[Matrix] Add option to use row-major matrix layout as default.

[fhahn]

Summary

This patch adds a -matrix-default-layout option which can be used to
set the default matrix layout to row-major or column-major (default).

The initial patch updates codegen for loads, stores, binary operators
and matrix multiply.

TODO: additional tests.

[fhahn]

Rebased on top of D75566.

[anemet]

TODO: additional tests.

Is this WIP?

llvm/lib/Transforms/Scalar/LowerMatrixIntrinsics.cpp

322

column vector?

783

is this computeColumnAddr or computeVectorAddr now?

984–988

This interface got me confused in particular in the context of this patch. Comment says tile-sized but then there is a flag whether it's tiled at all. Anyway we can improve this?

Sounds like this should be just called emitMatrixMultiply? You might want to insert a patch before this one that fixes this.

[fhahn]

In D76325#1948804, @anemet wrote:

TODO: additional tests.

Is this WIP?

Originally it was missing tests for binary operators/loads/stores. Added now additional tests now.

Update renaming function names/docs only referring to column vectors. O

[fhahn]

Adjust documentation in a few more places, add assertions to make sure operands of binary ops agree on the layout.

[fhahn]

llvm/lib/Transforms/Scalar/LowerMatrixIntrinsics.cpp

783

Fix and also in various other places.

984–988

Fixed in rG9e81249d7614

[anemet]

LGTM!

llvm/lib/Transforms/Scalar/LowerMatrixIntrinsics.cpp

 static cl::opt<bool> ForceFusion(
     "force-fuse-matrix", cl::init(false), cl::Hidden,
     cl::desc("Force matrix instruction fusion even if not profitable."));
 static cl::opt<bool> AllowContractEnabled(
     "matrix-allow-contract", cl::init(false), cl::Hidden,
     cl::desc("Allow the use of FMAs if available and profitable. This may "
              "result in different results, due to less rounding error."));
 
+enum class MatrixLayoutTy { ColumnMajor, RowMajor };
+
+static cl::opt<MatrixLayoutTy> MatrixLayout(
+    "matrix-default-layout", cl::init(MatrixLayoutTy::ColumnMajor),
+    cl::desc("Sets the default matrix layout"),
+    cl::values(clEnumValN(MatrixLayoutTy::ColumnMajor, "column-major",
+                          "Use column-major layout"),
+               clEnumValN(MatrixLayoutTy::RowMajor, "row-major",
+                          "Use row-major layout")));
+
 /// Helper function to either return Scope, if it is a subprogram or the
 /// attached subprogram for a local scope.
 static DISubprogram *getSubprogram(DIScope *Scope) {
   if (auto *Subprogram = dyn_cast<DISubprogram>(Scope))
     return Subprogram;
   return cast<DILocalScope>(Scope)->getSubprogram();
 }
 
 namespace {
 
-// Given an element poitner \p BasePtr to the start of a (sub) matrix, compute
-// the start address of column \p Col with type (\p EltType x \p NumRows)
-// assuming \p Stride elements between start two consecutive columns.
-// \p Stride must be >= \p NumRows.
+// Given an element pointer \p BasePtr to the start of a (sub) matrix, compute
+// the start address of vector \p VecIdx with type (\p EltType x \p NumElements)
+// assuming \p Stride elements between start two consecutive vectors.
+// \p Stride must be >= \p NumElements.
+// For column-major matrixes, the function computes the address of a column
+// vectors and \p NumElements must be set to the number of elements in a column
+// (= number of rows of the matrix). For row-major matrixes, the function
+// computes the address of a row vector and \p NumElements must be set to the
+// number of elements in a column (= number of columns of the matrix).
 //
-// Consider a 4x4 matrix like below
+// Consider a 4x4 matrix in column-mjaor layout like below
 //
 //      0       1      2      3
 // 0   v_0_0  v_0_1  v_0_2  v_0_3
 // 1   v_1_0  v_1_1  v_1_2  v_1_3
 // 2   v_2_0  v_2_1  v_2_2  v_2_3
 // 3   v_3_0  v_3_1  v_3_2  v_3_3
 
 // To compute the column addresses for a 2x3 sub-matrix at row 1 and column 1,
 // we need a pointer to the first element of the submatrix as base pointer.
-// Then we can use computeColumnAddr to compute the addresses for the columns
+// Then we can use computeVectorAddr to compute the addresses for the columns
 // of the sub-matrix.
 //
-// Column 0: computeColumnAddr(Base, 0 (column), 4 (stride), 2 (num rows), ..)
+// Column 0: computeVectorAddr(Base, 0 (column), 4 (stride), 2 (num rows), ..)
 //           -> just returns Base
-// Column 1: computeColumnAddr(Base, 1 (column), 4 (stride), 2 (num rows), ..)
+// Column 1: computeVectorAddr(Base, 1 (column), 4 (stride), 2 (num rows), ..)
 //           -> returns Base + (1 * 4)
-// Column 2: computeColumnAddr(Base, 2 (column), 4 (stride), 2 (num rows), ..)
+// Column 2: computeVectorAddr(Base, 2 (column), 4 (stride), 2 (num rows), ..)
 //           -> returns Base + (2 * 4)
 //
 // The graphic below illustrates the number of elements in a column (marked
 // with |) and the number of skipped elements (marked with }).
 //
 //         v_0_0  v_0_1 {v_0_2 {v_0_3
 //                Base   Col 1  Col 2
 //                  |     |      |
 //         v_1_0 |v_1_1 |v_1_2 |v_1_3
 //         v_2_0 |v_2_1 |v_2_2 |v_2_3
 //         v_3_0 {v_3_1 {v_3_2  v_3_3
 //
-Value *computeColumnAddr(Value *BasePtr, Value *Col, Value *Stride,
-                         unsigned NumRows, Type *EltType,
+Value *computeVectorAddr(Value *BasePtr, Value *VecIdx, Value *Stride,
+                         unsigned NumElements, Type *EltType,
                          IRBuilder<> &Builder) {
 
   assert((!isa<ConstantInt>(Stride) ||
-          cast<ConstantInt>(Stride)->getZExtValue() >= NumRows) &&
-         "Stride must be >= the number of rows.");
+          cast<ConstantInt>(Stride)->getZExtValue() >= NumElements) &&
+         "Stride must be >= the number of elements in the result vector.");
   unsigned AS = cast<PointerType>(BasePtr->getType())->getAddressSpace();
 
-  // Compute the start of the column with index Col as Col * Stride.
-  Value *ColumnStart = Builder.CreateMul(Col, Stride, "col.start");
+  // Compute the start of the vector with index VecIdx as VecIdx * Stride.
+  Value *VecStart = Builder.CreateMul(VecIdx, Stride, "vec.start");
 
-  // Get pointer to the start of the selected column. Skip GEP creation,
-  // if we select column 0.
-  if (isa<ConstantInt>(ColumnStart) && cast<ConstantInt>(ColumnStart)->isZero())
-    ColumnStart = BasePtr;
+  // Get pointer to the start of the selected vector. Skip GEP creation,
+  // if we select vector 0.
+  if (isa<ConstantInt>(VecStart) && cast<ConstantInt>(VecStart)->isZero())
+    VecStart = BasePtr;
   else
-    ColumnStart = Builder.CreateGEP(EltType, BasePtr, ColumnStart, "col.gep");
+    VecStart = Builder.CreateGEP(EltType, BasePtr, VecStart, "vec.gep");
 
-  // Cast elementwise column start pointer to a pointer to a column
-  // (EltType x NumRows)*.
-  Type *ColumnType = VectorType::get(EltType, NumRows);
-  Type *ColumnPtrType = PointerType::get(ColumnType, AS);
-  return Builder.CreatePointerCast(ColumnStart, ColumnPtrType, "col.cast");
+  // Cast elementwise vector start pointer to a pointer to a vector
+  // (EltType x NumElements)*.
+  Type *VecType = VectorType::get(EltType, NumElements);
+  Type *VecPtrType = PointerType::get(VecType, AS);
+  return Builder.CreatePointerCast(VecStart, VecPtrType, "vec.cast");
 }
 
 /// LowerMatrixIntrinsics contains the methods used to lower matrix intrinsics.
 ///
 /// Currently, the lowering for each matrix intrinsic is done as follows:
 /// 1. Propagate the shape information from intrinsics to connected
 /// instructions.
-/// 2. Lower instructions with shape information.
+/// 2. Lower instructions with shape information (assuming column-major layout).
+///  The lowering works similarly using row-major layout.
 ///  2.1. Get column vectors for each argument. If we already lowered the
 ///       definition of an argument, use the produced column vectors directly.
 ///       If not, split the operand vector containing an embedded matrix into
 ///       a set of column vectors,
 ///  2.2. Lower the instruction in terms of columnwise operations, which yields
 ///       a set of column vectors containing result matrix. Note that we lower
 ///       all instructions that have shape information. Besides the intrinsics,
 ///       this includes stores for example.
class LowerMatrixIntrinsics {
   class MatrixTy {
     SmallVector<Value *, 16> Vectors;
 
     OpInfoTy OpInfo;
 
     bool IsColumnMajor = true;
 
   public:
-    MatrixTy() : Vectors() {}
+    MatrixTy()
+        : Vectors(),
+          IsColumnMajor(MatrixLayout == MatrixLayoutTy::ColumnMajor) {}
     MatrixTy(ArrayRef<Value *> Vectors)
-        : Vectors(Vectors.begin(), Vectors.end()) {}
+        : Vectors(Vectors.begin(), Vectors.end()),
+          IsColumnMajor(MatrixLayout == MatrixLayoutTy::ColumnMajor) {}
+    MatrixTy(unsigned NumRows, unsigned NumColumns, Type *EltTy)
+        : IsColumnMajor(MatrixLayout == MatrixLayoutTy::ColumnMajor) {
+
+      unsigned D = isColumnMajor() ? NumColumns : NumRows;
+      for (unsigned J = 0; J < D; ++J)
+        addVector(UndefValue::get(
+            VectorType::get(EltTy, isColumnMajor() ? NumRows : NumColumns)));
+    }
 
     Value *getVector(unsigned i) const { return Vectors[i]; }
     Value *getColumn(unsigned i) const {
       assert(isColumnMajor() && "only supported for column-major matrixes");
       return Vectors[i];
     }
+    Value *getRow(unsigned i) const {
+      assert(!isColumnMajor() && "only supported for row-major matrixes");
+      return Vectors[i];
+    }
 
-    void setColumn(unsigned i, Value *V) { Vectors[i] = V; }
+    void setVector(unsigned i, Value *V) { Vectors[i] = V; }
 
     Type *getElementType() { return getVectorTy()->getElementType(); }
 
+    unsigned getNumVectors() const {
+      if (isColumnMajor())
+        return getNumColumns();
+      return getNumRows();
+    }
+
     unsigned getNumColumns() const {
       if (isColumnMajor())
         return Vectors.size();
       else {
         assert(Vectors.size() > 0 && "Cannot call getNumRows without columns");
         return cast<VectorType>(Vectors[0]->getType())->getNumElements();
       }
     }
     unsigned getNumRows() const {
       if (isColumnMajor()) {
         assert(Vectors.size() > 0 && "Cannot call getNumRows without columns");
         return cast<VectorType>(Vectors[0]->getType())->getNumElements();
       } else
         return Vectors.size();
     }
 
-    const SmallVectorImpl<Value *> &getColumnVectors() const { return Vectors; }
-
-    SmallVectorImpl<Value *> &getColumnVectors() { return Vectors; }
-
-    void addColumn(Value *V) { Vectors.push_back(V); }
-
+    void addVector(Value *V) { Vectors.push_back(V); }
     VectorType *getColumnTy() {
       assert(isColumnMajor() && "only supported for column-major matrixes");
       return getVectorTy();
     }
 
     VectorType *getVectorTy() {
       return cast<VectorType>(Vectors[0]->getType());
     }
 
     iterator_range<SmallVector<Value *, 8>::iterator> columns() {
+      assert(isColumnMajor() &&
+             "columns() only supported for column-major matrixes");
       return make_range(Vectors.begin(), Vectors.end());
     }
 
-    /// Embed the columns of the matrix into a flat vector by concatenating
+    iterator_range<SmallVector<Value *, 8>::iterator> vectors() {
+      return make_range(Vectors.begin(), Vectors.end());
+    }
+
+    /// Embed the vectors of the matrix into a flat vector by concatenating
     /// them.
     Value *embedInVector(IRBuilder<> &Builder) const {
       return Vectors.size() == 1 ? Vectors[0]
                                  : concatenateVectors(Builder, Vectors);
     }
 
     MatrixTy &addNumLoads(unsigned N) {
       OpInfo.NumLoads += N;
public:
 
     unsigned getNumStores() const { return OpInfo.NumStores; }
     unsigned getNumLoads() const { return OpInfo.NumLoads; }
     unsigned getNumComputeOps() const { return OpInfo.NumComputeOps; }
 
     const OpInfoTy &getOpInfo() const { return OpInfo; }
 
     bool isColumnMajor() const { return IsColumnMajor; }
+
+    unsigned getStride() const {
+      if (isColumnMajor())
+        return getNumRows();
+      return getNumColumns();
+    }
+
+    /// Extract a vector of \p NumElts starting at index (\p I, \p J). If the
+    /// matrix is column-major, the result vector is extracted from a column
+    /// vector, otherwise from a row vector.
+    Value *extractVector(unsigned I, unsigned J, unsigned NumElts,
+                         IRBuilder<> &Builder) const {
+      Value *Vec = isColumnMajor() ? getColumn(J) : getRow(I);
+      Value *Undef = UndefValue::get(Vec->getType());
+      Constant *Mask =
+          createSequentialMask(Builder, isColumnMajor() ? I : J, NumElts, 0);
+      return Builder.CreateShuffleVector(Vec, Undef, Mask, "block");
+    }
   };
 
   struct ShapeInfo {
     unsigned NumRows;
     unsigned NumColumns;
 
+    bool IsColumnMajor;
+
     ShapeInfo(unsigned NumRows = 0, unsigned NumColumns = 0)
-        : NumRows(NumRows), NumColumns(NumColumns) {}
+        : NumRows(NumRows), NumColumns(NumColumns),
+          IsColumnMajor(MatrixLayout == MatrixLayoutTy::ColumnMajor) {}
 
