[mlir] make vector to llvm conversion truly partial

[ftynse]

Summary

Historically, the Vector to LLVM dialect conversion subsumed the Standard to
LLVM dialect conversion patterns. This was necessary because the conversion
infrastructure did not have sufficient support for reconciling type
conversions. This support is now available. Only keep the patterns related to
the Vector dialect in the Vector to LLVM conversion and require type casts
operations to be inserted if necessary. These casts will be removed by
following conversions if possible. Update integration tests to also run the
Standard to LLVM conversion.

There is a significant amount of test churn, which is due to (a) unnecessarily
strict tests in VectorToLLVM and (b) many patterns actually targeting Standard
dialect ops instead of LLVM dialect ops leading to tests actually exercising a
Vector->Standard->LLVM conversion. This churn is a good illustration of the
reason to make the conversion partial: now the tests only check the code in the
Vector to LLVM conversion and will not be randomly broken by changes in
Standard to LLVM conversion.

Arguably, it may be possible to extract Vector to Standard patterns into a
separate pass, but given the ongoing splitting of the Standard dialect, such
pass will be short-lived and will require further refactoring.

Depends On D95626

[aartbik]

Oh this is nice! I always felt like vector dialect lowered too much, making the tests more LLVM-centric than they should be, but this resolves it nicely.
Please wait for Nicolas to approve too, but LGTM

mlir/lib/Conversion/VectorToLLVM/ConvertVectorToLLVMPass.cpp

73–74

don't you want to remove this for real now instead of leaving it commented out ;-)

88

I have the weird request to remove all empty lines in this new code, just because all the architectural specific additions were in one straight block, but I can see how that request is very subjective, so feel free to ignore as well....

[nicolasvasilache]

Great thanks, double benefit of killing most LLVM types :)

Thanks!

[ftynse]

Address review.

mlir/integration_test/Dialect/Linalg/CPU/benchmark_matmul.mlir

 // RUN: export M=24 && export K=64 && export N=192 && export ITERS=10 && \
 // RUN: cat %s | sed 's@${M}@'"$M"'@g'| sed 's@${K}@'"$K"'@g' | sed 's@${N}@'"$N"'@g'| sed 's@${ITERS}@'"$ITERS"'@g'| \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.matmul register-tile-sizes=12,32,16 vectorize" | \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.fill register-tile-sizes=4,32 vectorize" | \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.copy register-tile-sizes=4,32 vectorize" | \
 
 // RUN: mlir-opt -canonicalize -convert-vector-to-scf -lower-affine -convert-linalg-to-loops | \
-// RUN: mlir-opt -canonicalize -convert-scf-to-std -convert-vector-to-llvm | \
+// RUN: mlir-opt -canonicalize -convert-scf-to-std -convert-vector-to-llvm -convert-std-to-llvm | \
 // RUN: mlir-cpu-runner -O3 -e main -entry-point-result=void \
 // Activate to dump assembly
 // R_UN:   -dump-object-file -object-filename=/tmp/a.o \
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_c_runner_utils%shlibext | \
 // Use tee to both print to stderr and FileCheck
 // RUN: tee -a /dev/stderr | FileCheck %s
 
 

mlir/integration_test/Dialect/Linalg/CPU/benchmark_matmul_column_major.mlir

 // RUN: export M=24 && export K=64 && export N=192 && export ITERS=10 && \
 // RUN: cat %s | sed 's@${M}@'"$M"'@g'| sed 's@${K}@'"$K"'@g' | sed 's@${N}@'"$N"'@g'| sed 's@${ITERS}@'"$ITERS"'@g'| \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.matmul_column_major register-tile-sizes=16,0,32 vectorize" | \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.fill register-tile-sizes=4,16 vectorize" | \
 
 // TODO: linalg.copy vectorization in the presence of permutation map fails. Enable when addressed.
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.copy register-tile-sizes=4,16 vectorize" | \
 
 // RUN: mlir-opt -canonicalize -convert-vector-to-scf -lower-affine -convert-linalg-to-loops | \
-// RUN: mlir-opt -canonicalize -convert-scf-to-std -convert-vector-to-llvm | \
+// RUN: mlir-opt -canonicalize -convert-scf-to-std -convert-vector-to-llvm -convert-std-to-llvm | \
 // RUN: mlir-cpu-runner -O3 -e main -entry-point-result=void \
 // Activate to dump assembly
 // R_UN:   -dump-object-file -object-filename=/tmp/a.o \
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_c_runner_utils%shlibext | \
 // Use tee to both print to stderr and FileCheck
 // RUN: tee -a /dev/stderr | FileCheck %s
 
 !elem_type_a = type f32

mlir/integration_test/Dialect/Linalg/CPU/benchmark_matmul_column_major_as_row_major.mlir

 // RUN: export M=24 && export K=64 && export N=192 && export ITERS=10 && \
 // RUN: cat %s | sed 's@${M}@'"$M"'@g'| sed 's@${K}@'"$K"'@g' | sed 's@${N}@'"$N"'@g'| sed 's@${ITERS}@'"$ITERS"'@g'| \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.matmul_column_major register-tile-sizes=16,0,32 vectorize" | \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.matmul register-tile-sizes=12,32,16 vectorize" | \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.fill register-tile-sizes=4,16 vectorize" | \
 
 // TODO: linalg.copy vectorization in the presence of permutation map fails. Enable when addressed.
 // R_UN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.copy register-tile-sizes=4,16 vectorize" | \
 
 // RUN: mlir-opt -canonicalize -convert-vector-to-scf -lower-affine -convert-linalg-to-loops | \
-// RUN: mlir-opt -canonicalize -convert-scf-to-std -convert-vector-to-llvm | \
+// RUN: mlir-opt -canonicalize -convert-scf-to-std -convert-vector-to-llvm -convert-std-to-llvm | \
 // RUN: mlir-cpu-runner -O3 -e main -entry-point-result=void \
 // Activate to dump assembly
 // R_UN:   -dump-object-file -object-filename=/tmp/a.o \
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_c_runner_utils%shlibext | \
 // Use tee to both print to stderr and FileCheck
 // RUN: tee -a /dev/stderr | FileCheck %s
 
 !elem_type_a = type f32

mlir/integration_test/Dialect/Linalg/CPU/benchmark_matmul_i8_i8_i32.mlir

 // RUN: export M=24 && export K=64 && export N=192 && export ITERS=10 && \
 // RUN: cat %s | sed 's@${M}@'"$M"'@g'| sed 's@${K}@'"$K"'@g' | sed 's@${N}@'"$N"'@g'| sed 's@${ITERS}@'"$ITERS"'@g'| \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.matmul_i8_i8_i32 register-tile-sizes=12,32,16 vectorize" | \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.fill register-tile-sizes=4,32 vectorize" | \
 // RUN: mlir-opt -test-linalg-codegen-strategy="anchor-op=linalg.copy register-tile-sizes=4,32 vectorize" | \
 // RUN: mlir-opt -canonicalize -convert-vector-to-scf -lower-affine -convert-linalg-to-loops | \
 
-// RUN: mlir-opt -canonicalize -convert-scf-to-std -convert-vector-to-llvm -mlir-disable-threading | \
+// RUN: mlir-opt -canonicalize -convert-scf-to-std -convert-vector-to-llvm -convert-std-to-llvm -mlir-disable-threading | \
 // RUN: mlir-cpu-runner -O3 -e main -entry-point-result=void \
 // Activate to dump assembly
 // R_UN:   -dump-object-file -object-filename=/tmp/a.o \
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_c_runner_utils%shlibext | \
 // Use tee to both print to stderr and FileCheck
 // RUN: tee -a /dev/stderr | FileCheck %s
 
