16> [[COL_LOAD10]], i64 4
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT21:%.*]] = insertelement <1 x i16> poison, i16 [[TMP11]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT22:%.*]] = shufflevector <1 x i16> [[SPLAT_SPLATINSERT21]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP12:%.*]] = mul <1 x i16> [[BLOCK20]], [[SPLAT_SPLAT22]]
-; CHECK-NEXT:    [[TMP13:%.*]] = add <1 x i16> [[TMP10]], [[TMP12]]
-; CHECK-NEXT:    [[BLOCK23:%.*]] = shufflevector <1 x i16> [[COL_LOAD9]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP14:%.*]] = extractelement <6 x i16> [[COL_LOAD10]], i64 5
-; CHECK-NEXT:    [[SPLAT_SPLATINSERT24:%.*]] = insertelement <1 x i16> poison, i16 [[TMP14]], i32 0
-; CHECK-NEXT:    [[SPLAT_SPLAT25:%.*]] = shufflevector <1 x i16> [[SPLAT_SPLATINSERT24]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP15:%.*]] = mul <1 x i16> [[BLOCK23]], [[SPLAT_SPLAT25]]
-; CHECK-NEXT:    [[TMP16:%.*]] = add <1 x i16> [[TMP13]], [[TMP15]]
-; CHECK-NEXT:    [[TMP17:%.*]] = shufflevector <1 x i16> [[TMP16]], <1 x i16> poison, <1 x i32> zeroinitializer
-; CHECK-NEXT:    [[TMP18:%.*]] = shufflevector <1 x i16> undef, <1 x i16> [[TMP17]], <1 x i32> <i32 1>
-; CHECK-NEXT:    ret <1 x i16> [[TMP18]]
+; CHECK-NEXT:    [[LHS:%.*]] = load <6 x i16>, ptr [[LHS_ADDRESS:%.*]], align 16
+; CHECK-NEXT:    [[RHS:%.*]] = load <6 x i16>, ptr [[RHS_ADDRESS:%.*]], align 16
+; CHECK-NEXT:    [[TMP0:%.*]] = mul <6 x i16> [[LHS]], [[RHS]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call i16 @llvm.vector.reduce.add.v6i16(<6 x i16> [[TMP0]])
+; CHECK-NEXT:    [[TMP2:%.*]] = insertelement <1 x i16> poison, i16 [[TMP1]], i64 0
+; CHECK-NEXT:    ret <1 x i16> [[TMP2]]
 ;
 entry:
   %lhs = load <6 x i16>, ptr %lhs_address
   %rhs = load <6 x i16>, ptr %rhs_address
   %result = tail call <1 x i16> @llvm.matrix.multiply.v1i16.v6i16.v6i16(<6 x i16> %lhs, <6 x i16> %rhs, i32 1, i32 6, i32 1)
   ret <1 x i16> %result
 }
 
 declare <1 x i16> @llvm.matrix.multiply.v1i16.v6i16.v6i16(<6 x i16>, <6 x i16>, i32, i32, i32)