     ShapeInfo(Value *NumRows, Value *NumColumns)
-        : NumRows(cast<ConstantInt>(NumRows)->getZExtValue()),
-          NumColumns(cast<ConstantInt>(NumColumns)->getZExtValue()) {}
+        : ShapeInfo(cast<ConstantInt>(NumRows)->getZExtValue(),
+                    cast<ConstantInt>(NumColumns)->getZExtValue()) {}
 
     bool operator==(const ShapeInfo &other) {
       return NumRows == other.NumRows && NumColumns == other.NumColumns;
     }
     bool operator!=(const ShapeInfo &other) { return !(*this == other); }
 
     /// Returns true if shape-information is defined, meaning both dimensions
     /// are != 0.
     operator bool() const {
       assert(NumRows == 0 || NumColumns != 0);
       return NumRows != 0;
     }
+
+    unsigned getStride() const {
+      if (IsColumnMajor)
+        return NumRows;
+      return NumColumns;
+    }
+
+    unsigned getNumVectors() const {
+      if (IsColumnMajor)
+        return NumColumns;
+      return NumRows;
+    }
   };
 
   /// Maps instructions to their shape information. The shape information
   /// describes the shape to be used while lowering. This matches the shape of
   /// the result value of the instruction, with the only exceptions being store
   /// instructions and the matrix_columnwise_store intrinsics. For those, the
   /// shape information indicates that those instructions should be lowered
   /// using shape information as well.
public:
   //
   /// Return the estimated number of vector ops required for an operation on
   /// \p VT * N.
   unsigned getNumOps(Type *ST, unsigned N) {
     return std::ceil((ST->getPrimitiveSizeInBits() * N).getFixedSize() /
                      double(TTI.getRegisterBitWidth(true)));
   }
 
-  /// Return the set of column vectors that a matrix value is lowered to.
+  /// Return the set of vectors that a matrix value is lowered to.
   ///
-  /// If we lowered \p MatrixVal, just return the cache result column matrix.
-  /// Otherwie split the flat vector \p MatrixVal containing a matrix with
-  /// shape \p SI into column vectors.
+  /// If we lowered \p MatrixVal, just return the cache result matrix. Otherwise
+  /// split the flat vector \p MatrixVal containing a matrix with shape \p SI
+  /// into vectors.
   MatrixTy getMatrix(Value *MatrixVal, const ShapeInfo &SI,
                      IRBuilder<> &Builder) {
     VectorType *VType = dyn_cast<VectorType>(MatrixVal->getType());
     assert(VType && "MatrixVal must be a vector type");
     assert(VType->getNumElements() == SI.NumRows * SI.NumColumns &&
            "The vector size must match the number of matrix elements");
 
     // Check if we lowered MatrixVal using shape information. In that case,
-    // return the existing column matrix, if it matches the requested shape
+    // return the existing matrix, if it matches the requested shape
     // information. If there is a mis-match, embed the result in a flat
     // vector and split it later.
     auto Found = Inst2ColumnMatrix.find(MatrixVal);
     if (Found != Inst2ColumnMatrix.end()) {
       MatrixTy &M = Found->second;
       // Return the found matrix, if its shape matches the requested shape
       // information
       if (SI.NumRows == M.getNumRows() && SI.NumColumns == M.getNumColumns())
         return M;
 
       MatrixVal = M.embedInVector(Builder);
     }
 
     // Otherwise split MatrixVal.
     SmallVector<Value *, 16> SplitVecs;
     Value *Undef = UndefValue::get(VType);
     for (unsigned MaskStart = 0; MaskStart < VType->getNumElements();
-         MaskStart += SI.NumRows) {
-      Constant *Mask = createSequentialMask(Builder, MaskStart, SI.NumRows, 0);
+         MaskStart += SI.getStride()) {
+      Constant *Mask =
+          createSequentialMask(Builder, MaskStart, SI.getStride(), 0);
       Value *V = Builder.CreateShuffleVector(MatrixVal, Undef, Mask, "split");
       SplitVecs.push_back(V);
     }
 
     return {SplitVecs};
   }
 
   /// If \p V already has a known shape return false.  Otherwise set the shape
bool Visit() {
     RemarkGen.emitRemarks();
 
     for (Instruction *Inst : reverse(ToRemove))
       Inst->eraseFromParent();
 
     return Changed;
   }
 
-  LoadInst *createColumnLoad(Value *ColumnPtr, Type *EltType,
+  LoadInst *createVectorLoad(Value *ColumnPtr, Type *EltType,
                              IRBuilder<> &Builder) {
     return Builder.CreateAlignedLoad(
         ColumnPtr, Align(DL.getABITypeAlignment(EltType)), "col.load");
   }
 
-  StoreInst *createColumnStore(Value *ColumnValue, Value *ColumnPtr,
+  StoreInst *createVectorStore(Value *ColumnValue, Value *ColumnPtr,
                                Type *EltType, IRBuilder<> &Builder) {
     return Builder.CreateAlignedStore(ColumnValue, ColumnPtr,
                                       DL.getABITypeAlign(EltType));
   }
 
-
   /// Turns \p BasePtr into an elementwise pointer to \p EltType.
   Value *createElementPtr(Value *BasePtr, Type *EltType, IRBuilder<> &Builder) {
     unsigned AS = cast<PointerType>(BasePtr->getType())->getAddressSpace();
     Type *EltPtrType = PointerType::get(EltType, AS);
     return Builder.CreatePointerCast(BasePtr, EltPtrType);
   }
 
   /// Replace intrinsic calls
case Intrinsic::matrix_columnwise_store:
       break;
     default:
       return false;
     }
     return true;
   }
 
   /// Load a matrix with \p Shape starting at \p Ptr and using \p Stride between
-  /// columns.
+  /// vectors.
   MatrixTy loadMatrix(Type *Ty, Value *Ptr, Value *Stride, ShapeInfo Shape,
                       IRBuilder<> &Builder) {
     auto VType = cast<VectorType>(Ty);
     Value *EltPtr = createElementPtr(Ptr, VType->getElementType(), Builder);
     MatrixTy Result;
-    // Distance between start of one column and the start of the next
-    for (unsigned C = 0, E = Shape.NumColumns; C < E; ++C) {
-      Value *GEP =
-          computeColumnAddr(EltPtr, Builder.getInt32(C), Stride, Shape.NumRows,
-                            VType->getElementType(), Builder);
-      Value *Column = createColumnLoad(GEP, VType->getElementType(), Builder);
-      Result.addColumn(Column);
+    for (unsigned I = 0, E = Shape.getNumVectors(); I < E; ++I) {
+      Value *GEP = computeVectorAddr(EltPtr, Builder.getInt32(I), Stride,
+                                     Shape.getStride(), VType->getElementType(),
+                                     Builder);
+      Value *Vector = createVectorLoad(GEP, VType->getElementType(), Builder);
+      Result.addVector(Vector);
     }
-    return Result.addNumLoads(getNumOps(Result.getColumnTy()) *
-                              Result.getNumColumns());
+    return Result.addNumLoads(getNumOps(Result.getVectorTy()) *
+                              Result.getNumVectors());
   }
 
   /// Loads a sub-matrix with shape \p ResultShape from a \p R x \p C matrix,
   /// starting at \p MatrixPtr[I][J].
   MatrixTy loadMatrix(Value *MatrixPtr, ShapeInfo MatrixShape, unsigned I,
                       unsigned J, ShapeInfo ResultShape, Type *EltTy,
                       IRBuilder<> &Builder) {
 
     Value *Offset = Builder.CreateAdd(
         Builder.CreateMul(Builder.getInt32(J),
-                          Builder.getInt32(MatrixShape.NumRows)),
+                          Builder.getInt32(MatrixShape.getStride())),
         Builder.getInt32(I));
 
     unsigned AS = cast<PointerType>(MatrixPtr->getType())->getAddressSpace();
     Value *EltPtr =
         Builder.CreatePointerCast(MatrixPtr, PointerType::get(EltTy, AS));
     Value *TileStart = Builder.CreateGEP(EltTy, EltPtr, Offset);
     Type *TileTy =
         VectorType::get(EltTy, ResultShape.NumRows * ResultShape.NumColumns);
     Type *TilePtrTy = PointerType::get(TileTy, AS);
     Value *TilePtr =
         Builder.CreatePointerCast(TileStart, TilePtrTy, "col.cast");
 
-    return loadMatrix(TileTy, TilePtr, Builder.getInt32(MatrixShape.NumRows),
-                      ResultShape, Builder);
+    return loadMatrix(TileTy, TilePtr,
+                      Builder.getInt32(MatrixShape.getStride()), ResultShape,
+                      Builder);
   }
 
   /// Lower a load instruction with shape information.
   void LowerLoad(Instruction *Inst, Value *Ptr, Value *Stride,
                  ShapeInfo Shape) {
     IRBuilder<> Builder(Inst);
     finalizeLowering(Inst,
                      loadMatrix(Inst->getType(), Ptr, Stride, Shape, Builder),
                      Builder);
   }
 
   /// Lowers llvm.matrix.columnwise.load.
   ///
   /// The intrinsic loads a matrix from memory using a stride between columns.
   void LowerColumnwiseLoad(CallInst *Inst) {
+    assert(MatrixLayout == MatrixLayoutTy::ColumnMajor &&
+           "Intrinsic only supports column-major layout!");
     Value *Ptr = Inst->getArgOperand(0);
     Value *Stride = Inst->getArgOperand(1);
     LowerLoad(Inst, Ptr, Stride,
               {Inst->getArgOperand(2), Inst->getArgOperand(3)});
   }
 
   /// Stores a sub-matrix \p StoreVal into the \p R x \p C matrix starting at \p
   /// MatrixPtr[I][J].
   void storeMatrix(const MatrixTy &StoreVal, Value *MatrixPtr,
                    ShapeInfo MatrixShape, unsigned I, unsigned J, Type *EltTy,
                    IRBuilder<> &Builder) {
     Value *Offset = Builder.CreateAdd(
         Builder.CreateMul(Builder.getInt32(J),
-                          Builder.getInt32(MatrixShape.NumRows)),
+                          Builder.getInt32(MatrixShape.getStride())),
         Builder.getInt32(I));
 
     unsigned AS = cast<PointerType>(MatrixPtr->getType())->getAddressSpace();
     Value *EltPtr =
         Builder.CreatePointerCast(MatrixPtr, PointerType::get(EltTy, AS));
     Value *TileStart = Builder.CreateGEP(EltTy, EltPtr, Offset);
     Type *TileTy = VectorType::get(EltTy, StoreVal.getNumRows() *
                                               StoreVal.getNumColumns());
     Type *TilePtrTy = PointerType::get(TileTy, AS);
     Value *TilePtr =
         Builder.CreatePointerCast(TileStart, TilePtrTy, "col.cast");
 
     storeMatrix(TileTy, StoreVal, TilePtr,
-                Builder.getInt32(MatrixShape.NumRows), Builder);
+                Builder.getInt32(MatrixShape.getStride()), Builder);
   }
 
   /// Store matrix \p StoreVal starting at \p Ptr and using \p Stride between
-  /// columns.
+  /// vectors.
   MatrixTy storeMatrix(Type *Ty, MatrixTy StoreVal, Value *Ptr, Value *Stride,
                        IRBuilder<> &Builder) {
     auto VType = cast<VectorType>(Ty);
     Value *EltPtr = createElementPtr(Ptr, VType->getElementType(), Builder);
-    for (auto C : enumerate(StoreVal.columns())) {
-      Value *GEP = computeColumnAddr(EltPtr, Builder.getInt32(C.index()),
-                                     Stride, StoreVal.getNumRows(),
+    for (auto Vec : enumerate(StoreVal.vectors())) {
+      Value *GEP = computeVectorAddr(EltPtr, Builder.getInt32(Vec.index()),
+                                     Stride, StoreVal.getStride(),
                                      VType->getElementType(), Builder);
-      createColumnStore(C.value(), GEP, VType->getElementType(), Builder);
+      createVectorStore(Vec.value(), GEP, VType->getElementType(), Builder);
     }
-    return MatrixTy().addNumStores(getNumOps(StoreVal.getColumnTy()) *
-                                   StoreVal.getNumColumns());
+    return MatrixTy().addNumStores(getNumOps(StoreVal.getVectorTy()) *
+                                   StoreVal.getNumVectors());
   }
 
   /// Lower a store instruction with shape information.
   void LowerStore(Instruction *Inst, Value *Matrix, Value *Ptr, Value *Stride,
                   ShapeInfo Shape) {
     IRBuilder<> Builder(Inst);
     auto StoreVal = getMatrix(Matrix, Shape, Builder);
     finalizeLowering(
         Inst, storeMatrix(Matrix->getType(), StoreVal, Ptr, Stride, Builder),
         Builder);
   }
 
   /// Lowers llvm.matrix.columnwise.store.
   ///
   /// The intrinsic store a matrix back memory using a stride between columns.
   void LowerColumnwiseStore(CallInst *Inst) {
+    assert(MatrixLayout == MatrixLayoutTy::ColumnMajor &&
+           "Intrinsic only supports column-major layout!");
     Value *Matrix = Inst->getArgOperand(0);
     Value *Ptr = Inst->getArgOperand(1);
     Value *Stride = Inst->getArgOperand(2);
     LowerStore(Inst, Matrix, Ptr, Stride,
                {Inst->getArgOperand(3), Inst->getArgOperand(4)});
   }
 
-  /// Extract a column vector of \p NumElts starting at index (\p I, \p J) from
-  /// the matrix \p LM represented as a vector of column vectors.
-  Value *extractVector(const MatrixTy &LM, unsigned I, unsigned J,
-                       unsigned NumElts, IRBuilder<> &Builder) {
-    Value *Col = LM.getColumn(J);
-    Value *Undef = UndefValue::get(Col->getType());
-    Constant *Mask = createSequentialMask(Builder, I, NumElts, 0);
-    return Builder.CreateShuffleVector(Col, Undef, Mask, "block");
-  }
-
   // Set elements I..I+NumElts-1 to Block
   Value *insertVector(Value *Col, unsigned I, Value *Block,
                       IRBuilder<> &Builder) {
 
     // First, bring Block to the same size as Col
     unsigned BlockNumElts =
         cast<VectorType>(Block->getType())->getNumElements();
     unsigned NumElts = cast<VectorType>(Col->getType())->getNumElements();
for (auto I = Inst->use_begin(), E = Inst->use_end(); I != E;) {
       if (ShapeMap.find(U.getUser()) == ShapeMap.end()) {
         if (!Flattened)
           Flattened = Matrix.embedInVector(Builder);
         U.set(Flattened);
       }
     }
   }
 