 

mlir/integration_test/Dialect/Standard/CPU/test-ceil-floor-pos-neg.mlir

-// RUN: mlir-opt %s -convert-vector-to-scf -lower-affine -convert-scf-to-std -std-expand -convert-vector-to-llvm | \
+// RUN: mlir-opt %s -convert-vector-to-scf -lower-affine -convert-scf-to-std -std-expand -convert-vector-to-llvm -convert-std-to-llvm | \
 // RUN: mlir-cpu-runner -e entry -entry-point-result=void  \
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_c_runner_utils%shlibext | \
 // RUN: FileCheck %s
 
 func @transfer_read_2d(%A : memref<40xi32>, %base1: index) {
   %i42 = constant -42: i32
   %f = vector.transfer_read %A[%base1], %i42
       {permutation_map = affine_map<(d0) -> (d0)>} :

mlir/integration_test/Dialect/Vector/CPU/test-transfer-read-2d.mlir

-// RUN: mlir-opt %s -convert-vector-to-scf -lower-affine -convert-scf-to-std -convert-vector-to-llvm | \
+// RUN: mlir-opt %s -convert-vector-to-scf -lower-affine -convert-scf-to-std -convert-vector-to-llvm -convert-std-to-llvm | \
 // RUN: mlir-cpu-runner -e entry -entry-point-result=void  \
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_c_runner_utils%shlibext | \
 // RUN: FileCheck %s
 
 func @transfer_read_2d(%A : memref<?x?xf32>, %base1: index, %base2: index) {
   %fm42 = constant -42.0: f32
   %f = vector.transfer_read %A[%base1, %base2], %fm42
       {permutation_map = affine_map<(d0, d1) -> (d0, d1)>} :

mlir/integration_test/Dialect/Vector/CPU/test-transfer-to-loops.mlir

-// RUN: mlir-opt %s -convert-vector-to-scf -lower-affine -convert-scf-to-std -convert-vector-to-llvm | \
+// RUN: mlir-opt %s -convert-vector-to-scf -lower-affine -convert-scf-to-std -convert-vector-to-llvm -convert-std-to-llvm | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void  \
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext,%mlir_integration_test_dir/libmlir_c_runner_utils%shlibext | \
 // RUN: FileCheck %s
 
 #map0 = affine_map<(d0, d1) -> (d1, d0)>
 #map1 = affine_map<(d0, d1) -> (d1)>
 
 func private @print_memref_f32(memref<*xf32>)

mlir/integration_test/Dialect/Vector/CPU/test-vector-distribute.mlir

 // RUN: mlir-opt %s -test-vector-to-forloop -convert-vector-to-scf \
-// RUN:   -lower-affine -convert-scf-to-std -convert-vector-to-llvm | \
+// RUN:   -lower-affine -convert-scf-to-std -convert-vector-to-llvm -convert-std-to-llvm | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void  \
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext | \
 // RUN: FileCheck %s
 
 // RUN: mlir-opt %s -convert-vector-to-scf -lower-affine \
-// RUN: -convert-scf-to-std -convert-vector-to-llvm | mlir-cpu-runner -e main \
+// RUN: -convert-scf-to-std -convert-vector-to-llvm -convert-std-to-llvm | mlir-cpu-runner -e main \
 // RUN: -entry-point-result=void \
 // RUN: -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext | \
 // RUN: FileCheck %s
 
 // RUN: mlir-opt %s -test-vector-to-forloop | FileCheck %s -check-prefix=TRANSFORM
 
 
 func private @print_memref_f32(memref<*xf32>)

mlir/lib/Conversion/ArmSVEToLLVM/ArmSVEToLLVM.cpp

convertScalableVectorTypeToLLVM(ScalableVectorType svType,
   if (!elementType)
     return {};
 
   auto sVectorType =
       LLVM::LLVMScalableVectorType::get(elementType, svType.getShape().back());
   return sVectorType;
 }
 
+template <typename OpTy>
+class ForwardOperands : public OpConversionPattern<OpTy> {
+  using OpConversionPattern<OpTy>::OpConversionPattern;
+
+  LogicalResult
+  matchAndRewrite(OpTy op, ArrayRef<Value> operands,
+                  ConversionPatternRewriter &rewriter) const final {
+    if (ValueRange(operands).getTypes() == op->getOperands().getTypes())
+      return rewriter.notifyMatchFailure(op, "operand types already match");
+
+    rewriter.updateRootInPlace(op, [&]() { op->setOperands(operands); });
+    return success();
+  }
+};
+
+class ReturnOpTypeConversion : public OpConversionPattern<ReturnOp> {
+public:
+  using OpConversionPattern<ReturnOp>::OpConversionPattern;
+
+  LogicalResult
+  matchAndRewrite(ReturnOp op, ArrayRef<Value> operands,
+                  ConversionPatternRewriter &rewriter) const final {
+    rewriter.updateRootInPlace(op, [&]() { op->setOperands(operands); });
+    return success();
+  }
+};
+
+static Optional<Value> addUnrealizedCast(OpBuilder &builder,
+                                         ScalableVectorType svType,
+                                         ValueRange inputs, Location loc) {
+  if (inputs.size() != 1 ||
+      !inputs[0].getType().isa<LLVM::LLVMScalableVectorType>())
+    return Value();
+  return builder.create<UnrealizedConversionCastOp>(loc, svType, inputs)
+      .getResult(0);
+}
+
 /// Populate the given list with patterns that convert from ArmSVE to LLVM.
 void mlir::populateArmSVEToLLVMConversionPatterns(
     LLVMTypeConverter &converter, OwningRewritePatternList &patterns) {
   converter.addConversion([&converter](ScalableVectorType svType) {
     return convertScalableVectorTypeToLLVM(svType, converter);
   });
+  converter.addSourceMaterialization(addUnrealizedCast);
+
   // clang-format off
+  patterns.insert<ForwardOperands<CallOp>,
+                  ForwardOperands<CallIndirectOp>,
+                  ForwardOperands<ReturnOp>>(converter,
+                                             &converter.getContext());
   patterns.insert<SdotOpLowering,
                   SmmlaOpLowering,
                   UdotOpLowering,
                   UmmlaOpLowering,
                   VectorScaleOpLowering>(converter);
   // clang-format on
 }

mlir/lib/Conversion/VectorToLLVM/ConvertVectorToLLVM.cpp

static SmallVector<int64_t, 4> getI64SubArray(ArrayAttr arrayAttr,
   SmallVector<int64_t, 4> res;
   res.reserve(arrayAttr.size() - dropFront - dropBack);
   for (auto it = range.begin() + dropFront, eit = range.end() - dropBack;
        it != eit; ++it)
     res.push_back((*it).getValue().getSExtValue());
   return res;
 }
 
+static Value createCastToIndexLike(ConversionPatternRewriter &rewriter,
+                                   Location loc, Type targetType, Value value) {
+  if (targetType == value.getType())
+    return value;
+
+  bool targetIsIndex = targetType.isIndex();
+  bool valueIsIndex = value.getType().isIndex();
+  if (targetIsIndex ^ valueIsIndex)
+    return rewriter.create<IndexCastOp>(loc, targetType, value);
+
+  auto targetIntegerType = targetType.dyn_cast<IntegerType>();
+  auto valueIntegerType = value.getType().dyn_cast<IntegerType>();
+  assert(targetIntegerType && valueIntegerType &&
+         "unexpected cast between types other than integers and index");
+  assert(targetIntegerType.getSignedness() == valueIntegerType.getSignedness());
+
+  if (targetIntegerType.getWidth() > valueIntegerType.getWidth())
+    return rewriter.create<SignExtendIOp>(loc, targetIntegerType, value);
+  return rewriter.create<TruncateIOp>(loc, targetIntegerType, value);
+}
+
 // Helper that returns a vector comparison that constructs a mask:
 //     mask = [0,1,..,n-1] + [o,o,..,o] < [b,b,..,b]
 //
 // NOTE: The LLVM::GetActiveLaneMaskOp intrinsic would provide an alternative,
 //       much more compact, IR for this operation, but LLVM eventually
 //       generates more elaborate instructions for this intrinsic since it
 //       is very conservative on the boundary conditions.
 static Value buildVectorComparison(ConversionPatternRewriter &rewriter,
static Value buildVectorComparison(ConversionPatternRewriter &rewriter,
   } else {
     indices = rewriter.create<ConstantOp>(
         loc, rewriter.getI64VectorAttr(
                  llvm::to_vector<4>(llvm::seq<int64_t>(0, dim))));
     idxType = rewriter.getI64Type();
   }
   // Add in an offset if requested.
   if (off) {
-    Value o = rewriter.create<IndexCastOp>(loc, idxType, *off);
+    Value o = createCastToIndexLike(rewriter, loc, idxType, *off);
     Value ov = rewriter.create<SplatOp>(loc, indices.getType(), o);
     indices = rewriter.create<AddIOp>(loc, ov, indices);
   }
   // Construct the vector comparison.
-  Value bound = rewriter.create<IndexCastOp>(loc, idxType, b);
+  Value bound = createCastToIndexLike(rewriter, loc, idxType, b);
   Value bounds = rewriter.create<SplatOp>(loc, indices.getType(), bound);
   return rewriter.create<CmpIOp>(loc, CmpIPredicate::slt, indices, bounds);
 }
 