   /// Compute \p Result += \p A * \p B for input matrices with left-associating
   /// addition.
   void emitMatrixMultiply(MatrixTy &Result, const MatrixTy &A,
                           const MatrixTy &B, bool AllowContraction,
                           IRBuilder<> &Builder, bool isTiled) {
     const unsigned VF = std::max<unsigned>(
         TTI.getRegisterBitWidth(true) /
             Result.getElementType()->getPrimitiveSizeInBits().getFixedSize(),
         1U);
     unsigned R = Result.getNumRows();
     unsigned C = Result.getNumColumns();
     unsigned M = A.getNumColumns();
 
+    bool IsFP = Result.getElementType()->isFloatingPointTy();
+    assert(A.isColumnMajor() == B.isColumnMajor() &&
+           Result.isColumnMajor() == A.isColumnMajor() &&
+           "operands must agree on matrix layout");
+    unsigned NumComputeOps = 0;
+    if (A.isColumnMajor()) {
+      // Multiply columns from the first operand with scalars from the second
+      // operand. Then move along the K axes and accumulate the columns.  With
+      // this the adds can be vectorized without reassociation.
       for (unsigned J = 0; J < C; ++J) {
         unsigned BlockSize = VF;
-
         // If Result is zero, we don't need to accumulate in the K==0 iteration.
         bool isSumZero = isa<ConstantAggregateZero>(Result.getColumn(J));
 
-      unsigned NumOps = 0;
         for (unsigned I = 0; I < R; I += BlockSize) {
           // Gradually lower the vectorization factor to cover the remainder.
           while (I + BlockSize > R)
             BlockSize /= 2;
 
-        Value *Sum =
-            isTiled ? extractVector(Result, I, J, BlockSize, Builder) : nullptr;
+          Value *Sum = isTiled ? Result.extractVector(I, J, BlockSize, Builder)
+                               : nullptr;
           for (unsigned K = 0; K < M; ++K) {
-          Value *L = extractVector(A, I, K, BlockSize, Builder);
+            Value *L = A.extractVector(I, K, BlockSize, Builder);
             Value *RH = Builder.CreateExtractElement(B.getColumn(J), K);
             Value *Splat = Builder.CreateVectorSplat(BlockSize, RH, "splat");
             Sum = createMulAdd(isSumZero && K == 0 ? nullptr : Sum, L, Splat,
                                Result.getElementType()->isFloatingPointTy(),
-                             Builder, AllowContraction, NumOps);
+                               Builder, AllowContraction, NumComputeOps);
+          }
+          Result.setVector(J,
+                           insertVector(Result.getVector(J), I, Sum, Builder));
         }
-        Result.setColumn(J, insertVector(Result.getColumn(J), I, Sum, Builder));
       }
+    } else {
+      // Multiply rows from the second operand with scalars from the first
+      // operand. Then move along the K axes and accumulate the rows.  With this
+      // the adds can be vectorized without reassociation.
+      for (unsigned I = 0; I < R; ++I) {
+        unsigned BlockSize = VF;
+        bool isSumZero = isa<ConstantAggregateZero>(Result.getRow(I));
+        for (unsigned J = 0; J < C; J += BlockSize) {
+          // Gradually lower the vectorization factor to cover the remainder.
+          while (J + BlockSize > C)
+            BlockSize /= 2;
 
-      Result.addNumComputeOps(NumOps);
+          Value *Sum = nullptr;
+          for (unsigned K = 0; K < M; ++K) {
+            Value *R = B.extractVector(K, J, BlockSize, Builder);
+            Value *LH = Builder.CreateExtractElement(A.getVector(I), K);
+            Value *Splat = Builder.CreateVectorSplat(BlockSize, LH, "splat");
+            Sum = createMulAdd(isSumZero && K == 0 ? nullptr : Sum, Splat, R,
+                               IsFP, Builder, AllowContraction, NumComputeOps);
+          }
+          Result.setVector(I,
+                           insertVector(Result.getVector(I), J, Sum, Builder));
+        }
       }
     }
+    Result.addNumComputeOps(NumComputeOps);
+  }
 
   /// Ensure that the memory in \p Load does not alias \p Store by potentially
   /// copying it to a new location.  This new or otherwise the original location
   /// is returned.
   Value *getNonAliasingPointer(LoadInst *Load, StoreInst *Store,
                                CallInst *MatMul) {
     MemoryLocation StoreLoc = MemoryLocation::get(Store);
     MemoryLocation LoadLoc = MemoryLocation::get(Load);
bool isFusionProfitable(CallInst *MatMul) {
     unsigned Op1Regs = (M + VF - 1) / VF * C;
     return Op0Regs + Op1Regs > TTI.getNumberOfRegisters(true);
   }
 
   MatrixTy getZeroMatrix(Type *EltType, unsigned R, unsigned C) {
     MatrixTy Res;
     Type *ColumType = VectorType::get(EltType, R);
     for (unsigned I = 0; I < C; ++I)
-      Res.addColumn(ConstantAggregateZero::get(ColumType));
+      Res.addVector(ConstantAggregateZero::get(ColumType));
     return Res;
   }
 
   void emitSIMDTiling(CallInst *MatMul, LoadInst *LoadOp0, LoadInst *LoadOp1,
                       StoreInst *Store,
                       SmallPtrSetImpl<Instruction *> &FusedInsts) {
+    assert(MatrixLayout == MatrixLayoutTy::ColumnMajor &&
+           "Tiling only supported for column-major matrixes at the moment!");
     if (!isFusionProfitable(MatMul))
       return;
 
     ShapeInfo LShape(MatMul->getArgOperand(2), MatMul->getArgOperand(3));
     ShapeInfo RShape(MatMul->getArgOperand(3), MatMul->getArgOperand(4));
 
     const unsigned R = LShape.NumRows;
     const unsigned C = RShape.NumColumns;
public:
   ///
   /// Currently we only lower {ld, ld} -> matmul -> st chains.
   //
   /// No need to return a MatrixTy object for the result of the operation, since
   /// the single store user will be lowered as part of this. Instructions that
   /// are completely eliminated by fusion are added to \p FusedInsts.
   void LowerMatrixMultiplyFused(CallInst *MatMul,
                                 SmallPtrSetImpl<Instruction *> &FusedInsts) {
-    if (!FuseMatrix || !MatMul->hasOneUse())
+    if (!FuseMatrix || !MatMul->hasOneUse() ||
+        MatrixLayout != MatrixLayoutTy::ColumnMajor)
       return;
 
     auto *LoadOp0 = dyn_cast<LoadInst>(MatMul->getOperand(0));
     auto *LoadOp1 = dyn_cast<LoadInst>(MatMul->getOperand(1));
     auto *Store = dyn_cast<StoreInst>(*MatMul->user_begin());
     if (LoadOp0 && LoadOp1 && Store) {
       // The store address must dominate the MatMul instruction, otherwise
       // we create invalid IR.
void LowerMultiply(CallInst *MatMul) {
     const MatrixTy &Lhs = getMatrix(MatMul->getArgOperand(0), LShape, Builder);
     const MatrixTy &Rhs = getMatrix(MatMul->getArgOperand(1), RShape, Builder);
 
     const unsigned R = LShape.NumRows;
     const unsigned C = RShape.NumColumns;
     assert(LShape.NumColumns == RShape.NumRows);
 
     // Initialize the output
-    MatrixTy Result;
-    for (unsigned J = 0; J < C; ++J)
-      Result.addColumn(UndefValue::get(VectorType::get(EltType, R)));
+    MatrixTy Result(R, C, EltType);
 
     bool AllowContract = AllowContractEnabled || (isa<FPMathOperator>(MatMul) &&
                                                   MatMul->hasAllowContract());
     emitMatrixMultiply(Result, Lhs, Rhs, AllowContract, Builder, false);
     finalizeLowering(MatMul, Result, Builder);
   }
 
   /// Lowers llvm.matrix.transpose.
   void LowerTranspose(CallInst *Inst) {
     MatrixTy Result;
     IRBuilder<> Builder(Inst);
     Value *InputVal = Inst->getArgOperand(0);
     VectorType *VectorTy = cast<VectorType>(InputVal->getType());
     ShapeInfo ArgShape(Inst->getArgOperand(1), Inst->getArgOperand(2));
     MatrixTy InputMatrix = getMatrix(InputVal, ArgShape, Builder);
+    assert(InputMatrix.isColumnMajor() &&
+           "Row-major code-gen not supported yet!");
 
     for (unsigned Row = 0; Row < ArgShape.NumRows; ++Row) {
       // Build a single column vector for this row. First initialize it.
       Value *ResultColumn = UndefValue::get(
           VectorType::get(VectorTy->getElementType(), ArgShape.NumColumns));
 
       // Go through the elements of this row and insert it into the resulting
       // column vector.
       for (auto C : enumerate(InputMatrix.columns())) {
         Value *Elt = Builder.CreateExtractElement(C.value(), Row);
         // We insert at index Column since that is the row index after the
         // transpose.
         ResultColumn =
             Builder.CreateInsertElement(ResultColumn, Elt, C.index());
       }
-      Result.addColumn(ResultColumn);
+      Result.addVector(ResultColumn);
     }
 
     // TODO: Improve estimate of operations needed for transposes. Currently we
     // just count the insertelement/extractelement instructions, but do not
     // account for later simplifications/combines.
     finalizeLowering(
         Inst,
         Result.addNumComputeOps(2 * ArgShape.NumRows * ArgShape.NumColumns),
         Builder);
   }
 
   /// Lower load instructions, if shape information is available.
   bool VisitLoad(Instruction *Inst, Value *Ptr, IRBuilder<> &Builder) {
     auto I = ShapeMap.find(Inst);
     if (I == ShapeMap.end())
       return false;
 
-    LowerLoad(Inst, Ptr, Builder.getInt32(I->second.NumRows), I->second);
+    LowerLoad(Inst, Ptr, Builder.getInt32(I->second.getStride()), I->second);
     return true;
   }
 
   bool VisitStore(Instruction *Inst, Value *StoredVal, Value *Ptr,
                   IRBuilder<> &Builder) {
     auto I = ShapeMap.find(StoredVal);
     if (I == ShapeMap.end())
       return false;
 
-    LowerStore(Inst, StoredVal, Ptr, Builder.getInt32(I->second.NumRows), I->second);
+    LowerStore(Inst, StoredVal, Ptr, Builder.getInt32(I->second.getStride()),
+               I->second);
     return true;
   }
 
   /// Lower binary operators, if shape information is available.
   bool VisitBinaryOperator(BinaryOperator *Inst) {
     auto I = ShapeMap.find(Inst);
     if (I == ShapeMap.end())
       return false;
 
     Value *Lhs = Inst->getOperand(0);
     Value *Rhs = Inst->getOperand(1);
 
     IRBuilder<> Builder(Inst);
     ShapeInfo &Shape = I->second;
 
-    MatrixTy LoweredLhs = getMatrix(Lhs, Shape, Builder);
-    MatrixTy LoweredRhs = getMatrix(Rhs, Shape, Builder);
-
-    // Add each column and store the result back into the opmapping
     MatrixTy Result;
-    auto BuildColumnOp = [&Builder, Inst](Value *LHS, Value *RHS) {
+    MatrixTy A = getMatrix(Lhs, Shape, Builder);
+    MatrixTy B = getMatrix(Rhs, Shape, Builder);
+    assert(A.isColumnMajor() == B.isColumnMajor() &&
+           Result.isColumnMajor() == A.isColumnMajor() &&
+           "operands must agree on matrix layout");
+
+    // Helper to perform binary op on vectors.
+    auto BuildVectorOp = [&Builder, Inst](Value *LHS, Value *RHS) {
       switch (Inst->getOpcode()) {
       case Instruction::Add:
         return Builder.CreateAdd(LHS, RHS);
       case Instruction::Mul:
         return Builder.CreateMul(LHS, RHS);
       case Instruction::Sub:
         return Builder.CreateSub(LHS, RHS);
       case Instruction::FAdd:
         return Builder.CreateFAdd(LHS, RHS);
       case Instruction::FMul:
         return Builder.CreateFMul(LHS, RHS);
       case Instruction::FSub:
         return Builder.CreateFSub(LHS, RHS);
       default:
         llvm_unreachable("Unsupported binary operator for matrix");
       }
     };
-    for (unsigned C = 0; C < Shape.NumColumns; ++C)
-      Result.addColumn(
-          BuildColumnOp(LoweredLhs.getColumn(C), LoweredRhs.getColumn(C)));
+
+    for (unsigned I = 0; I < Shape.getNumVectors(); ++I)
+      Result.addVector(BuildVectorOp(A.getVector(I), B.getVector(I)));
 
     finalizeLowering(Inst,
-                     Result.addNumComputeOps(getNumOps(Result.getColumnTy()) *
-                                             Result.getNumColumns()),
+                     Result.addNumComputeOps(getNumOps(Result.getVectorTy()) *
+                                             Result.getNumVectors()),
                      Builder);
     return true;
   }
 
   /// Helper to linearize a matrix expression tree into a string. Currently
   /// matrix expressions are linarized by starting at an expression leaf and
   /// linearizing bottom up.
   struct ExprLinearizer {
     unsigned LengthToBreak = 100;
     std::string Str;
     raw_string_ostream Stream;
     unsigned LineLength = 0;
     const DataLayout &DL;
 
-    /// Mapping from instructions to column matrixes. It is used to identify
+    /// Mapping from instructions to matrixes. It is used to identify
     /// matrix instructions.
-    const MapVector<Value *, MatrixTy> &Inst2ColumnMatrix;
+    const MapVector<Value *, MatrixTy> &Inst2Matrix;
 
     /// Mapping from values to the leaves of all expressions that the value is
     /// part of.
     const DenseMap<Value *, SmallPtrSet<Value *, 2>> &Shared;
 
     /// Set of matrix expressions in the scope of a given DISubprogram.
     const SmallSetVector<Value *, 32> &ExprsInSubprogram;
 
     /// Leaf node of the expression to linearize.
     Value *Leaf;
 
     /// Used to keep track of sub-expressions that get reused while linearizing
     /// the expression. Re-used sub-expressions are marked as (reused).
     SmallPtrSet<Value *, 8> ReusedExprs;
 
     ExprLinearizer(const DataLayout &DL,
-                   const MapVector<Value *, MatrixTy> &Inst2ColumnMatrix,
+                   const MapVector<Value *, MatrixTy> &Inst2Matrix,
                    const DenseMap<Value *, SmallPtrSet<Value *, 2>> &Shared,
                    const SmallSetVector<Value *, 32> &ExprsInSubprogram,
                    Value *Leaf)
-        : Str(), Stream(Str), DL(DL), Inst2ColumnMatrix(Inst2ColumnMatrix),
-          Shared(Shared), ExprsInSubprogram(ExprsInSubprogram), Leaf(Leaf) {}
+        : Str(), Stream(Str), DL(DL), Inst2Matrix(Inst2Matrix), Shared(Shared),
+          ExprsInSubprogram(ExprsInSubprogram), Leaf(Leaf) {}
 
     void indent(unsigned N) {
       LineLength += N;
       for (unsigned i = 0; i < N; i++)
         Stream << " ";
     }
 
     void lineBreak() {
struct ExprLinearizer {
     }
 
     /// Returns true if \p V is a matrix value in the given subprogram.
     bool isMatrix(Value *V) const { return ExprsInSubprogram.count(V); }
 
     /// If \p V is a matrix value, print its shape as as NumRows x NumColumns to
     /// \p SS.
     void prettyPrintMatrixType(Value *V, raw_string_ostream &SS) {
-      auto M = Inst2ColumnMatrix.find(V);
-      if (M == Inst2ColumnMatrix.end())
+      auto M = Inst2Matrix.find(V);
+      if (M == Inst2Matrix.end())
         SS << "unknown";
       else {
         SS << M->second.getNumRows();
         SS << "x";
         SS << M->second.getNumColumns();
       }
     }
 

   //     mapping.  Leaves are lowered matrix instructions without other matrix
   //     users (like stores) in the current subprogram.
   /// 3. For each leaf, create a remark containing a linearizied version of the
   ///    matrix expression. The expression is linearized by a recursive
   ///    bottom-up traversal of the matrix operands, starting at a leaf. Note
   ///    that multiple leaves can share sub-expressions. Shared subexpressions
   ///    are explicitly marked as shared().
   struct RemarkGenerator {
-    const MapVector<Value *, MatrixTy> &Inst2ColumnMatrix;
+    const MapVector<Value *, MatrixTy> &Inst2Matrix;
     OptimizationRemarkEmitter &ORE;
     Function &Func;
     const DataLayout &DL;
 
-    RemarkGenerator(const MapVector<Value *, MatrixTy> &Inst2ColumnMatrix,
+    RemarkGenerator(const MapVector<Value *, MatrixTy> &Inst2Matrix,
                     OptimizationRemarkEmitter &ORE, Function &Func)
-        : Inst2ColumnMatrix(Inst2ColumnMatrix), ORE(ORE), Func(Func),
+        : Inst2Matrix(Inst2Matrix), ORE(ORE), Func(Func),
           DL(Func.getParent()->getDataLayout()) {}
 
     /// Return all leaves of the expressions in \p ExprsInSubprogram. Those are
-    /// instructions in Inst2ColumnMatrix returning void or without any users in
+    /// instructions in Inst2Matrix returning void or without any users in
     /// \p ExprsInSubprogram. Currently that should only include stores.
     SmallVector<Value *, 4>
     getExpressionLeaves(const SmallSetVector<Value *, 32> &ExprsInSubprogram) {
       SmallVector<Value *, 4> Leaves;
       for (auto *Expr : ExprsInSubprogram)
         if (Expr->getType()->isVoidTy() ||
             !any_of(Expr->users(), [&ExprsInSubprogram](User *U) {
               return ExprsInSubprogram.count(U);
sumOpInfos(Value *Root, SmallPtrSetImpl<Value *> &ReusedExprs,
       // Already counted this expression. Stop.
       if (!ReusedExprs.insert(Root).second)
         return {};
 
       OpInfoTy SharedCount;
       OpInfoTy Count;
 
       auto I = Shared.find(Root);
-      auto CM = Inst2ColumnMatrix.find(Root);
+      auto CM = Inst2Matrix.find(Root);
       if (I->second.size() == 1)
         Count = CM->second.getOpInfo();
       else
         SharedCount = CM->second.getOpInfo();
 
       for (Value *Op : cast<Instruction>(Root)->operand_values()) {
         auto C = sumOpInfos(Op, ReusedExprs, ExprsInSubprogram, Shared);
         Count += C.first;
         SharedCount += C.second;
       }
       return {Count, SharedCount};
     }
 
     void emitRemarks() {
       if (!ORE.allowExtraAnalysis(DEBUG_TYPE))
         return;
 
       // Map matrix operations to their containting subprograms, by traversing
       // the inlinedAt chain. If the function does not have a DISubprogram, we
       // only map them to the containing function.
       MapVector<DISubprogram *, SmallVector<Value *, 8>> Subprog2Exprs;
-      for (auto &KV : Inst2ColumnMatrix) {
+      for (auto &KV : Inst2Matrix) {
         if (Func.getSubprogram()) {
           auto *I = cast<Instruction>(KV.first);
           DILocation *Context = I->getDebugLoc();
           while (Context) {
             auto I =
                 Subprog2Exprs.insert({getSubprogram(Context->getScope()), {}});
             I.first->second.push_back(KV.first);
             Context = DebugLoc(Context).getInlinedAt();
void emitRemarks() {
       }
     }
 
     std::string
     linearize(Value *L,
               const DenseMap<Value *, SmallPtrSet<Value *, 2>> &Shared,
               const SmallSetVector<Value *, 32> &ExprsInSubprogram,
               const DataLayout &DL) {
-      ExprLinearizer Lin(DL, Inst2ColumnMatrix, Shared, ExprsInSubprogram, L);
+      ExprLinearizer Lin(DL, Inst2Matrix, Shared, ExprsInSubprogram, L);
       Lin.linearizeExpr(L, 0, false, false);
       return Lin.getResult();
     }
   };
 };
 } // namespace
 
 PreservedAnalyses LowerMatrixIntrinsicsPass::run(Function &F,

llvm/test/Transforms/LowerMatrixIntrinsics/multiply-add-sub-double-row-major.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --verbose
+
+; RUN: opt -lower-matrix-intrinsics -matrix-default-layout=row-major -S < %s | FileCheck --check-prefix=RM %s
+
+; Check row-major code generation for loads, stores, binary operators (fadd/fsub) and multiply.
+; %a.ptr is a pointer to a 2x3 matrix, %b.ptr to a 3x2 matrix and %c.ptr to a 2x2 matrix.
+; Load, store and binary operators on %a should operate on 3 element vectors and on 2 element vectors for %b.
+define void @multiply_sub_add_2x3_3x2(<6 x double>* %a.ptr, <6 x double>* %b.ptr, <4 x double>* %c.ptr) {
+; RM-LABEL: @multiply_sub_add_2x3_3x2(
+; RM-NEXT:  entry:
+; RM-NEXT:    [[TMP0:%.*]] = bitcast <6 x double>* [[A_PTR:%.*]] to double*
+; RM-NEXT:    [[VEC_CAST:%.*]] = bitcast double* [[TMP0]] to <3 x double>*
+; RM-NEXT:    [[COL_LOAD:%.*]] = load <3 x double>, <3 x double>* [[VEC_CAST]], align 8
+; RM-NEXT:    [[VEC_GEP:%.*]] = getelementptr double, double* [[TMP0]], i32 3
+; RM-NEXT:    [[VEC_CAST1:%.*]] = bitcast double* [[VEC_GEP]] to <3 x double>*
+; RM-NEXT:    [[COL_LOAD2:%.*]] = load <3 x double>, <3 x double>* [[VEC_CAST1]], align 8
+; RM-NEXT:    [[TMP1:%.*]] = bitcast <6 x double>* [[B_PTR:%.*]] to double*
+; RM-NEXT:    [[VEC_CAST3:%.*]] = bitcast double* [[TMP1]] to <2 x double>*
+; RM-NEXT:    [[COL_LOAD4:%.*]] = load <2 x double>, <2 x double>* [[VEC_CAST3]], align 8
+; RM-NEXT:    [[VEC_GEP5:%.*]] = getelementptr double, double* [[TMP1]], i32 2
+; RM-NEXT:    [[VEC_CAST6:%.*]] = bitcast double* [[VEC_GEP5]] to <2 x double>*
+; RM-NEXT:    [[COL_LOAD7:%.*]] = load <2 x double>, <2 x double>* [[VEC_CAST6]], align 8
+; RM-NEXT:    [[VEC_GEP8:%.*]] = getelementptr double, double* [[TMP1]], i32 4
+; RM-NEXT:    [[VEC_CAST9:%.*]] = bitcast double* [[VEC_GEP8]] to <2 x double>*
+; RM-NEXT:    [[COL_LOAD10:%.*]] = load <2 x double>, <2 x double>* [[VEC_CAST9]], align 8
+; RM-NEXT:    [[TMP2:%.*]] = fadd <3 x double> [[COL_LOAD]], [[COL_LOAD]]
+; RM-NEXT:    [[TMP3:%.*]] = fadd <3 x double> [[COL_LOAD2]], [[COL_LOAD2]]
+; RM-NEXT:    [[TMP4:%.*]] = bitcast <6 x double>* [[A_PTR]] to double*
+; RM-NEXT:    [[VEC_CAST11:%.*]] = bitcast double* [[TMP4]] to <3 x double>*
+; RM-NEXT:    store <3 x double> [[TMP2]], <3 x double>* [[VEC_CAST11]], align 8
+; RM-NEXT:    [[VEC_GEP12:%.*]] = getelementptr double, double* [[TMP4]], i32 3
+; RM-NEXT:    [[VEC_CAST13:%.*]] = bitcast double* [[VEC_GEP12]] to <3 x double>*
+; RM-NEXT:    store <3 x double> [[TMP3]], <3 x double>* [[VEC_CAST13]], align 8
+; RM-NEXT:    [[TMP5:%.*]] = fsub <2 x double> [[COL_LOAD4]], <double 1.000000e+00, double 1.000000e+00>
+; RM-NEXT:    [[TMP6:%.*]] = fsub <2 x double> [[COL_LOAD7]], <double 1.000000e+00, double 1.000000e+00>
+; RM-NEXT:    [[TMP7:%.*]] = fsub <2 x double> [[COL_LOAD10]], <double 1.000000e+00, double 1.000000e+00>
+; RM-NEXT:    [[TMP8:%.*]] = bitcast <6 x double>* [[B_PTR]] to double*
+; RM-NEXT:    [[VEC_CAST14:%.*]] = bitcast double* [[TMP8]] to <2 x double>*
+; RM-NEXT:    store <2 x double> [[TMP5]], <2 x double>* [[VEC_CAST14]], align 8
+; RM-NEXT:    [[VEC_GEP15:%.*]] = getelementptr double, double* [[TMP8]], i32 2
+; RM-NEXT:    [[VEC_CAST16:%.*]] = bitcast double* [[VEC_GEP15]] to <2 x double>*
+; RM-NEXT:    store <2 x double> [[TMP6]], <2 x double>* [[VEC_CAST16]], align 8
+; RM-NEXT:    [[VEC_GEP17:%.*]] = getelementptr double, double* [[TMP8]], i32 4
+; RM-NEXT:    [[VEC_CAST18:%.*]] = bitcast double* [[VEC_GEP17]] to <2 x double>*
+; RM-NEXT:    store <2 x double> [[TMP7]], <2 x double>* [[VEC_CAST18]], align 8
+; RM-NEXT:    [[BLOCK:%.*]] = shufflevector <2 x double> [[TMP5]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP9:%.*]] = extractelement <3 x double> [[TMP2]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x double> undef, double [[TMP9]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP10:%.*]] = fmul <1 x double> [[SPLAT_SPLAT]], [[BLOCK]]
+; RM-NEXT:    [[BLOCK19:%.*]] = shufflevector <2 x double> [[TMP6]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP11:%.*]] = extractelement <3 x double> [[TMP2]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <1 x double> undef, double [[TMP11]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT20]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP12:%.*]] = fmul <1 x double> [[SPLAT_SPLAT21]], [[BLOCK19]]
+; RM-NEXT:    [[TMP13:%.*]] = fadd <1 x double> [[TMP10]], [[TMP12]]
+; RM-NEXT:    [[BLOCK22:%.*]] = shufflevector <2 x double> [[TMP7]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP14:%.*]] = extractelement <3 x double> [[TMP2]], i64 2
+; RM-NEXT:    [[SPLAT_SPLATINSERT23:%.*]] = insertelement <1 x double> undef, double [[TMP14]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT24:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT23]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP15:%.*]] = fmul <1 x double> [[SPLAT_SPLAT24]], [[BLOCK22]]
+; RM-NEXT:    [[TMP16:%.*]] = fadd <1 x double> [[TMP13]], [[TMP15]]
+; RM-NEXT:    [[TMP17:%.*]] = shufflevector <1 x double> [[TMP16]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP18:%.*]] = shufflevector <2 x double> undef, <2 x double> [[TMP17]], <2 x i32> <i32 2, i32 1>
+; RM-NEXT:    [[BLOCK25:%.*]] = shufflevector <2 x double> [[TMP5]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP19:%.*]] = extractelement <3 x double> [[TMP2]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT26:%.*]] = insertelement <1 x double> undef, double [[TMP19]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT27:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT26]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP20:%.*]] = fmul <1 x double> [[SPLAT_SPLAT27]], [[BLOCK25]]
+; RM-NEXT:    [[BLOCK28:%.*]] = shufflevector <2 x double> [[TMP6]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP21:%.*]] = extractelement <3 x double> [[TMP2]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT29:%.*]] = insertelement <1 x double> undef, double [[TMP21]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT30:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT29]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP22:%.*]] = fmul <1 x double> [[SPLAT_SPLAT30]], [[BLOCK28]]
+; RM-NEXT:    [[TMP23:%.*]] = fadd <1 x double> [[TMP20]], [[TMP22]]
+; RM-NEXT:    [[BLOCK31:%.*]] = shufflevector <2 x double> [[TMP7]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP24:%.*]] = extractelement <3 x double> [[TMP2]], i64 2
+; RM-NEXT:    [[SPLAT_SPLATINSERT32:%.*]] = insertelement <1 x double> undef, double [[TMP24]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT33:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT32]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP25:%.*]] = fmul <1 x double> [[SPLAT_SPLAT33]], [[BLOCK31]]
+; RM-NEXT:    [[TMP26:%.*]] = fadd <1 x double> [[TMP23]], [[TMP25]]
+; RM-NEXT:    [[TMP27:%.*]] = shufflevector <1 x double> [[TMP26]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP28:%.*]] = shufflevector <2 x double> [[TMP18]], <2 x double> [[TMP27]], <2 x i32> <i32 0, i32 2>
+; RM-NEXT:    [[BLOCK34:%.*]] = shufflevector <2 x double> [[TMP5]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP29:%.*]] = extractelement <3 x double> [[TMP3]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT35:%.*]] = insertelement <1 x double> undef, double [[TMP29]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT36:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT35]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP30:%.*]] = fmul <1 x double> [[SPLAT_SPLAT36]], [[BLOCK34]]
+; RM-NEXT:    [[BLOCK37:%.*]] = shufflevector <2 x double> [[TMP6]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP31:%.*]] = extractelement <3 x double> [[TMP3]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT38:%.*]] = insertelement <1 x double> undef, double [[TMP31]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT39:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT38]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP32:%.*]] = fmul <1 x double> [[SPLAT_SPLAT39]], [[BLOCK37]]
+; RM-NEXT:    [[TMP33:%.*]] = fadd <1 x double> [[TMP30]], [[TMP32]]
+; RM-NEXT:    [[BLOCK40:%.*]] = shufflevector <2 x double> [[TMP7]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP34:%.*]] = extractelement <3 x double> [[TMP3]], i64 2
+; RM-NEXT:    [[SPLAT_SPLATINSERT41:%.*]] = insertelement <1 x double> undef, double [[TMP34]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT42:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT41]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP35:%.*]] = fmul <1 x double> [[SPLAT_SPLAT42]], [[BLOCK40]]
+; RM-NEXT:    [[TMP36:%.*]] = fadd <1 x double> [[TMP33]], [[TMP35]]
+; RM-NEXT:    [[TMP37:%.*]] = shufflevector <1 x double> [[TMP36]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP38:%.*]] = shufflevector <2 x double> undef, <2 x double> [[TMP37]], <2 x i32> <i32 2, i32 1>
+; RM-NEXT:    [[BLOCK43:%.*]] = shufflevector <2 x double> [[TMP5]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP39:%.*]] = extractelement <3 x double> [[TMP3]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT44:%.*]] = insertelement <1 x double> undef, double [[TMP39]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT45:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT44]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP40:%.*]] = fmul <1 x double> [[SPLAT_SPLAT45]], [[BLOCK43]]
+; RM-NEXT:    [[BLOCK46:%.*]] = shufflevector <2 x double> [[TMP6]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP41:%.*]] = extractelement <3 x double> [[TMP3]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT47:%.*]] = insertelement <1 x double> undef, double [[TMP41]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT48:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT47]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP42:%.*]] = fmul <1 x double> [[SPLAT_SPLAT48]], [[BLOCK46]]
+; RM-NEXT:    [[TMP43:%.*]] = fadd <1 x double> [[TMP40]], [[TMP42]]
+; RM-NEXT:    [[BLOCK49:%.*]] = shufflevector <2 x double> [[TMP7]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP44:%.*]] = extractelement <3 x double> [[TMP3]], i64 2
+; RM-NEXT:    [[SPLAT_SPLATINSERT50:%.*]] = insertelement <1 x double> undef, double [[TMP44]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT51:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT50]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP45:%.*]] = fmul <1 x double> [[SPLAT_SPLAT51]], [[BLOCK49]]
+; RM-NEXT:    [[TMP46:%.*]] = fadd <1 x double> [[TMP43]], [[TMP45]]
+; RM-NEXT:    [[TMP47:%.*]] = shufflevector <1 x double> [[TMP46]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP48:%.*]] = shufflevector <2 x double> [[TMP38]], <2 x double> [[TMP47]], <2 x i32> <i32 0, i32 2>
+; RM-NEXT:    [[TMP49:%.*]] = bitcast <4 x double>* [[C_PTR:%.*]] to double*
+; RM-NEXT:    [[VEC_CAST52:%.*]] = bitcast double* [[TMP49]] to <2 x double>*
+; RM-NEXT:    [[COL_LOAD53:%.*]] = load <2 x double>, <2 x double>* [[VEC_CAST52]], align 8
+; RM-NEXT:    [[VEC_GEP54:%.*]] = getelementptr double, double* [[TMP49]], i32 2
+; RM-NEXT:    [[VEC_CAST55:%.*]] = bitcast double* [[VEC_GEP54]] to <2 x double>*
+; RM-NEXT:    [[COL_LOAD56:%.*]] = load <2 x double>, <2 x double>* [[VEC_CAST55]], align 8
+; RM-NEXT:    [[TMP50:%.*]] = fsub <2 x double> [[COL_LOAD53]], [[TMP28]]
+; RM-NEXT:    [[TMP51:%.*]] = fsub <2 x double> [[COL_LOAD56]], [[TMP48]]
+; RM-NEXT:    [[TMP52:%.*]] = bitcast <4 x double>* [[C_PTR]] to double*
+; RM-NEXT:    [[VEC_CAST57:%.*]] = bitcast double* [[TMP52]] to <2 x double>*
+; RM-NEXT:    store <2 x double> [[TMP50]], <2 x double>* [[VEC_CAST57]], align 8
+; RM-NEXT:    [[VEC_GEP58:%.*]] = getelementptr double, double* [[TMP52]], i32 2
+; RM-NEXT:    [[VEC_CAST59:%.*]] = bitcast double* [[VEC_GEP58]] to <2 x double>*
+; RM-NEXT:    store <2 x double> [[TMP51]], <2 x double>* [[VEC_CAST59]], align 8
+; RM-NEXT:    ret void
+;
+entry:
+  %a = load <6 x double>, <6 x double>* %a.ptr
+  %b = load <6 x double>, <6 x double>* %b.ptr
+  %add = fadd <6 x double> %a, %a
+  store <6 x double> %add, <6 x double>* %a.ptr
+  %sub = fsub <6 x double> %b, <double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0>
+  store <6 x double> %sub, <6 x double>* %b.ptr
+  %mul = call <4 x double> @llvm.matrix.multiply.v4f64.v6f64.v6f64(<6 x double> %add, <6 x double> %sub, i32 2, i32 3, i32 2)
+  %c = load <4 x double>, <4 x double>* %c.ptr
+  %res = fsub <4 x double> %c, %mul
+  store <4 x double> %res, <4 x double>* %c.ptr
+  ret void
+}
+
+declare <4 x double> @llvm.matrix.multiply.v4f64.v6f64.v6f64(<6 x double>, <6 x double>, i32, i32, i32)