 // Helper that returns data layout alignment of a memref.
 LogicalResult getMemRefAlignment(LLVMTypeConverter &typeConverter,
                                  MemRefType memrefType, unsigned &align) {
   Type elementTy = typeConverter.convertType(memrefType.getElementType());
replaceTransferOpWithLoadOrStore(ConversionPatternRewriter &rewriter,
   return success();
 }
 
 static LogicalResult
 replaceTransferOpWithMasked(ConversionPatternRewriter &rewriter,
                             LLVMTypeConverter &typeConverter, Location loc,
                             TransferReadOp xferOp, ArrayRef<Value> operands,
                             Value dataPtr, Value mask) {
-  auto toLLVMTy = [&](Type t) { return typeConverter.convertType(t); };
   VectorType fillType = xferOp.getVectorType();
   Value fill = rewriter.create<SplatOp>(loc, fillType, xferOp.padding());
-  fill = rewriter.create<LLVM::DialectCastOp>(loc, toLLVMTy(fillType), fill);
 
   Type vecTy = typeConverter.convertType(xferOp.getVectorType());
   if (!vecTy)
     return failure();
 
   unsigned align;
   if (failed(getMemRefAlignment(
           typeConverter, xferOp.getShapedType().cast<MemRefType>(), align)))

mlir/lib/Conversion/VectorToLLVM/ConvertVectorToLLVMPass.cpp

 #include "mlir/Conversion/StandardToLLVM/ConvertStandardToLLVMPass.h"
 #include "mlir/Dialect/AVX512/AVX512Dialect.h"
 #include "mlir/Dialect/ArmNeon/ArmNeonDialect.h"
 #include "mlir/Dialect/ArmSVE/ArmSVEDialect.h"
 #include "mlir/Dialect/LLVMIR/LLVMAVX512Dialect.h"
 #include "mlir/Dialect/LLVMIR/LLVMArmNeonDialect.h"
 #include "mlir/Dialect/LLVMIR/LLVMArmSVEDialect.h"
 #include "mlir/Dialect/LLVMIR/LLVMDialect.h"
+#include "mlir/Dialect/StandardOps/IR/Ops.h"
 #include "mlir/Dialect/Vector/VectorOps.h"
 #include "mlir/Transforms/GreedyPatternRewriteDriver.h"
 
 using namespace mlir;
 using namespace mlir::vector;
 
 namespace {
 struct LowerVectorToLLVMPass
void LowerVectorToLLVMPass::runOnOperation() {
   }
 
   // Convert to the LLVM IR dialect.
   LLVMTypeConverter converter(&getContext());
   OwningRewritePatternList patterns;
   populateVectorToLLVMMatrixConversionPatterns(converter, patterns);
   populateVectorToLLVMConversionPatterns(
       converter, patterns, reassociateFPReductions, enableIndexOptimizations);
   populateVectorToLLVMMatrixConversionPatterns(converter, patterns);
-  populateStdToLLVMConversionPatterns(converter, patterns);
 
   // Architecture specific augmentations.
   LLVMConversionTarget target(getContext());
+  target.addLegalOp<LLVM::DialectCastOp>();
+  target.addLegalDialect<StandardOpsDialect>();
+  target.addLegalOp<UnrealizedConversionCastOp>();
   if (enableArmNeon) {
     target.addLegalDialect<LLVM::LLVMArmNeonDialect>();
     target.addIllegalDialect<arm_neon::ArmNeonDialect>();
     populateArmNeonToLLVMConversionPatterns(converter, patterns);
   }
   if (enableArmSVE) {
     target.addLegalDialect<LLVM::LLVMArmSVEDialect>();
     target.addIllegalDialect<arm_sve::ArmSVEDialect>();
+    auto hasScalableVectorType = [](TypeRange types) {
+      for (Type type : types)
+        if (type.isa<arm_sve::ScalableVectorType>())
+          return true;
+      return false;
+    };
+    // Remove any ArmSVE-specific types from function signatures and results.
+    populateFuncOpTypeConversionPattern(patterns, &getContext(), converter);
+    target.addDynamicallyLegalOp<FuncOp>([hasScalableVectorType](FuncOp op) {
+      return !hasScalableVectorType(op.getType().getInputs()) &&
+             !hasScalableVectorType(op.getType().getResults());
+    });
+    target.addDynamicallyLegalOp<CallOp, CallIndirectOp, ReturnOp>(
+        [hasScalableVectorType](Operation *op) {
+          return !hasScalableVectorType(op->getOperandTypes()) &&
+                 !hasScalableVectorType(op->getResultTypes());
+        });
     populateArmSVEToLLVMConversionPatterns(converter, patterns);
   }
   if (enableAVX512) {
     target.addLegalDialect<LLVM::LLVMAVX512Dialect>();
     target.addIllegalDialect<avx512::AVX512Dialect>();
     populateAVX512ToLLVMConversionPatterns(converter, patterns);
   }
 

mlir/lib/Dialect/LLVMIR/IR/LLVMDialect.cpp

static LogicalResult verifyCast(DialectCastOp op, Type llvmType, Type type,
   // Index is compatible with any integer.
   if (type.isIndex()) {
     if (succeeded(verifyCastWithIndex(llvmType)))
       return success();
 
     return op.emitOpError("invalid cast between index and non-integer type");
   }
 
+  if (type.isa<IntegerType>()) {
+    auto llvmIntegerType = llvmType.dyn_cast<IntegerType>();
+    if (!llvmIntegerType)
+      return op->emitOpError("invalid cast between integer and non-integer");
+    if (llvmIntegerType.getWidth() != type.getIntOrFloatBitWidth())
+      return op.emitOpError("invalid cast changing integer width");
+    return success();
+  }
+
   // Vectors are compatible if they are 1D non-scalable, and their element types
-  // are compatible.
-  if (auto vectorType = type.dyn_cast<VectorType>())
+  // are compatible. nD vectors are compatible with (n-1)D arrays containing 1D
+  // vector.
+  if (auto vectorType = type.dyn_cast<VectorType>()) {
+    if (vectorType == llvmType && !isElement)
       return op.emitOpError("vector types should not be casted");
 
+    if (vectorType.getRank() == 1) {
+      auto llvmVectorType = llvmType.dyn_cast<VectorType>();
+      if (!llvmVectorType || llvmVectorType.getRank() != 1)
+        return op.emitOpError("invalid cast for vector types");
+
+      return verifyCast(op, llvmVectorType.getElementType(),
+                        vectorType.getElementType(), /*isElement=*/true);
+    }
+
+    auto arrayType = llvmType.dyn_cast<LLVM::LLVMArrayType>();
+    if (!arrayType ||
+        arrayType.getNumElements() != vectorType.getShape().front())
+      return op.emitOpError("invalid cast for vector, expected array");
+    return verifyCast(op, arrayType.getElementType(),
+                      VectorType::get(vectorType.getShape().drop_front(),
+                                      vectorType.getElementType()),
+                      /*isElement=*/true);
+  }
+
   if (auto memrefType = type.dyn_cast<MemRefType>()) {
     // Bare pointer convention: statically-shaped memref is compatible with an
     // LLVM pointer to the element type.
     if (auto ptrType = llvmType.dyn_cast<LLVMPointerType>()) {
       if (!memrefType.hasStaticShape())
         return op->emitOpError(
             "unexpected bare pointer for dynamically shaped memref");
       if (memrefType.getMemorySpace() != ptrType.getAddressSpace())

mlir/test/Conversion/ArmSVEToLLVM/convert-to-llvm.mlir

-// RUN: mlir-opt %s -convert-vector-to-llvm="enable-arm-sve" | mlir-opt | FileCheck %s
+// RUN: mlir-opt %s -convert-vector-to-llvm="enable-arm-sve" -convert-std-to-llvm | mlir-opt | FileCheck %s
 
 func @arm_sve_sdot(%a: !arm_sve.vector<16xi8>,
                    %b: !arm_sve.vector<16xi8>,
                    %c: !arm_sve.vector<4xi32>)
     -> !arm_sve.vector<4xi32> {
   // CHECK: llvm_arm_sve.sdot
   %0 = arm_sve.sdot %c, %a, %b :
                !arm_sve.vector<16xi8> to !arm_sve.vector<4xi32>

mlir/test/Conversion/VectorToLLVM/vector-mask-to-llvm.mlir

 // RUN: mlir-opt %s --convert-vector-to-llvm='enable-index-optimizations=1' | FileCheck %s --check-prefix=CMP32
 // RUN: mlir-opt %s --convert-vector-to-llvm='enable-index-optimizations=0' | FileCheck %s --check-prefix=CMP64
 