llvm/test/Transforms/LowerMatrixIntrinsics/multiply-double-row-major.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --verbose
+
+; RUN: opt -lower-matrix-intrinsics -matrix-default-layout=row-major -S < %s | FileCheck --check-prefix=RM %s
+
+
+
+define <4 x double> @multiply_2x2(<4 x double> %a, <4 x double> %b) {
+; RM-LABEL: @multiply_2x2(
+; RM-NEXT:  entry:
+; RM-NEXT:    [[SPLIT:%.*]] = shufflevector <4 x double> [[A:%.*]], <4 x double> undef, <2 x i32> <i32 0, i32 1>
+; RM-NEXT:    [[SPLIT1:%.*]] = shufflevector <4 x double> [[A]], <4 x double> undef, <2 x i32> <i32 2, i32 3>
+; RM-NEXT:    [[SPLIT2:%.*]] = shufflevector <4 x double> [[B:%.*]], <4 x double> undef, <2 x i32> <i32 0, i32 1>
+; RM-NEXT:    [[SPLIT3:%.*]] = shufflevector <4 x double> [[B]], <4 x double> undef, <2 x i32> <i32 2, i32 3>
+; RM-NEXT:    [[BLOCK:%.*]] = shufflevector <2 x double> [[SPLIT2]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP0:%.*]] = extractelement <2 x double> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x double> undef, double [[TMP0]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP1:%.*]] = fmul <1 x double> [[SPLAT_SPLAT]], [[BLOCK]]
+; RM-NEXT:    [[BLOCK4:%.*]] = shufflevector <2 x double> [[SPLIT3]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP2:%.*]] = extractelement <2 x double> [[SPLIT]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT5:%.*]] = insertelement <1 x double> undef, double [[TMP2]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT6:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT5]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP3:%.*]] = fmul <1 x double> [[SPLAT_SPLAT6]], [[BLOCK4]]
+; RM-NEXT:    [[TMP4:%.*]] = fadd <1 x double> [[TMP1]], [[TMP3]]
+; RM-NEXT:    [[TMP5:%.*]] = shufflevector <1 x double> [[TMP4]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP6:%.*]] = shufflevector <2 x double> undef, <2 x double> [[TMP5]], <2 x i32> <i32 2, i32 1>
+; RM-NEXT:    [[BLOCK7:%.*]] = shufflevector <2 x double> [[SPLIT2]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP7:%.*]] = extractelement <2 x double> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT8:%.*]] = insertelement <1 x double> undef, double [[TMP7]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT9:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT8]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP8:%.*]] = fmul <1 x double> [[SPLAT_SPLAT9]], [[BLOCK7]]
+; RM-NEXT:    [[BLOCK10:%.*]] = shufflevector <2 x double> [[SPLIT3]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP9:%.*]] = extractelement <2 x double> [[SPLIT]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT11:%.*]] = insertelement <1 x double> undef, double [[TMP9]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT12:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT11]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP10:%.*]] = fmul <1 x double> [[SPLAT_SPLAT12]], [[BLOCK10]]
+; RM-NEXT:    [[TMP11:%.*]] = fadd <1 x double> [[TMP8]], [[TMP10]]
+; RM-NEXT:    [[TMP12:%.*]] = shufflevector <1 x double> [[TMP11]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP13:%.*]] = shufflevector <2 x double> [[TMP6]], <2 x double> [[TMP12]], <2 x i32> <i32 0, i32 2>
+; RM-NEXT:    [[BLOCK13:%.*]] = shufflevector <2 x double> [[SPLIT2]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP14:%.*]] = extractelement <2 x double> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT14:%.*]] = insertelement <1 x double> undef, double [[TMP14]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT15:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT14]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP15:%.*]] = fmul <1 x double> [[SPLAT_SPLAT15]], [[BLOCK13]]
+; RM-NEXT:    [[BLOCK16:%.*]] = shufflevector <2 x double> [[SPLIT3]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP16:%.*]] = extractelement <2 x double> [[SPLIT1]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT17:%.*]] = insertelement <1 x double> undef, double [[TMP16]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT18:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT17]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP17:%.*]] = fmul <1 x double> [[SPLAT_SPLAT18]], [[BLOCK16]]
+; RM-NEXT:    [[TMP18:%.*]] = fadd <1 x double> [[TMP15]], [[TMP17]]
+; RM-NEXT:    [[TMP19:%.*]] = shufflevector <1 x double> [[TMP18]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP20:%.*]] = shufflevector <2 x double> undef, <2 x double> [[TMP19]], <2 x i32> <i32 2, i32 1>
+; RM-NEXT:    [[BLOCK19:%.*]] = shufflevector <2 x double> [[SPLIT2]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP21:%.*]] = extractelement <2 x double> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <1 x double> undef, double [[TMP21]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT20]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP22:%.*]] = fmul <1 x double> [[SPLAT_SPLAT21]], [[BLOCK19]]
+; RM-NEXT:    [[BLOCK22:%.*]] = shufflevector <2 x double> [[SPLIT3]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP23:%.*]] = extractelement <2 x double> [[SPLIT1]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT23:%.*]] = insertelement <1 x double> undef, double [[TMP23]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT24:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT23]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP24:%.*]] = fmul <1 x double> [[SPLAT_SPLAT24]], [[BLOCK22]]
+; RM-NEXT:    [[TMP25:%.*]] = fadd <1 x double> [[TMP22]], [[TMP24]]
+; RM-NEXT:    [[TMP26:%.*]] = shufflevector <1 x double> [[TMP25]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP27:%.*]] = shufflevector <2 x double> [[TMP20]], <2 x double> [[TMP26]], <2 x i32> <i32 0, i32 2>
+; RM-NEXT:    [[TMP28:%.*]] = shufflevector <2 x double> [[TMP13]], <2 x double> [[TMP27]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; RM-NEXT:    ret <4 x double> [[TMP28]]
+;
+entry:
+  %c = call <4 x double> @llvm.matrix.multiply.v4f64.v4f64.v4f64(<4 x double> %a, <4 x double> %b, i32 2, i32 2, i32 2)
+  ret <4 x double> %c
+}
+
+declare <4 x double> @llvm.matrix.multiply.v4f64.v4f64.v4f64(<4 x double>, <4 x double>, i32, i32, i32)
+
+define <4 x double> @multiply_1x2(<2 x double> %a, <2 x double> %b) {
+
+; RM-LABEL: @multiply_1x2(
+; RM-NEXT:  entry:
+; RM-NEXT:    [[SPLIT:%.*]] = shufflevector <2 x double> [[A:%.*]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[SPLIT1:%.*]] = shufflevector <2 x double> [[A]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[SPLIT2:%.*]] = shufflevector <2 x double> [[B:%.*]], <2 x double> undef, <2 x i32> <i32 0, i32 1>
+; RM-NEXT:    [[BLOCK:%.*]] = shufflevector <2 x double> [[SPLIT2]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP0:%.*]] = extractelement <1 x double> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x double> undef, double [[TMP0]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP1:%.*]] = fmul <1 x double> [[SPLAT_SPLAT]], [[BLOCK]]
+; RM-NEXT:    [[TMP2:%.*]] = shufflevector <1 x double> [[TMP1]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP3:%.*]] = shufflevector <2 x double> undef, <2 x double> [[TMP2]], <2 x i32> <i32 2, i32 1>
+; RM-NEXT:    [[BLOCK3:%.*]] = shufflevector <2 x double> [[SPLIT2]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP4:%.*]] = extractelement <1 x double> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT4:%.*]] = insertelement <1 x double> undef, double [[TMP4]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT5:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT4]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP5:%.*]] = fmul <1 x double> [[SPLAT_SPLAT5]], [[BLOCK3]]
+; RM-NEXT:    [[TMP6:%.*]] = shufflevector <1 x double> [[TMP5]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP7:%.*]] = shufflevector <2 x double> [[TMP3]], <2 x double> [[TMP6]], <2 x i32> <i32 0, i32 2>
+; RM-NEXT:    [[BLOCK6:%.*]] = shufflevector <2 x double> [[SPLIT2]], <2 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP8:%.*]] = extractelement <1 x double> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT7:%.*]] = insertelement <1 x double> undef, double [[TMP8]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT8:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT7]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP9:%.*]] = fmul <1 x double> [[SPLAT_SPLAT8]], [[BLOCK6]]
+; RM-NEXT:    [[TMP10:%.*]] = shufflevector <1 x double> [[TMP9]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP11:%.*]] = shufflevector <2 x double> undef, <2 x double> [[TMP10]], <2 x i32> <i32 2, i32 1>
+; RM-NEXT:    [[BLOCK9:%.*]] = shufflevector <2 x double> [[SPLIT2]], <2 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP12:%.*]] = extractelement <1 x double> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT10:%.*]] = insertelement <1 x double> undef, double [[TMP12]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT11:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT10]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP13:%.*]] = fmul <1 x double> [[SPLAT_SPLAT11]], [[BLOCK9]]
+; RM-NEXT:    [[TMP14:%.*]] = shufflevector <1 x double> [[TMP13]], <1 x double> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP15:%.*]] = shufflevector <2 x double> [[TMP11]], <2 x double> [[TMP14]], <2 x i32> <i32 0, i32 2>
+; RM-NEXT:    [[TMP16:%.*]] = shufflevector <2 x double> [[TMP7]], <2 x double> [[TMP15]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; RM-NEXT:    ret <4 x double> [[TMP16]]
+;
+entry:
+  %c = call <4 x double> @llvm.matrix.multiply.v4f64.v2f64.v2f64(<2 x double> %a, <2 x double> %b, i32 2, i32 1, i32 2)
+  ret <4 x double> %c
+}
+
+declare <4 x double> @llvm.matrix.multiply.v4f64.v2f64.v2f64(<2 x double>, <2 x double>, i32, i32, i32)
+
+define <9 x double> @multiply_2x3(<6 x double> %a, <6 x double> %b) {
+; RM-LABEL: @multiply_2x3(
+; RM-NEXT:  entry:
+; RM-NEXT:    [[SPLIT:%.*]] = shufflevector <6 x double> [[A:%.*]], <6 x double> undef, <2 x i32> <i32 0, i32 1>
+; RM-NEXT:    [[SPLIT1:%.*]] = shufflevector <6 x double> [[A]], <6 x double> undef, <2 x i32> <i32 2, i32 3>
+; RM-NEXT:    [[SPLIT2:%.*]] = shufflevector <6 x double> [[A]], <6 x double> undef, <2 x i32> <i32 4, i32 5>
+; RM-NEXT:    [[SPLIT3:%.*]] = shufflevector <6 x double> [[B:%.*]], <6 x double> undef, <3 x i32> <i32 0, i32 1, i32 2>
+; RM-NEXT:    [[SPLIT4:%.*]] = shufflevector <6 x double> [[B]], <6 x double> undef, <3 x i32> <i32 3, i32 4, i32 5>
+; RM-NEXT:    [[BLOCK:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP0:%.*]] = extractelement <2 x double> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x double> undef, double [[TMP0]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP1:%.*]] = fmul <1 x double> [[SPLAT_SPLAT]], [[BLOCK]]
+; RM-NEXT:    [[BLOCK5:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP2:%.*]] = extractelement <2 x double> [[SPLIT]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT6:%.*]] = insertelement <1 x double> undef, double [[TMP2]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT6]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP3:%.*]] = fmul <1 x double> [[SPLAT_SPLAT7]], [[BLOCK5]]
+; RM-NEXT:    [[TMP4:%.*]] = fadd <1 x double> [[TMP1]], [[TMP3]]
+; RM-NEXT:    [[TMP5:%.*]] = shufflevector <1 x double> [[TMP4]], <1 x double> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP6:%.*]] = shufflevector <3 x double> undef, <3 x double> [[TMP5]], <3 x i32> <i32 3, i32 1, i32 2>
+; RM-NEXT:    [[BLOCK8:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP7:%.*]] = extractelement <2 x double> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT9:%.*]] = insertelement <1 x double> undef, double [[TMP7]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT10:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT9]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP8:%.*]] = fmul <1 x double> [[SPLAT_SPLAT10]], [[BLOCK8]]
+; RM-NEXT:    [[BLOCK11:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP9:%.*]] = extractelement <2 x double> [[SPLIT]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT12:%.*]] = insertelement <1 x double> undef, double [[TMP9]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT13:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT12]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP10:%.*]] = fmul <1 x double> [[SPLAT_SPLAT13]], [[BLOCK11]]
+; RM-NEXT:    [[TMP11:%.*]] = fadd <1 x double> [[TMP8]], [[TMP10]]
+; RM-NEXT:    [[TMP12:%.*]] = shufflevector <1 x double> [[TMP11]], <1 x double> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP13:%.*]] = shufflevector <3 x double> [[TMP6]], <3 x double> [[TMP12]], <3 x i32> <i32 0, i32 3, i32 2>
+; RM-NEXT:    [[BLOCK14:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP14:%.*]] = extractelement <2 x double> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT15:%.*]] = insertelement <1 x double> undef, double [[TMP14]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT16:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT15]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP15:%.*]] = fmul <1 x double> [[SPLAT_SPLAT16]], [[BLOCK14]]
+; RM-NEXT:    [[BLOCK17:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP16:%.*]] = extractelement <2 x double> [[SPLIT]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT18:%.*]] = insertelement <1 x double> undef, double [[TMP16]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT19:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT18]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP17:%.*]] = fmul <1 x double> [[SPLAT_SPLAT19]], [[BLOCK17]]
+; RM-NEXT:    [[TMP18:%.*]] = fadd <1 x double> [[TMP15]], [[TMP17]]
+; RM-NEXT:    [[TMP19:%.*]] = shufflevector <1 x double> [[TMP18]], <1 x double> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP20:%.*]] = shufflevector <3 x double> [[TMP13]], <3 x double> [[TMP19]], <3 x i32> <i32 0, i32 1, i32 3>
+; RM-NEXT:    [[BLOCK20:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP21:%.*]] = extractelement <2 x double> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT21:%.*]] = insertelement <1 x double> undef, double [[TMP21]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT22:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT21]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP22:%.*]] = fmul <1 x double> [[SPLAT_SPLAT22]], [[BLOCK20]]
+; RM-NEXT:    [[BLOCK23:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP23:%.*]] = extractelement <2 x double> [[SPLIT1]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT24:%.*]] = insertelement <1 x double> undef, double [[TMP23]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT25:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT24]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP24:%.*]] = fmul <1 x double> [[SPLAT_SPLAT25]], [[BLOCK23]]
+; RM-NEXT:    [[TMP25:%.*]] = fadd <1 x double> [[TMP22]], [[TMP24]]
+; RM-NEXT:    [[TMP26:%.*]] = shufflevector <1 x double> [[TMP25]], <1 x double> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP27:%.*]] = shufflevector <3 x double> undef, <3 x double> [[TMP26]], <3 x i32> <i32 3, i32 1, i32 2>
+; RM-NEXT:    [[BLOCK26:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP28:%.*]] = extractelement <2 x double> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT27:%.*]] = insertelement <1 x double> undef, double [[TMP28]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT28:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT27]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP29:%.*]] = fmul <1 x double> [[SPLAT_SPLAT28]], [[BLOCK26]]
+; RM-NEXT:    [[BLOCK29:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP30:%.*]] = extractelement <2 x double> [[SPLIT1]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT30:%.*]] = insertelement <1 x double> undef, double [[TMP30]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT31:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT30]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP31:%.*]] = fmul <1 x double> [[SPLAT_SPLAT31]], [[BLOCK29]]
+; RM-NEXT:    [[TMP32:%.*]] = fadd <1 x double> [[TMP29]], [[TMP31]]
+; RM-NEXT:    [[TMP33:%.*]] = shufflevector <1 x double> [[TMP32]], <1 x double> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP34:%.*]] = shufflevector <3 x double> [[TMP27]], <3 x double> [[TMP33]], <3 x i32> <i32 0, i32 3, i32 2>
+; RM-NEXT:    [[BLOCK32:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP35:%.*]] = extractelement <2 x double> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT33:%.*]] = insertelement <1 x double> undef, double [[TMP35]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT34:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT33]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP36:%.*]] = fmul <1 x double> [[SPLAT_SPLAT34]], [[BLOCK32]]
+; RM-NEXT:    [[BLOCK35:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP37:%.*]] = extractelement <2 x double> [[SPLIT1]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT36:%.*]] = insertelement <1 x double> undef, double [[TMP37]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT37:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT36]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP38:%.*]] = fmul <1 x double> [[SPLAT_SPLAT37]], [[BLOCK35]]
+; RM-NEXT:    [[TMP39:%.*]] = fadd <1 x double> [[TMP36]], [[TMP38]]
+; RM-NEXT:    [[TMP40:%.*]] = shufflevector <1 x double> [[TMP39]], <1 x double> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP41:%.*]] = shufflevector <3 x double> [[TMP34]], <3 x double> [[TMP40]], <3 x i32> <i32 0, i32 1, i32 3>
+; RM-NEXT:    [[BLOCK38:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP42:%.*]] = extractelement <2 x double> [[SPLIT2]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT39:%.*]] = insertelement <1 x double> undef, double [[TMP42]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT40:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT39]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP43:%.*]] = fmul <1 x double> [[SPLAT_SPLAT40]], [[BLOCK38]]
+; RM-NEXT:    [[BLOCK41:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP44:%.*]] = extractelement <2 x double> [[SPLIT2]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT42:%.*]] = insertelement <1 x double> undef, double [[TMP44]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT43:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT42]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP45:%.*]] = fmul <1 x double> [[SPLAT_SPLAT43]], [[BLOCK41]]
+; RM-NEXT:    [[TMP46:%.*]] = fadd <1 x double> [[TMP43]], [[TMP45]]
+; RM-NEXT:    [[TMP47:%.*]] = shufflevector <1 x double> [[TMP46]], <1 x double> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP48:%.*]] = shufflevector <3 x double> undef, <3 x double> [[TMP47]], <3 x i32> <i32 3, i32 1, i32 2>
+; RM-NEXT:    [[BLOCK44:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP49:%.*]] = extractelement <2 x double> [[SPLIT2]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT45:%.*]] = insertelement <1 x double> undef, double [[TMP49]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT46:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT45]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP50:%.*]] = fmul <1 x double> [[SPLAT_SPLAT46]], [[BLOCK44]]
+; RM-NEXT:    [[BLOCK47:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP51:%.*]] = extractelement <2 x double> [[SPLIT2]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT48:%.*]] = insertelement <1 x double> undef, double [[TMP51]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT49:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT48]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP52:%.*]] = fmul <1 x double> [[SPLAT_SPLAT49]], [[BLOCK47]]
+; RM-NEXT:    [[TMP53:%.*]] = fadd <1 x double> [[TMP50]], [[TMP52]]
+; RM-NEXT:    [[TMP54:%.*]] = shufflevector <1 x double> [[TMP53]], <1 x double> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP55:%.*]] = shufflevector <3 x double> [[TMP48]], <3 x double> [[TMP54]], <3 x i32> <i32 0, i32 3, i32 2>
+; RM-NEXT:    [[BLOCK50:%.*]] = shufflevector <3 x double> [[SPLIT3]], <3 x double> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP56:%.*]] = extractelement <2 x double> [[SPLIT2]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT51:%.*]] = insertelement <1 x double> undef, double [[TMP56]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT52:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT51]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP57:%.*]] = fmul <1 x double> [[SPLAT_SPLAT52]], [[BLOCK50]]
+; RM-NEXT:    [[BLOCK53:%.*]] = shufflevector <3 x double> [[SPLIT4]], <3 x double> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP58:%.*]] = extractelement <2 x double> [[SPLIT2]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT54:%.*]] = insertelement <1 x double> undef, double [[TMP58]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT55:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT54]], <1 x double> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP59:%.*]] = fmul <1 x double> [[SPLAT_SPLAT55]], [[BLOCK53]]
+; RM-NEXT:    [[TMP60:%.*]] = fadd <1 x double> [[TMP57]], [[TMP59]]
+; RM-NEXT:    [[TMP61:%.*]] = shufflevector <1 x double> [[TMP60]], <1 x double> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP62:%.*]] = shufflevector <3 x double> [[TMP55]], <3 x double> [[TMP61]], <3 x i32> <i32 0, i32 1, i32 3>
+; RM-NEXT:    [[TMP63:%.*]] = shufflevector <3 x double> [[TMP20]], <3 x double> [[TMP41]], <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5>
+; RM-NEXT:    [[TMP64:%.*]] = shufflevector <3 x double> [[TMP62]], <3 x double> undef, <6 x i32> <i32 0, i32 1, i32 2, i32 undef, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP65:%.*]] = shufflevector <6 x double> [[TMP63]], <6 x double> [[TMP64]], <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
+; RM-NEXT:    ret <9 x double> [[TMP65]]
+;
+entry:
+  %c = call <9 x double> @llvm.matrix.multiply.v6f64.v6f64.v6f64(<6 x double> %a, <6 x double> %b, i32 3, i32 2, i32 3)
+  ret <9 x double> %c
+}
+
+declare <9 x double> @llvm.matrix.multiply.v6f64.v6f64.v6f64(<6 x double>, <6 x double>, i32, i32, i32)