-// CMP32-LABEL: llvm.func @genbool_var_1d(
-// CMP32-SAME: %[[A:.*]]: i64)
-// CMP32: %[[T0:.*]] = llvm.mlir.constant(dense<[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]> : vector<11xi32>) : vector<11xi32>
-// CMP32: %[[T1:.*]] = llvm.trunc %[[A]] : i64 to i32
-// CMP32: %[[T2:.*]] = llvm.mlir.undef : vector<11xi32>
-// CMP32: %[[T3:.*]] = llvm.mlir.constant(0 : i32) : i32
-// CMP32: %[[T4:.*]] = llvm.insertelement %[[T1]], %[[T2]][%[[T3]] : i32] : vector<11xi32>
-// CMP32: %[[T5:.*]] = llvm.shufflevector %[[T4]], %[[T2]] [0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32] : vector<11xi32>, vector<11xi32>
-// CMP32: %[[T6:.*]] = llvm.icmp "slt" %[[T0]], %[[T5]] : vector<11xi32>
-// CMP32: llvm.return %[[T6]] : vector<11xi1>
-
-// CMP64-LABEL: llvm.func @genbool_var_1d(
-// CMP64-SAME: %[[A:.*]]: i64)
-// CMP64: %[[T0:.*]] = llvm.mlir.constant(dense<[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]> : vector<11xi64>) : vector<11xi64>
-// CMP64: %[[T1:.*]] = llvm.mlir.undef : vector<11xi64>
-// CMP64: %[[T2:.*]] = llvm.mlir.constant(0 : i32) : i32
-// CMP64: %[[T3:.*]] = llvm.insertelement %[[A]], %[[T1]][%[[T2]] : i32] : vector<11xi64>
-// CMP64: %[[T4:.*]] = llvm.shufflevector %[[T3]], %[[T1]] [0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32] : vector<11xi64>, vector<11xi64>
-// CMP64: %[[T5:.*]] = llvm.icmp "slt" %[[T0]], %[[T4]] : vector<11xi64>
-// CMP64: llvm.return %[[T5]] : vector<11xi1>
+// CMP32-LABEL: @genbool_var_1d(
+// CMP32-SAME: %[[ARG:.*]]: index)
+// CMP32: %[[A:.*]] = llvm.mlir.cast %[[ARG]] : index to i64
+// CMP32: %[[T0:.*]] = constant dense<[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]> : vector<11xi32>
+// CMP32: %[[T1:.*]] = trunci %[[A]] : i64 to i32
+// CMP32: %[[T2:.*]] = splat %[[T1]] : vector<11xi32>
+// CMP32: %[[T3:.*]] = cmpi slt, %[[T0]], %[[T2]] : vector<11xi32>
+// CMP32: return %[[T3]] : vector<11xi1>
+
+// CMP64-LABEL: @genbool_var_1d(
+// CMP64-SAME: %[[ARG:.*]]: index)
+// CMP64: %[[A:.*]] = llvm.mlir.cast %[[ARG]] : index to i64
+// CMP64: %[[T0:.*]] = constant dense<[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]> : vector<11xi64>
+// CMP64: %[[T1:.*]] = splat %[[A]] : vector<11xi64>
+// CMP64: %[[T2:.*]] = cmpi slt, %[[T0]], %[[T1]] : vector<11xi64>
+// CMP64: return %[[T2]] : vector<11xi1>
 
 func @genbool_var_1d(%arg0: index) -> vector<11xi1> {
   %0 = vector.create_mask %arg0 : vector<11xi1>
   return %0 : vector<11xi1>
 }
 
-// CMP32-LABEL: llvm.func @transfer_read_1d
-// CMP32: %[[C:.*]] = llvm.mlir.constant(dense<[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]> : vector<16xi32>) : vector<16xi32>
-// CMP32: %[[A:.*]] = llvm.add %{{.*}}, %[[C]] : vector<16xi32>
-// CMP32: %[[M:.*]] = llvm.icmp "slt" %[[A]], %{{.*}} : vector<16xi32>
+// CMP32-LABEL: @transfer_read_1d
+// CMP32: %[[C:.*]] = constant dense<[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]> : vector<16xi32>
+// CMP32: %[[A:.*]] = addi %{{.*}}, %[[C]] : vector<16xi32>
+// CMP32: %[[M:.*]] = cmpi slt, %[[A]], %{{.*}} : vector<16xi32>
 // CMP32: %[[L:.*]] = llvm.intr.masked.load %{{.*}}, %[[M]], %{{.*}}
-// CMP32: llvm.return %[[L]] : vector<16xf32>
+// CMP32: return %[[L]] : vector<16xf32>
 
-// CMP64-LABEL: llvm.func @transfer_read_1d
-// CMP64: %[[C:.*]] = llvm.mlir.constant(dense<[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]> : vector<16xi64>) : vector<16xi64>
-// CMP64: %[[A:.*]] = llvm.add %{{.*}}, %[[C]] : vector<16xi64>
-// CMP64: %[[M:.*]] = llvm.icmp "slt" %[[A]], %{{.*}} : vector<16xi64>
+// CMP64-LABEL: @transfer_read_1d
+// CMP64: %[[C:.*]] = constant dense<[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15]> : vector<16xi64>
+// CMP64: %[[A:.*]] = addi %{{.*}}, %[[C]] : vector<16xi64>
+// CMP64: %[[M:.*]] = cmpi slt, %[[A]], %{{.*}} : vector<16xi64>
 // CMP64: %[[L:.*]] = llvm.intr.masked.load %{{.*}}, %[[M]], %{{.*}}
-// CMP64: llvm.return %[[L]] : vector<16xf32>
+// CMP64: return %[[L]] : vector<16xf32>
 
 func @transfer_read_1d(%A : memref<?xf32>, %i: index) -> vector<16xf32> {
   %d = constant -1.0: f32
   %f = vector.transfer_read %A[%i], %d {permutation_map = affine_map<(d0) -> (d0)>} : memref<?xf32>, vector<16xf32>
   return %f : vector<16xf32>
 }

mlir/test/Conversion/VectorToLLVM/vector-reduction-to-llvm.mlir

 // RUN: mlir-opt %s -convert-vector-to-llvm | FileCheck %s
 // RUN: mlir-opt %s -convert-vector-to-llvm='reassociate-fp-reductions' | FileCheck %s --check-prefix=REASSOC
 
 //
-// CHECK-LABEL: llvm.func @reduce_add_f32(
+// CHECK-LABEL: @reduce_add_f32(
 // CHECK-SAME: %[[A:.*]]: vector<16xf32>)
 //      CHECK: %[[C:.*]] = llvm.mlir.constant(0.000000e+00 : f32) : f32
 //      CHECK: %[[V:.*]] = "llvm.intr.vector.reduce.fadd"(%[[C]], %[[A]])
 // CHECK-SAME: {reassoc = false} : (f32, vector<16xf32>) -> f32
-//      CHECK: llvm.return %[[V]] : f32
+//      CHECK: return %[[V]] : f32
 //
-// REASSOC-LABEL: llvm.func @reduce_add_f32(
+// REASSOC-LABEL: @reduce_add_f32(
 // REASSOC-SAME: %[[A:.*]]: vector<16xf32>)
 //      REASSOC: %[[C:.*]] = llvm.mlir.constant(0.000000e+00 : f32) : f32
 //      REASSOC: %[[V:.*]] = "llvm.intr.vector.reduce.fadd"(%[[C]], %[[A]])
 // REASSOC-SAME: {reassoc = true} : (f32, vector<16xf32>) -> f32
-//      REASSOC: llvm.return %[[V]] : f32
+//      REASSOC: return %[[V]] : f32
 //
 func @reduce_add_f32(%arg0: vector<16xf32>) -> f32 {
   %0 = vector.reduction "add", %arg0 : vector<16xf32> into f32
   return %0 : f32
 }
 