llvm/test/Transforms/LowerMatrixIntrinsics/multiply-fused-multiple-blocks.ll

 ; CHECK-NEXT:    store <2 x double> [[TMP35]], <2 x double>* [[COL_CAST213]], align 8
 ; CHECK-NEXT:    br label [[END]]
 ; CHECK:       end:
 ; CHECK-NEXT:    [[ST_B92:%.*]] = ptrtoint <9 x double>* [[C]] to i64
 ; CHECK-NEXT:    [[ST_E93:%.*]] = add nuw nsw i64 [[ST_B92]], 72
 ; CHECK-NEXT:    [[LD_B97:%.*]] = ptrtoint <6 x double>* [[A]] to i64
 ; CHECK-NEXT:    [[TMP36:%.*]] = icmp ugt i64 [[ST_E93]], [[LD_B97]]
 ; CHECK-NEXT:    br i1 [[TMP36]], label [[ALIAS_CONT94:%.*]], label [[NO_ALIAS96:%.*]]
-; CHECK:       alias_cont92:
+; CHECK:       alias_cont91:
 ; CHECK-NEXT:    [[LD_E98:%.*]] = add nuw nsw i64 [[LD_B97]], 48
 ; CHECK-NEXT:    [[TMP37:%.*]] = icmp ugt i64 [[LD_E98]], [[ST_B92]]
 ; CHECK-NEXT:    br i1 [[TMP37]], label [[COPY95:%.*]], label [[NO_ALIAS96]]
-; CHECK:       copy93:
+; CHECK:       copy92:
 ; CHECK-NEXT:    [[TMP38:%.*]] = alloca <6 x double>, align 64
 ; CHECK-NEXT:    [[TMP39:%.*]] = bitcast <6 x double>* [[TMP38]] to i8*
 ; CHECK-NEXT:    [[TMP40:%.*]] = bitcast <6 x double>* [[A]] to i8*
 ; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 64 dereferenceable(48) [[TMP39]], i8* nonnull align 16 dereferenceable(48) [[TMP40]], i64 48, i1 false)
 ; CHECK-NEXT:    br label [[NO_ALIAS96]]
-; CHECK:       no_alias94:
+; CHECK:       no_alias93:
 ; CHECK-NEXT:    [[TMP41:%.*]] = phi <6 x double>* [ [[A]], [[END]] ], [ [[A]], [[ALIAS_CONT94]] ], [ [[TMP38]], [[COPY95]] ]
 ; CHECK-NEXT:    [[ST_B99:%.*]] = ptrtoint <9 x double>* [[C]] to i64
 ; CHECK-NEXT:    [[ST_E100:%.*]] = add nuw nsw i64 [[ST_B99]], 72
 ; CHECK-NEXT:    [[LD_B104:%.*]] = ptrtoint <6 x double>* [[B]] to i64
 ; CHECK-NEXT:    [[TMP42:%.*]] = icmp ugt i64 [[ST_E100]], [[LD_B104]]
 ; CHECK-NEXT:    br i1 [[TMP42]], label [[ALIAS_CONT101:%.*]], label [[NO_ALIAS103:%.*]]
-; CHECK:       alias_cont99:
+; CHECK:       alias_cont98:
 ; CHECK-NEXT:    [[LD_E105:%.*]] = add nuw nsw i64 [[LD_B104]], 48
 ; CHECK-NEXT:    [[TMP43:%.*]] = icmp ugt i64 [[LD_E105]], [[ST_B99]]
 ; CHECK-NEXT:    br i1 [[TMP43]], label [[COPY102:%.*]], label [[NO_ALIAS103]]
-; CHECK:       copy100:
+; CHECK:       copy99:
 ; CHECK-NEXT:    [[TMP44:%.*]] = alloca <6 x double>, align 64
 ; CHECK-NEXT:    [[TMP45:%.*]] = bitcast <6 x double>* [[TMP44]] to i8*
 ; CHECK-NEXT:    [[TMP46:%.*]] = bitcast <6 x double>* [[B]] to i8*
 ; CHECK-NEXT:    call void @llvm.memcpy.p0i8.p0i8.i64(i8* nonnull align 64 dereferenceable(48) [[TMP45]], i8* nonnull align 16 dereferenceable(48) [[TMP46]], i64 48, i1 false)
 ; CHECK-NEXT:    br label [[NO_ALIAS103]]
-; CHECK:       no_alias101:
+; CHECK:       no_alias100:
 ; CHECK-NEXT:    [[TMP47:%.*]] = phi <6 x double>* [ [[B]], [[NO_ALIAS96]] ], [ [[B]], [[ALIAS_CONT101]] ], [ [[TMP44]], [[COPY102]] ]
 ; CHECK-NEXT:    [[COL_CAST107:%.*]] = bitcast <6 x double>* [[TMP41]] to <2 x double>*
 ; CHECK-NEXT:    [[COL_LOAD108:%.*]] = load <2 x double>, <2 x double>* [[COL_CAST107]], align 8
 ; CHECK-NEXT:    [[COL_GEP109:%.*]] = getelementptr <6 x double>, <6 x double>* [[TMP41]], i64 0, i64 3
 ; CHECK-NEXT:    [[COL_CAST110:%.*]] = bitcast double* [[COL_GEP109]] to <2 x double>*
 ; CHECK-NEXT:    [[COL_LOAD111:%.*]] = load <2 x double>, <2 x double>* [[COL_CAST110]], align 8
 ; CHECK-NEXT:    [[COL_CAST113:%.*]] = bitcast <6 x double>* [[TMP47]] to <2 x double>*
 ; CHECK-NEXT:    [[COL_LOAD114:%.*]] = load <2 x double>, <2 x double>* [[COL_CAST113]], align 8

llvm/test/Transforms/LowerMatrixIntrinsics/multiply-i32-row-major.ll

+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --verbose
+
+; RUN: opt -lower-matrix-intrinsics -matrix-default-layout=row-major -S < %s | FileCheck --check-prefix=RM %s
+
+
+
+define <4 x i32> @multiply_2x2(<4 x i32> %a, <4 x i32> %b) {
+; RM-LABEL: @multiply_2x2(
+; RM-NEXT:  entry:
+; RM-NEXT:    [[SPLIT:%.*]] = shufflevector <4 x i32> [[A:%.*]], <4 x i32> undef, <2 x i32> <i32 0, i32 1>
+; RM-NEXT:    [[SPLIT1:%.*]] = shufflevector <4 x i32> [[A]], <4 x i32> undef, <2 x i32> <i32 2, i32 3>
+; RM-NEXT:    [[SPLIT2:%.*]] = shufflevector <4 x i32> [[B:%.*]], <4 x i32> undef, <2 x i32> <i32 0, i32 1>
+; RM-NEXT:    [[SPLIT3:%.*]] = shufflevector <4 x i32> [[B]], <4 x i32> undef, <2 x i32> <i32 2, i32 3>
+; RM-NEXT:    [[BLOCK:%.*]] = shufflevector <2 x i32> [[SPLIT2]], <2 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP0:%.*]] = extractelement <2 x i32> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x i32> undef, i32 [[TMP0]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP1:%.*]] = mul <1 x i32> [[SPLAT_SPLAT]], [[BLOCK]]
+; RM-NEXT:    [[BLOCK4:%.*]] = shufflevector <2 x i32> [[SPLIT3]], <2 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP2:%.*]] = extractelement <2 x i32> [[SPLIT]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT5:%.*]] = insertelement <1 x i32> undef, i32 [[TMP2]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT6:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT5]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP3:%.*]] = mul <1 x i32> [[SPLAT_SPLAT6]], [[BLOCK4]]
+; RM-NEXT:    [[TMP4:%.*]] = add <1 x i32> [[TMP1]], [[TMP3]]
+; RM-NEXT:    [[TMP5:%.*]] = shufflevector <1 x i32> [[TMP4]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP6:%.*]] = shufflevector <2 x i32> undef, <2 x i32> [[TMP5]], <2 x i32> <i32 2, i32 1>
+; RM-NEXT:    [[BLOCK7:%.*]] = shufflevector <2 x i32> [[SPLIT2]], <2 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP7:%.*]] = extractelement <2 x i32> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT8:%.*]] = insertelement <1 x i32> undef, i32 [[TMP7]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT9:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT8]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP8:%.*]] = mul <1 x i32> [[SPLAT_SPLAT9]], [[BLOCK7]]
+; RM-NEXT:    [[BLOCK10:%.*]] = shufflevector <2 x i32> [[SPLIT3]], <2 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP9:%.*]] = extractelement <2 x i32> [[SPLIT]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT11:%.*]] = insertelement <1 x i32> undef, i32 [[TMP9]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT12:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT11]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP10:%.*]] = mul <1 x i32> [[SPLAT_SPLAT12]], [[BLOCK10]]
+; RM-NEXT:    [[TMP11:%.*]] = add <1 x i32> [[TMP8]], [[TMP10]]
+; RM-NEXT:    [[TMP12:%.*]] = shufflevector <1 x i32> [[TMP11]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP13:%.*]] = shufflevector <2 x i32> [[TMP6]], <2 x i32> [[TMP12]], <2 x i32> <i32 0, i32 2>
+; RM-NEXT:    [[BLOCK13:%.*]] = shufflevector <2 x i32> [[SPLIT2]], <2 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP14:%.*]] = extractelement <2 x i32> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT14:%.*]] = insertelement <1 x i32> undef, i32 [[TMP14]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT15:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT14]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP15:%.*]] = mul <1 x i32> [[SPLAT_SPLAT15]], [[BLOCK13]]
+; RM-NEXT:    [[BLOCK16:%.*]] = shufflevector <2 x i32> [[SPLIT3]], <2 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP16:%.*]] = extractelement <2 x i32> [[SPLIT1]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT17:%.*]] = insertelement <1 x i32> undef, i32 [[TMP16]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT18:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT17]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP17:%.*]] = mul <1 x i32> [[SPLAT_SPLAT18]], [[BLOCK16]]
+; RM-NEXT:    [[TMP18:%.*]] = add <1 x i32> [[TMP15]], [[TMP17]]
+; RM-NEXT:    [[TMP19:%.*]] = shufflevector <1 x i32> [[TMP18]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP20:%.*]] = shufflevector <2 x i32> undef, <2 x i32> [[TMP19]], <2 x i32> <i32 2, i32 1>
+; RM-NEXT:    [[BLOCK19:%.*]] = shufflevector <2 x i32> [[SPLIT2]], <2 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP21:%.*]] = extractelement <2 x i32> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT20:%.*]] = insertelement <1 x i32> undef, i32 [[TMP21]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT21:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT20]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP22:%.*]] = mul <1 x i32> [[SPLAT_SPLAT21]], [[BLOCK19]]
+; RM-NEXT:    [[BLOCK22:%.*]] = shufflevector <2 x i32> [[SPLIT3]], <2 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP23:%.*]] = extractelement <2 x i32> [[SPLIT1]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT23:%.*]] = insertelement <1 x i32> undef, i32 [[TMP23]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT24:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT23]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP24:%.*]] = mul <1 x i32> [[SPLAT_SPLAT24]], [[BLOCK22]]
+; RM-NEXT:    [[TMP25:%.*]] = add <1 x i32> [[TMP22]], [[TMP24]]
+; RM-NEXT:    [[TMP26:%.*]] = shufflevector <1 x i32> [[TMP25]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP27:%.*]] = shufflevector <2 x i32> [[TMP20]], <2 x i32> [[TMP26]], <2 x i32> <i32 0, i32 2>
+; RM-NEXT:    [[TMP28:%.*]] = shufflevector <2 x i32> [[TMP13]], <2 x i32> [[TMP27]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; RM-NEXT:    ret <4 x i32> [[TMP28]]
+;
+entry:
+  %c = call <4 x i32> @llvm.matrix.multiply.v4f64.v4f64.v4f64(<4 x i32> %a, <4 x i32> %b, i32 2, i32 2, i32 2)
+  ret <4 x i32> %c
+}
+
+declare <4 x i32> @llvm.matrix.multiply.v4f64.v4f64.v4f64(<4 x i32>, <4 x i32>, i32, i32, i32)
+
+define <4 x i32> @multiply_1x2(<2 x i32> %a, <2 x i32> %b) {
+
+; RM-LABEL: @multiply_1x2(
+; RM-NEXT:  entry:
+; RM-NEXT:    [[SPLIT:%.*]] = shufflevector <2 x i32> [[A:%.*]], <2 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[SPLIT1:%.*]] = shufflevector <2 x i32> [[A]], <2 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[SPLIT2:%.*]] = shufflevector <2 x i32> [[B:%.*]], <2 x i32> undef, <2 x i32> <i32 0, i32 1>
+; RM-NEXT:    [[BLOCK:%.*]] = shufflevector <2 x i32> [[SPLIT2]], <2 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP0:%.*]] = extractelement <1 x i32> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x i32> undef, i32 [[TMP0]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP1:%.*]] = mul <1 x i32> [[SPLAT_SPLAT]], [[BLOCK]]
+; RM-NEXT:    [[TMP2:%.*]] = shufflevector <1 x i32> [[TMP1]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP3:%.*]] = shufflevector <2 x i32> undef, <2 x i32> [[TMP2]], <2 x i32> <i32 2, i32 1>
+; RM-NEXT:    [[BLOCK3:%.*]] = shufflevector <2 x i32> [[SPLIT2]], <2 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP4:%.*]] = extractelement <1 x i32> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT4:%.*]] = insertelement <1 x i32> undef, i32 [[TMP4]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT5:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT4]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP5:%.*]] = mul <1 x i32> [[SPLAT_SPLAT5]], [[BLOCK3]]
+; RM-NEXT:    [[TMP6:%.*]] = shufflevector <1 x i32> [[TMP5]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP7:%.*]] = shufflevector <2 x i32> [[TMP3]], <2 x i32> [[TMP6]], <2 x i32> <i32 0, i32 2>
+; RM-NEXT:    [[BLOCK6:%.*]] = shufflevector <2 x i32> [[SPLIT2]], <2 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP8:%.*]] = extractelement <1 x i32> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT7:%.*]] = insertelement <1 x i32> undef, i32 [[TMP8]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT8:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT7]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP9:%.*]] = mul <1 x i32> [[SPLAT_SPLAT8]], [[BLOCK6]]
+; RM-NEXT:    [[TMP10:%.*]] = shufflevector <1 x i32> [[TMP9]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP11:%.*]] = shufflevector <2 x i32> undef, <2 x i32> [[TMP10]], <2 x i32> <i32 2, i32 1>
+; RM-NEXT:    [[BLOCK9:%.*]] = shufflevector <2 x i32> [[SPLIT2]], <2 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP12:%.*]] = extractelement <1 x i32> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT10:%.*]] = insertelement <1 x i32> undef, i32 [[TMP12]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT11:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT10]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP13:%.*]] = mul <1 x i32> [[SPLAT_SPLAT11]], [[BLOCK9]]
+; RM-NEXT:    [[TMP14:%.*]] = shufflevector <1 x i32> [[TMP13]], <1 x i32> undef, <2 x i32> <i32 0, i32 undef>
+; RM-NEXT:    [[TMP15:%.*]] = shufflevector <2 x i32> [[TMP11]], <2 x i32> [[TMP14]], <2 x i32> <i32 0, i32 2>
+; RM-NEXT:    [[TMP16:%.*]] = shufflevector <2 x i32> [[TMP7]], <2 x i32> [[TMP15]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+; RM-NEXT:    ret <4 x i32> [[TMP16]]
+;
+entry:
+  %c = call <4 x i32> @llvm.matrix.multiply.v4f64.v2f64.v2f64(<2 x i32> %a, <2 x i32> %b, i32 2, i32 1, i32 2)
+  ret <4 x i32> %c
+}
+
+declare <4 x i32> @llvm.matrix.multiply.v4f64.v2f64.v2f64(<2 x i32>, <2 x i32>, i32, i32, i32)
+
+define <9 x i32> @multiply_2x3(<6 x i32> %a, <6 x i32> %b) {
+; RM-LABEL: @multiply_2x3(
+; RM-NEXT:  entry:
+; RM-NEXT:    [[SPLIT:%.*]] = shufflevector <6 x i32> [[A:%.*]], <6 x i32> undef, <2 x i32> <i32 0, i32 1>
+; RM-NEXT:    [[SPLIT1:%.*]] = shufflevector <6 x i32> [[A]], <6 x i32> undef, <2 x i32> <i32 2, i32 3>
+; RM-NEXT:    [[SPLIT2:%.*]] = shufflevector <6 x i32> [[A]], <6 x i32> undef, <2 x i32> <i32 4, i32 5>
+; RM-NEXT:    [[SPLIT3:%.*]] = shufflevector <6 x i32> [[B:%.*]], <6 x i32> undef, <3 x i32> <i32 0, i32 1, i32 2>
+; RM-NEXT:    [[SPLIT4:%.*]] = shufflevector <6 x i32> [[B]], <6 x i32> undef, <3 x i32> <i32 3, i32 4, i32 5>
+; RM-NEXT:    [[BLOCK:%.*]] = shufflevector <3 x i32> [[SPLIT3]], <3 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP0:%.*]] = extractelement <2 x i32> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x i32> undef, i32 [[TMP0]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP1:%.*]] = mul <1 x i32> [[SPLAT_SPLAT]], [[BLOCK]]
+; RM-NEXT:    [[BLOCK5:%.*]] = shufflevector <3 x i32> [[SPLIT4]], <3 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP2:%.*]] = extractelement <2 x i32> [[SPLIT]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT6:%.*]] = insertelement <1 x i32> undef, i32 [[TMP2]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT7:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT6]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP3:%.*]] = mul <1 x i32> [[SPLAT_SPLAT7]], [[BLOCK5]]