 //
-// CHECK-LABEL: llvm.func @reduce_mul_f32(
+// CHECK-LABEL: @reduce_mul_f32(
 // CHECK-SAME: %[[A:.*]]: vector<16xf32>)
 //      CHECK: %[[C:.*]] = llvm.mlir.constant(1.000000e+00 : f32) : f32
 //      CHECK: %[[V:.*]] = "llvm.intr.vector.reduce.fmul"(%[[C]], %[[A]])
 // CHECK-SAME: {reassoc = false} : (f32, vector<16xf32>) -> f32
-//      CHECK: llvm.return %[[V]] : f32
+//      CHECK: return %[[V]] : f32
 //
-// REASSOC-LABEL: llvm.func @reduce_mul_f32(
+// REASSOC-LABEL: @reduce_mul_f32(
 // REASSOC-SAME: %[[A:.*]]: vector<16xf32>)
 //      REASSOC: %[[C:.*]] = llvm.mlir.constant(1.000000e+00 : f32) : f32
 //      REASSOC: %[[V:.*]] = "llvm.intr.vector.reduce.fmul"(%[[C]], %[[A]])
 // REASSOC-SAME: {reassoc = true} : (f32, vector<16xf32>) -> f32
-//      REASSOC: llvm.return %[[V]] : f32
+//      REASSOC: return %[[V]] : f32
 //
 func @reduce_mul_f32(%arg0: vector<16xf32>) -> f32 {
   %0 = vector.reduction "mul", %arg0 : vector<16xf32> into f32
   return %0 : f32
 }

mlir/test/Conversion/VectorToLLVM/vector-to-llvm.mlir

 // RUN: mlir-opt %s -convert-vector-to-llvm -split-input-file | FileCheck %s
 
 
 func @bitcast_f32_to_i32_vector(%input: vector<16xf32>) -> vector<16xi32> {
   %0 = vector.bitcast %input : vector<16xf32> to vector<16xi32>
   return %0 : vector<16xi32>
 }
 
-// CHECK-LABEL: llvm.func @bitcast_f32_to_i32_vector(
+// CHECK-LABEL: @bitcast_f32_to_i32_vector
 // CHECK-SAME:  %[[input:.*]]: vector<16xf32>
 // CHECK:       llvm.bitcast %[[input]] : vector<16xf32> to vector<16xi32>
 
 // -----
 
 func @bitcast_i8_to_f32_vector(%input: vector<64xi8>) -> vector<16xf32> {
   %0 = vector.bitcast %input : vector<64xi8> to vector<16xf32>
   return %0 : vector<16xf32>
 }
 
-// CHECK-LABEL: llvm.func @bitcast_i8_to_f32_vector(
+// CHECK-LABEL: @bitcast_i8_to_f32_vector
 // CHECK-SAME:  %[[input:.*]]: vector<64xi8>
 // CHECK:       llvm.bitcast %[[input]] : vector<64xi8> to vector<16xf32>
 
 // -----
 
 func @broadcast_vec1d_from_scalar(%arg0: f32) -> vector<2xf32> {
   %0 = vector.broadcast %arg0 : f32 to vector<2xf32>
   return %0 : vector<2xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_vec1d_from_scalar(
+// CHECK-LABEL: @broadcast_vec1d_from_scalar
 // CHECK-SAME:  %[[A:.*]]: f32)
-// CHECK:       %[[T0:.*]] = llvm.mlir.undef : vector<2xf32>
-// CHECK:       %[[T1:.*]] = llvm.mlir.constant(0 : i32) : i32
-// CHECK:       %[[T2:.*]] = llvm.insertelement %[[A]], %[[T0]][%[[T1]] : i32] : vector<2xf32>
-// CHECK:       %[[T3:.*]] = llvm.shufflevector %[[T2]], %[[T0]] [0 : i32, 0 : i32] : vector<2xf32>, vector<2xf32>
-// CHECK:       llvm.return %[[T3]] : vector<2xf32>
+// CHECK:       %[[T0:.*]] = splat %[[A]] : vector<2xf32>
+// CHECK:       return %[[T0]] : vector<2xf32>
 
 // -----
 
 func @broadcast_vec2d_from_scalar(%arg0: f32) -> vector<2x3xf32> {
   %0 = vector.broadcast %arg0 : f32 to vector<2x3xf32>
   return %0 : vector<2x3xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_vec2d_from_scalar(
+// CHECK-LABEL: @broadcast_vec2d_from_scalar(
 // CHECK-SAME:  %[[A:.*]]: f32)
-// CHECK:       %[[T0:.*]] = llvm.mlir.undef : !llvm.array<2 x vector<3xf32>>
-// CHECK:       %[[T1:.*]] = llvm.mlir.undef : vector<3xf32>
-// CHECK:       %[[T2:.*]] = llvm.mlir.constant(0 : i32) : i32
-// CHECK:       %[[T3:.*]] = llvm.insertelement %[[A]], %[[T1]][%[[T2]] : i32] : vector<3xf32>
-// CHECK:       %[[T4:.*]] = llvm.shufflevector %[[T3]], %[[T3]] [0 : i32, 0 : i32, 0 : i32] : vector<3xf32>, vector<3xf32>
-// CHECK:       %[[T5:.*]] = llvm.insertvalue %[[T4]], %[[T0]][0] : !llvm.array<2 x vector<3xf32>>
-// CHECK:       %[[T6:.*]] = llvm.insertvalue %[[T4]], %[[T5]][1] : !llvm.array<2 x vector<3xf32>>
-// CHECK:       llvm.return %[[T6]] : !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T0:.*]] = splat %[[A]] : vector<2x3xf32>
+// CHECK:       return %[[T0]] : vector<2x3xf32>
 
 // -----
 
 func @broadcast_vec3d_from_scalar(%arg0: f32) -> vector<2x3x4xf32> {
   %0 = vector.broadcast %arg0 : f32 to vector<2x3x4xf32>
   return %0 : vector<2x3x4xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_vec3d_from_scalar(
+// CHECK-LABEL: @broadcast_vec3d_from_scalar(
 // CHECK-SAME:  %[[A:.*]]: f32)
-// CHECK:       %[[T0:.*]] = llvm.mlir.undef : !llvm.array<2 x array<3 x vector<4xf32>>>
-// CHECK:       %[[T1:.*]] = llvm.mlir.undef : vector<4xf32>
-// CHECK:       %[[T2:.*]] = llvm.mlir.constant(0 : i32) : i32
-// CHECK:       %[[T3:.*]] = llvm.insertelement %[[A]], %[[T1]][%[[T2]] : i32] : vector<4xf32>
-// CHECK:       %[[T4:.*]] = llvm.shufflevector %[[T3]], %[[T3]] [0 : i32, 0 : i32, 0 : i32, 0 : i32] : vector<4xf32>, vector<4xf32>
-// CHECK:       %[[T5:.*]] = llvm.insertvalue %[[T4]], %[[T0]][0, 0] : !llvm.array<2 x array<3 x vector<4xf32>>>
-// CHECK:       %[[T6:.*]] = llvm.insertvalue %[[T4]], %[[T5]][0, 1] : !llvm.array<2 x array<3 x vector<4xf32>>>
-// CHECK:       %[[T7:.*]] = llvm.insertvalue %[[T4]], %[[T6]][0, 2] : !llvm.array<2 x array<3 x vector<4xf32>>>
-// CHECK:       %[[T8:.*]] = llvm.insertvalue %[[T4]], %[[T7]][1, 0] : !llvm.array<2 x array<3 x vector<4xf32>>>
-// CHECK:       %[[T9:.*]] = llvm.insertvalue %[[T4]], %[[T8]][1, 1] : !llvm.array<2 x array<3 x vector<4xf32>>>
-// CHECK:       %[[T10:.*]] = llvm.insertvalue %[[T4]], %[[T9]][1, 2] : !llvm.array<2 x array<3 x vector<4xf32>>>
-// CHECK:       llvm.return %[[T10]] : !llvm.array<2 x array<3 x vector<4xf32>>>
+// CHECK:       %[[T0:.*]] = splat %[[A]] : vector<2x3x4xf32>
+// CHECK:       return %[[T0]] : vector<2x3x4xf32>
 