+; RM-NEXT:    [[TMP4:%.*]] = add <1 x i32> [[TMP1]], [[TMP3]]
+; RM-NEXT:    [[TMP5:%.*]] = shufflevector <1 x i32> [[TMP4]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP6:%.*]] = shufflevector <3 x i32> undef, <3 x i32> [[TMP5]], <3 x i32> <i32 3, i32 1, i32 2>
+; RM-NEXT:    [[BLOCK8:%.*]] = shufflevector <3 x i32> [[SPLIT3]], <3 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP7:%.*]] = extractelement <2 x i32> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT9:%.*]] = insertelement <1 x i32> undef, i32 [[TMP7]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT10:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT9]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP8:%.*]] = mul <1 x i32> [[SPLAT_SPLAT10]], [[BLOCK8]]
+; RM-NEXT:    [[BLOCK11:%.*]] = shufflevector <3 x i32> [[SPLIT4]], <3 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP9:%.*]] = extractelement <2 x i32> [[SPLIT]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT12:%.*]] = insertelement <1 x i32> undef, i32 [[TMP9]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT13:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT12]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP10:%.*]] = mul <1 x i32> [[SPLAT_SPLAT13]], [[BLOCK11]]
+; RM-NEXT:    [[TMP11:%.*]] = add <1 x i32> [[TMP8]], [[TMP10]]
+; RM-NEXT:    [[TMP12:%.*]] = shufflevector <1 x i32> [[TMP11]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP13:%.*]] = shufflevector <3 x i32> [[TMP6]], <3 x i32> [[TMP12]], <3 x i32> <i32 0, i32 3, i32 2>
+; RM-NEXT:    [[BLOCK14:%.*]] = shufflevector <3 x i32> [[SPLIT3]], <3 x i32> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP14:%.*]] = extractelement <2 x i32> [[SPLIT]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT15:%.*]] = insertelement <1 x i32> undef, i32 [[TMP14]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT16:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT15]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP15:%.*]] = mul <1 x i32> [[SPLAT_SPLAT16]], [[BLOCK14]]
+; RM-NEXT:    [[BLOCK17:%.*]] = shufflevector <3 x i32> [[SPLIT4]], <3 x i32> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP16:%.*]] = extractelement <2 x i32> [[SPLIT]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT18:%.*]] = insertelement <1 x i32> undef, i32 [[TMP16]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT19:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT18]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP17:%.*]] = mul <1 x i32> [[SPLAT_SPLAT19]], [[BLOCK17]]
+; RM-NEXT:    [[TMP18:%.*]] = add <1 x i32> [[TMP15]], [[TMP17]]
+; RM-NEXT:    [[TMP19:%.*]] = shufflevector <1 x i32> [[TMP18]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP20:%.*]] = shufflevector <3 x i32> [[TMP13]], <3 x i32> [[TMP19]], <3 x i32> <i32 0, i32 1, i32 3>
+; RM-NEXT:    [[BLOCK20:%.*]] = shufflevector <3 x i32> [[SPLIT3]], <3 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP21:%.*]] = extractelement <2 x i32> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT21:%.*]] = insertelement <1 x i32> undef, i32 [[TMP21]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT22:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT21]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP22:%.*]] = mul <1 x i32> [[SPLAT_SPLAT22]], [[BLOCK20]]
+; RM-NEXT:    [[BLOCK23:%.*]] = shufflevector <3 x i32> [[SPLIT4]], <3 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP23:%.*]] = extractelement <2 x i32> [[SPLIT1]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT24:%.*]] = insertelement <1 x i32> undef, i32 [[TMP23]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT25:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT24]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP24:%.*]] = mul <1 x i32> [[SPLAT_SPLAT25]], [[BLOCK23]]
+; RM-NEXT:    [[TMP25:%.*]] = add <1 x i32> [[TMP22]], [[TMP24]]
+; RM-NEXT:    [[TMP26:%.*]] = shufflevector <1 x i32> [[TMP25]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP27:%.*]] = shufflevector <3 x i32> undef, <3 x i32> [[TMP26]], <3 x i32> <i32 3, i32 1, i32 2>
+; RM-NEXT:    [[BLOCK26:%.*]] = shufflevector <3 x i32> [[SPLIT3]], <3 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP28:%.*]] = extractelement <2 x i32> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT27:%.*]] = insertelement <1 x i32> undef, i32 [[TMP28]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT28:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT27]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP29:%.*]] = mul <1 x i32> [[SPLAT_SPLAT28]], [[BLOCK26]]
+; RM-NEXT:    [[BLOCK29:%.*]] = shufflevector <3 x i32> [[SPLIT4]], <3 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP30:%.*]] = extractelement <2 x i32> [[SPLIT1]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT30:%.*]] = insertelement <1 x i32> undef, i32 [[TMP30]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT31:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT30]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP31:%.*]] = mul <1 x i32> [[SPLAT_SPLAT31]], [[BLOCK29]]
+; RM-NEXT:    [[TMP32:%.*]] = add <1 x i32> [[TMP29]], [[TMP31]]
+; RM-NEXT:    [[TMP33:%.*]] = shufflevector <1 x i32> [[TMP32]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP34:%.*]] = shufflevector <3 x i32> [[TMP27]], <3 x i32> [[TMP33]], <3 x i32> <i32 0, i32 3, i32 2>
+; RM-NEXT:    [[BLOCK32:%.*]] = shufflevector <3 x i32> [[SPLIT3]], <3 x i32> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP35:%.*]] = extractelement <2 x i32> [[SPLIT1]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT33:%.*]] = insertelement <1 x i32> undef, i32 [[TMP35]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT34:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT33]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP36:%.*]] = mul <1 x i32> [[SPLAT_SPLAT34]], [[BLOCK32]]
+; RM-NEXT:    [[BLOCK35:%.*]] = shufflevector <3 x i32> [[SPLIT4]], <3 x i32> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP37:%.*]] = extractelement <2 x i32> [[SPLIT1]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT36:%.*]] = insertelement <1 x i32> undef, i32 [[TMP37]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT37:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT36]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP38:%.*]] = mul <1 x i32> [[SPLAT_SPLAT37]], [[BLOCK35]]
+; RM-NEXT:    [[TMP39:%.*]] = add <1 x i32> [[TMP36]], [[TMP38]]
+; RM-NEXT:    [[TMP40:%.*]] = shufflevector <1 x i32> [[TMP39]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP41:%.*]] = shufflevector <3 x i32> [[TMP34]], <3 x i32> [[TMP40]], <3 x i32> <i32 0, i32 1, i32 3>
+; RM-NEXT:    [[BLOCK38:%.*]] = shufflevector <3 x i32> [[SPLIT3]], <3 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP42:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT39:%.*]] = insertelement <1 x i32> undef, i32 [[TMP42]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT40:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT39]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP43:%.*]] = mul <1 x i32> [[SPLAT_SPLAT40]], [[BLOCK38]]
+; RM-NEXT:    [[BLOCK41:%.*]] = shufflevector <3 x i32> [[SPLIT4]], <3 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP44:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT42:%.*]] = insertelement <1 x i32> undef, i32 [[TMP44]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT43:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT42]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP45:%.*]] = mul <1 x i32> [[SPLAT_SPLAT43]], [[BLOCK41]]
+; RM-NEXT:    [[TMP46:%.*]] = add <1 x i32> [[TMP43]], [[TMP45]]
+; RM-NEXT:    [[TMP47:%.*]] = shufflevector <1 x i32> [[TMP46]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP48:%.*]] = shufflevector <3 x i32> undef, <3 x i32> [[TMP47]], <3 x i32> <i32 3, i32 1, i32 2>
+; RM-NEXT:    [[BLOCK44:%.*]] = shufflevector <3 x i32> [[SPLIT3]], <3 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP49:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT45:%.*]] = insertelement <1 x i32> undef, i32 [[TMP49]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT46:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT45]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP50:%.*]] = mul <1 x i32> [[SPLAT_SPLAT46]], [[BLOCK44]]
+; RM-NEXT:    [[BLOCK47:%.*]] = shufflevector <3 x i32> [[SPLIT4]], <3 x i32> undef, <1 x i32> <i32 1>
+; RM-NEXT:    [[TMP51:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT48:%.*]] = insertelement <1 x i32> undef, i32 [[TMP51]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT49:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT48]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP52:%.*]] = mul <1 x i32> [[SPLAT_SPLAT49]], [[BLOCK47]]
+; RM-NEXT:    [[TMP53:%.*]] = add <1 x i32> [[TMP50]], [[TMP52]]
+; RM-NEXT:    [[TMP54:%.*]] = shufflevector <1 x i32> [[TMP53]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP55:%.*]] = shufflevector <3 x i32> [[TMP48]], <3 x i32> [[TMP54]], <3 x i32> <i32 0, i32 3, i32 2>
+; RM-NEXT:    [[BLOCK50:%.*]] = shufflevector <3 x i32> [[SPLIT3]], <3 x i32> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP56:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 0
+; RM-NEXT:    [[SPLAT_SPLATINSERT51:%.*]] = insertelement <1 x i32> undef, i32 [[TMP56]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT52:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT51]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP57:%.*]] = mul <1 x i32> [[SPLAT_SPLAT52]], [[BLOCK50]]
+; RM-NEXT:    [[BLOCK53:%.*]] = shufflevector <3 x i32> [[SPLIT4]], <3 x i32> undef, <1 x i32> <i32 2>
+; RM-NEXT:    [[TMP58:%.*]] = extractelement <2 x i32> [[SPLIT2]], i64 1
+; RM-NEXT:    [[SPLAT_SPLATINSERT54:%.*]] = insertelement <1 x i32> undef, i32 [[TMP58]], i32 0
+; RM-NEXT:    [[SPLAT_SPLAT55:%.*]] = shufflevector <1 x i32> [[SPLAT_SPLATINSERT54]], <1 x i32> undef, <1 x i32> zeroinitializer
+; RM-NEXT:    [[TMP59:%.*]] = mul <1 x i32> [[SPLAT_SPLAT55]], [[BLOCK53]]
+; RM-NEXT:    [[TMP60:%.*]] = add <1 x i32> [[TMP57]], [[TMP59]]
+; RM-NEXT:    [[TMP61:%.*]] = shufflevector <1 x i32> [[TMP60]], <1 x i32> undef, <3 x i32> <i32 0, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP62:%.*]] = shufflevector <3 x i32> [[TMP55]], <3 x i32> [[TMP61]], <3 x i32> <i32 0, i32 1, i32 3>
+; RM-NEXT:    [[TMP63:%.*]] = shufflevector <3 x i32> [[TMP20]], <3 x i32> [[TMP41]], <6 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5>
+; RM-NEXT:    [[TMP64:%.*]] = shufflevector <3 x i32> [[TMP62]], <3 x i32> undef, <6 x i32> <i32 0, i32 1, i32 2, i32 undef, i32 undef, i32 undef>
+; RM-NEXT:    [[TMP65:%.*]] = shufflevector <6 x i32> [[TMP63]], <6 x i32> [[TMP64]], <9 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
+; RM-NEXT:    ret <9 x i32> [[TMP65]]
+;
+entry:
+  %c = call <9 x i32> @llvm.matrix.multiply.v6f64.v6f64.v6f64(<6 x i32> %a, <6 x i32> %b, i32 3, i32 2, i32 3)
+  ret <9 x i32> %c
+}
+
+declare <9 x i32> @llvm.matrix.multiply.v6f64.v6f64.v6f64(<6 x i32>, <6 x i32>, i32, i32, i32)