 // -----
 
 func @broadcast_vec1d_from_vec1d(%arg0: vector<2xf32>) -> vector<2xf32> {
   %0 = vector.broadcast %arg0 : vector<2xf32> to vector<2xf32>
   return %0 : vector<2xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_vec1d_from_vec1d(
+// CHECK-LABEL: @broadcast_vec1d_from_vec1d(
 // CHECK-SAME:  %[[A:.*]]: vector<2xf32>)
-// CHECK:       llvm.return %[[A]] : vector<2xf32>
+// CHECK:       return %[[A]] : vector<2xf32>
 
 // -----
 
 func @broadcast_vec2d_from_vec1d(%arg0: vector<2xf32>) -> vector<3x2xf32> {
   %0 = vector.broadcast %arg0 : vector<2xf32> to vector<3x2xf32>
   return %0 : vector<3x2xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_vec2d_from_vec1d(
+// CHECK-LABEL: @broadcast_vec2d_from_vec1d(
 // CHECK-SAME:  %[[A:.*]]: vector<2xf32>)
-// CHECK:       %[[T0:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<3x2xf32>) : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T1:.*]] = llvm.insertvalue %[[A]], %[[T0]][0] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T2:.*]] = llvm.insertvalue %[[A]], %[[T1]][1] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T3:.*]] = llvm.insertvalue %[[A]], %[[T2]][2] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       llvm.return %[[T3]] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T0:.*]] = constant dense<0.000000e+00> : vector<3x2xf32>
+// CHECK:       %[[T1:.*]] = llvm.mlir.cast %[[T0]] : vector<3x2xf32> to !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T2:.*]] = llvm.insertvalue %[[A]], %[[T1]][0] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T3:.*]] = llvm.insertvalue %[[A]], %[[T2]][1] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T4:.*]] = llvm.insertvalue %[[A]], %[[T3]][2] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T5:.*]] = llvm.mlir.cast %[[T4]] : !llvm.array<3 x vector<2xf32>> to vector<3x2xf32>
+// CHECK:       return %[[T5]] : vector<3x2xf32>
 
 // -----
 
 func @broadcast_vec3d_from_vec1d(%arg0: vector<2xf32>) -> vector<4x3x2xf32> {
   %0 = vector.broadcast %arg0 : vector<2xf32> to vector<4x3x2xf32>
   return %0 : vector<4x3x2xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_vec3d_from_vec1d(
+// CHECK-LABEL: @broadcast_vec3d_from_vec1d(
 // CHECK-SAME:  %[[A:.*]]: vector<2xf32>)
-// CHECK:       %[[T0:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<3x2xf32>) : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T1:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<4x3x2xf32>) : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T2:.*]] = llvm.insertvalue %[[A]], %[[T0]][0] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T3:.*]] = llvm.insertvalue %[[A]], %[[T2]][1] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T4:.*]] = llvm.insertvalue %[[A]], %[[T3]][2] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T5:.*]] = llvm.insertvalue %[[T4]], %[[T1]][0] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T6:.*]] = llvm.insertvalue %[[T4]], %[[T5]][1] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T7:.*]] = llvm.insertvalue %[[T4]], %[[T6]][2] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T8:.*]] = llvm.insertvalue %[[T4]], %[[T7]][3] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       llvm.return %[[T8]] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T0:.*]] = constant dense<0.000000e+00> : vector<3x2xf32>
+// CHECK:       %[[T1:.*]] = constant dense<0.000000e+00> : vector<4x3x2xf32>
+
+// CHECK:       %[[T2:.*]] = llvm.mlir.cast %[[T0]] : vector<3x2xf32> to !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T3:.*]] = llvm.insertvalue %[[A]], %[[T2]][0] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T4:.*]] = llvm.insertvalue %[[A]], %[[T3]][1] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T5:.*]] = llvm.insertvalue %[[A]], %[[T4]][2] : !llvm.array<3 x vector<2xf32>>
+
+// CHECK:       %[[T6:.*]] = llvm.mlir.cast %[[T1]] : vector<4x3x2xf32> to !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T7:.*]] = llvm.insertvalue %[[T5]], %[[T6]][0] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T8:.*]] = llvm.insertvalue %[[T5]], %[[T7]][1] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T9:.*]] = llvm.insertvalue %[[T5]], %[[T8]][2] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T10:.*]] = llvm.insertvalue %[[T5]], %[[T9]][3] : !llvm.array<4 x array<3 x vector<2xf32>>>
+
+// CHECK:       %[[T11:.*]] = llvm.mlir.cast %[[T10]] : !llvm.array<4 x array<3 x vector<2xf32>>> to vector<4x3x2xf32>
+// CHECK:       return %[[T11]] : vector<4x3x2xf32>
 
 // -----
 
 func @broadcast_vec3d_from_vec2d(%arg0: vector<3x2xf32>) -> vector<4x3x2xf32> {
   %0 = vector.broadcast %arg0 : vector<3x2xf32> to vector<4x3x2xf32>
   return %0 : vector<4x3x2xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_vec3d_from_vec2d(
-// CHECK-SAME:  %[[A:.*]]: !llvm.array<3 x vector<2xf32>>)
-// CHECK:       %[[T0:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<4x3x2xf32>) : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T1:.*]] = llvm.insertvalue %[[A]], %[[T0]][0] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T2:.*]] = llvm.insertvalue %[[A]], %[[T1]][1] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T3:.*]] = llvm.insertvalue %[[A]], %[[T2]][2] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T4:.*]] = llvm.insertvalue %[[A]], %[[T3]][3] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       llvm.return %[[T4]] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK-LABEL: @broadcast_vec3d_from_vec2d(
+// CHECK-SAME:  %[[A:.*]]: vector<3x2xf32>)
+// CHECK:       %[[T0:.*]] = constant dense<0.000000e+00> : vector<4x3x2xf32>
+// CHECK:       %[[T1:.*]] = llvm.mlir.cast %[[A]] : vector<3x2xf32> to !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T2:.*]] = llvm.mlir.cast %[[T0]] : vector<4x3x2xf32> to !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T3:.*]] = llvm.insertvalue %[[T1]], %[[T2]][0] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T4:.*]] = llvm.mlir.cast %[[A]] : vector<3x2xf32> to !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T5:.*]] = llvm.insertvalue %[[T4]], %[[T3]][1] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T6:.*]] = llvm.mlir.cast %[[A]] : vector<3x2xf32> to !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T7:.*]] = llvm.insertvalue %[[T6]], %[[T5]][2] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T8:.*]] = llvm.mlir.cast %[[A]] : vector<3x2xf32> to !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T9:.*]] = llvm.insertvalue %[[T8]], %[[T7]][3] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T10:.*]] = llvm.mlir.cast %[[T9]] : !llvm.array<4 x array<3 x vector<2xf32>>> to vector<4x3x2xf32>
+// CHECK:       return %[[T10]] : vector<4x3x2xf32>
+
 
 // -----
 
 func @broadcast_stretch(%arg0: vector<1xf32>) -> vector<4xf32> {
   %0 = vector.broadcast %arg0 : vector<1xf32> to vector<4xf32>
   return %0 : vector<4xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_stretch(
+// CHECK-LABEL: @broadcast_stretch(
 // CHECK-SAME:  %[[A:.*]]: vector<1xf32>)
-// CHECK:       %[[T0:.*]] = llvm.mlir.constant(0 : i64) : i64
-// CHECK:       %[[T1:.*]] = llvm.extractelement %[[A]][%[[T0]] : i64] : vector<1xf32>
-// CHECK:       %[[T2:.*]] = llvm.mlir.undef : vector<4xf32>
-// CHECK:       %[[T3:.*]] = llvm.mlir.constant(0 : i32) : i32
-// CHECK:       %[[T4:.*]] = llvm.insertelement %[[T1]], %[[T2]][%3 : i32] : vector<4xf32>
-// CHECK:       %[[T5:.*]] = llvm.shufflevector %[[T4]], %[[T2]] [0 : i32, 0 : i32, 0 : i32, 0 : i32] : vector<4xf32>, vector<4xf32>
-// CHECK:       llvm.return %[[T5]] : vector<4xf32>
+// CHECK:       %[[T1:.*]] = llvm.mlir.constant(0 : i64) : i64
+// CHECK:       %[[T2:.*]] = llvm.extractelement %[[A]]{{\[}}%[[T1]] : i64] : vector<1xf32>
+// CHECK:       %[[T3:.*]] = splat %[[T2]] : vector<4xf32>
+// CHECK:       return %[[T3]] : vector<4xf32>
 
 // -----
 
 func @broadcast_stretch_at_start(%arg0: vector<1x4xf32>) -> vector<3x4xf32> {
   %0 = vector.broadcast %arg0 : vector<1x4xf32> to vector<3x4xf32>
   return %0 : vector<3x4xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_stretch_at_start(
-// CHECK-SAME:  %[[A:.*]]: !llvm.array<1 x vector<4xf32>>)
-// CHECK:       %[[T0:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<3x4xf32>) : !llvm.array<3 x vector<4xf32>>
-// CHECK:       %[[T1:.*]] = llvm.extractvalue %[[A]][0] : !llvm.array<1 x vector<4xf32>>
-// CHECK:       %[[T2:.*]] = llvm.insertvalue %[[T1]], %[[T0]][0] : !llvm.array<3 x vector<4xf32>>
-// CHECK:       %[[T3:.*]] = llvm.insertvalue %[[T1]], %[[T2]][1] : !llvm.array<3 x vector<4xf32>>
-// CHECK:       %[[T4:.*]] = llvm.insertvalue %[[T1]], %[[T3]][2] : !llvm.array<3 x vector<4xf32>>
-// CHECK:       llvm.return %[[T4]] : !llvm.array<3 x vector<4xf32>>
+// CHECK-LABEL: @broadcast_stretch_at_start(
+// CHECK-SAME:  %[[A:.*]]: vector<1x4xf32>)
+// CHECK:       %[[T1:.*]] = constant dense<0.000000e+00> : vector<3x4xf32>
+// CHECK:       %[[T2:.*]] = llvm.mlir.cast %[[A]] : vector<1x4xf32> to !llvm.array<1 x vector<4xf32>>
+// CHECK:       %[[T3:.*]] = llvm.extractvalue %[[T2]][0] : !llvm.array<1 x vector<4xf32>>
+// CHECK:       %[[T4:.*]] = llvm.mlir.cast %[[T1]] : vector<3x4xf32> to !llvm.array<3 x vector<4xf32>>
+// CHECK:       %[[T5:.*]] = llvm.insertvalue %[[T3]], %[[T4]][0] : !llvm.array<3 x vector<4xf32>>
+// CHECK:       %[[T6:.*]] = llvm.insertvalue %[[T3]], %[[T5]][1] : !llvm.array<3 x vector<4xf32>>
+// CHECK:       %[[T7:.*]] = llvm.insertvalue %[[T3]], %[[T6]][2] : !llvm.array<3 x vector<4xf32>>
+// CHECK:       %[[T8:.*]] = llvm.mlir.cast %[[T7]] : !llvm.array<3 x vector<4xf32>> to vector<3x4xf32>
+// CHECK:       return %[[T8]] : vector<3x4xf32>
 
 // -----
 
 func @broadcast_stretch_at_end(%arg0: vector<4x1xf32>) -> vector<4x3xf32> {
   %0 = vector.broadcast %arg0 : vector<4x1xf32> to vector<4x3xf32>
   return %0 : vector<4x3xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_stretch_at_end(
-// CHECK-SAME:  %[[A:.*]]: !llvm.array<4 x vector<1xf32>>)
-// CHECK:       %[[T0:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<4x3xf32>) : !llvm.array<4 x vector<3xf32>>
-// CHECK:       %[[T1:.*]] = llvm.extractvalue %[[A]][0] : !llvm.array<4 x vector<1xf32>>
-// CHECK:       %[[T2:.*]] = llvm.mlir.constant(0 : i64) : i64
-// CHECK:       %[[T3:.*]] = llvm.extractelement %[[T1]][%[[T2]] : i64] : vector<1xf32>
-// CHECK:       %[[T4:.*]] = llvm.mlir.undef : vector<3xf32>
-// CHECK:       %[[T5:.*]] = llvm.mlir.constant(0 : i32) : i32
-// CHECK:       %[[T6:.*]] = llvm.insertelement %[[T3]], %[[T4]][%[[T5]] : i32] : vector<3xf32>
-// CHECK:       %[[T7:.*]] = llvm.shufflevector %[[T6]], %[[T4]] [0 : i32, 0 : i32, 0 : i32] : vector<3xf32>, vector<3xf32>
-// CHECK:       %[[T8:.*]] = llvm.insertvalue %[[T7]], %[[T0]][0] : !llvm.array<4 x vector<3xf32>>
-// CHECK:       %[[T9:.*]] = llvm.extractvalue %[[A]][1] : !llvm.array<4 x vector<1xf32>>
-// CHECK:       %[[T10:.*]] = llvm.mlir.constant(0 : i64) : i64
-// CHECK:       %[[T11:.*]] = llvm.extractelement %[[T9]][%[[T10]] : i64] : vector<1xf32>
-// CHECK:       %[[T12:.*]] = llvm.mlir.undef : vector<3xf32>
-// CHECK:       %[[T13:.*]] = llvm.mlir.constant(0 : i32) : i32
-// CHECK:       %[[T14:.*]] = llvm.insertelement %[[T11]], %[[T12]][%[[T13]] : i32] : vector<3xf32>
-// CHECK:       %[[T15:.*]] = llvm.shufflevector %[[T14]], %[[T12]] [0 : i32, 0 : i32, 0 : i32] : vector<3xf32>, vector<3xf32>
-// CHECK:       %[[T16:.*]] = llvm.insertvalue %[[T15]], %[[T8]][1] : !llvm.array<4 x vector<3xf32>>
-// CHECK:       %[[T17:.*]] = llvm.extractvalue %[[A]][2] : !llvm.array<4 x vector<1xf32>>
-// CHECK:       %[[T18:.*]] = llvm.mlir.constant(0 : i64) : i64
-// CHECK:       %[[T19:.*]] = llvm.extractelement %[[T17]][%[[T18]] : i64] : vector<1xf32>
-// CHECK:       %[[T20:.*]] = llvm.mlir.undef : vector<3xf32>
-// CHECK:       %[[T21:.*]] = llvm.mlir.constant(0 : i32) : i32
-// CHECK:       %[[T22:.*]] = llvm.insertelement %[[T19]], %[[T20]][%[[T21]] : i32] : vector<3xf32>
-// CHECK:       %[[T23:.*]] = llvm.shufflevector %[[T22]], %[[T20]] [0 : i32, 0 : i32, 0 : i32] : vector<3xf32>, vector<3xf32>
-// CHECK:       %[[T24:.*]] = llvm.insertvalue %[[T23]], %[[T16]][2] : !llvm.array<4 x vector<3xf32>>
-// CHECK:       %[[T25:.*]] = llvm.extractvalue %[[A]][3] : !llvm.array<4 x vector<1xf32>>
-// CHECK:       %[[T26:.*]] = llvm.mlir.constant(0 : i64) : i64
-// CHECK:       %[[T27:.*]] = llvm.extractelement %[[T25]][%[[T26]] : i64] : vector<1xf32>
-// CHECK:       %[[T28:.*]] = llvm.mlir.undef : vector<3xf32>
-// CHECK:       %[[T29:.*]] = llvm.mlir.constant(0 : i32) : i32
-// CHECK:       %[[T30:.*]] = llvm.insertelement %[[T27]], %[[T28]][%[[T29]] : i32] : vector<3xf32>
-// CHECK:       %[[T31:.*]] = llvm.shufflevector %[[T30]], %[[T28]] [0 : i32, 0 : i32, 0 : i32] : vector<3xf32>, vector<3xf32>
-// CHECK:       %[[T32:.*]] = llvm.insertvalue %[[T31]], %[[T24]][3] : !llvm.array<4 x vector<3xf32>>
-// CHECK:       llvm.return %[[T32]] : !llvm.array<4 x vector<3xf32>>
+// CHECK-LABEL: @broadcast_stretch_at_end(
+// CHECK-SAME:  %[[A:.*]]: vector<4x1xf32>)
+// CHECK:       %[[T1:.*]] = constant dense<0.000000e+00> : vector<4x3xf32>
+// CHECK:       %[[T2:.*]] = llvm.mlir.cast %[[A]] : vector<4x1xf32> to !llvm.array<4 x vector<1xf32>>
+// CHECK:       %[[T3:.*]] = llvm.extractvalue %[[T2]][0] : !llvm.array<4 x vector<1xf32>>
+// CHECK:       %[[T4:.*]] = llvm.mlir.constant(0 : i64) : i64
+// CHECK:       %[[T5:.*]] = llvm.extractelement %[[T3]]{{\[}}%[[T4]] : i64] : vector<1xf32>
+// CHECK:       %[[T6:.*]] = splat %[[T5]] : vector<3xf32>
+// CHECK:       %[[T7:.*]] = llvm.mlir.cast %[[T1]] : vector<4x3xf32> to !llvm.array<4 x vector<3xf32>>
+// CHECK:       %[[T8:.*]] = llvm.insertvalue %[[T6]], %[[T7]][0] : !llvm.array<4 x vector<3xf32>>
+// CHECK:       %[[T9:.*]] = llvm.mlir.cast %[[A]] : vector<4x1xf32> to !llvm.array<4 x vector<1xf32>>
+// CHECK:       %[[T10:.*]] = llvm.extractvalue %[[T9]][1] : !llvm.array<4 x vector<1xf32>>
+// CHECK:       %[[T11:.*]] = llvm.mlir.constant(0 : i64) : i64
+// CHECK:       %[[T12:.*]] = llvm.extractelement %[[T10]]{{\[}}%[[T11]] : i64] : vector<1xf32>
+// CHECK:       %[[T13:.*]] = splat %[[T12]] : vector<3xf32>
+// CHECK:       %[[T14:.*]] = llvm.insertvalue %[[T13]], %[[T8]][1] : !llvm.array<4 x vector<3xf32>>
+// CHECK:       %[[T15:.*]] = llvm.mlir.cast %[[A]] : vector<4x1xf32> to !llvm.array<4 x vector<1xf32>>
+// CHECK:       %[[T16:.*]] = llvm.extractvalue %[[T15]][2] : !llvm.array<4 x vector<1xf32>>
+// CHECK:       %[[T17:.*]] = llvm.mlir.constant(0 : i64) : i64
+// CHECK:       %[[T18:.*]] = llvm.extractelement %[[T16]]{{\[}}%[[T17]] : i64] : vector<1xf32>
+// CHECK:       %[[T19:.*]] = splat %[[T18]] : vector<3xf32>
+// CHECK:       %[[T20:.*]] = llvm.insertvalue %[[T19]], %[[T14]][2] : !llvm.array<4 x vector<3xf32>>
+// CHECK:       %[[T21:.*]] = llvm.mlir.cast %[[A]] : vector<4x1xf32> to !llvm.array<4 x vector<1xf32>>
+// CHECK:       %[[T22:.*]] = llvm.extractvalue %[[T21]][3] : !llvm.array<4 x vector<1xf32>>
+// CHECK:       %[[T23:.*]] = llvm.mlir.constant(0 : i64) : i64
+// CHECK:       %[[T24:.*]] = llvm.extractelement %[[T22]]{{\[}}%[[T23]] : i64] : vector<1xf32>
+// CHECK:       %[[T25:.*]] = splat %[[T24]] : vector<3xf32>
+// CHECK:       %[[T26:.*]] = llvm.insertvalue %[[T25]], %[[T20]][3] : !llvm.array<4 x vector<3xf32>>
+// CHECK:       %[[T27:.*]] = llvm.mlir.cast %[[T26]] : !llvm.array<4 x vector<3xf32>> to vector<4x3xf32>
+// CHECK:       return %[[T27]] : vector<4x3xf32>
 
 // -----
 
 func @broadcast_stretch_in_middle(%arg0: vector<4x1x2xf32>) -> vector<4x3x2xf32> {
   %0 = vector.broadcast %arg0 : vector<4x1x2xf32> to vector<4x3x2xf32>
   return %0 : vector<4x3x2xf32>
 }
-// CHECK-LABEL: llvm.func @broadcast_stretch_in_middle(
-// CHECK-SAME:  %[[A:.*]]: !llvm.array<4 x array<1 x vector<2xf32>>>)
-// CHECK:       %[[T0:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<4x3x2xf32>) : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T1:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<3x2xf32>) : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T2:.*]] = llvm.extractvalue %[[A]][0, 0] : !llvm.array<4 x array<1 x vector<2xf32>>>
-// CHECK:       %[[T4:.*]] = llvm.insertvalue %[[T2]], %[[T1]][0] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T5:.*]] = llvm.insertvalue %[[T2]], %[[T4]][1] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T6:.*]] = llvm.insertvalue %[[T2]], %[[T5]][2] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T7:.*]] = llvm.insertvalue %[[T6]], %[[T0]][0] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T8:.*]] = llvm.extractvalue %[[A]][1, 0] : !llvm.array<4 x array<1 x vector<2xf32>>>
-// CHECK:       %[[T10:.*]] = llvm.insertvalue %[[T8]], %[[T1]][0] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T11:.*]] = llvm.insertvalue %[[T8]], %[[T10]][1] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T12:.*]] = llvm.insertvalue %[[T8]], %[[T11]][2] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T13:.*]] = llvm.insertvalue %[[T12]], %[[T7]][1] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T14:.*]] = llvm.extractvalue %[[A]][2, 0] : !llvm.array<4 x array<1 x vector<2xf32>>>
-// CHECK:       %[[T16:.*]] = llvm.insertvalue %[[T14]], %[[T1]][0] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T17:.*]] = llvm.insertvalue %[[T14]], %[[T16]][1] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T18:.*]] = llvm.insertvalue %[[T14]], %[[T17]][2] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T19:.*]] = llvm.insertvalue %[[T18]], %[[T13]][2] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       %[[T20:.*]] = llvm.extractvalue %[[A]][3, 0] : !llvm.array<4 x array<1 x vector<2xf32>>>
-// CHECK:       %[[T22:.*]] = llvm.insertvalue %[[T20]], %[[T1]][0] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T23:.*]] = llvm.insertvalue %[[T20]], %[[T22]][1] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T24:.*]] = llvm.insertvalue %[[T20]], %[[T23]][2] : !llvm.array<3 x vector<2xf32>>
-// CHECK:       %[[T25:.*]] = llvm.insertvalue %[[T24]], %[[T19]][3] : !llvm.array<4 x array<3 x vector<2xf32>>>
-// CHECK:       llvm.return %[[T25]] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK-LABEL: @broadcast_stretch_in_middle(
+// CHECK-SAME:  %[[A:.*]]: vector<4x1x2xf32>) -> vector<4x3x2xf32> {
+// CHECK:       %[[T1:.*]] = constant dense<0.000000e+00> : vector<4x3x2xf32>
+// CHECK:       %[[T2:.*]] = constant dense<0.000000e+00> : vector<3x2xf32>
+// CHECK:       %[[T3:.*]] = llvm.mlir.cast %[[A]] : vector<4x1x2xf32> to !llvm.array<4 x array<1 x vector<2xf32>>>
+// CHECK:       %[[T4:.*]] = llvm.extractvalue %[[T3]][0, 0] : !llvm.array<4 x array<1 x vector<2xf32>>>
+// CHECK:       %[[T5:.*]] = llvm.mlir.cast %[[T2]] : vector<3x2xf32> to !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T6:.*]] = llvm.insertvalue %[[T4]], %[[T5]][0] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T7:.*]] = llvm.insertvalue %[[T4]], %[[T6]][1] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T8:.*]] = llvm.insertvalue %[[T4]], %[[T7]][2] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T9:.*]] = llvm.mlir.cast %[[T1]] : vector<4x3x2xf32> to !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T10:.*]] = llvm.insertvalue %[[T8]], %[[T9]][0] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T11:.*]] = llvm.mlir.cast %[[A]] : vector<4x1x2xf32> to !llvm.array<4 x array<1 x vector<2xf32>>>
+// CHECK:       %[[T12:.*]] = llvm.extractvalue %[[T11]][1, 0] : !llvm.array<4 x array<1 x vector<2xf32>>>
+// CHECK:       %[[T13:.*]] = llvm.mlir.cast %[[T2]] : vector<3x2xf32> to !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T14:.*]] = llvm.insertvalue %[[T12]], %[[T13]][0] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T15:.*]] = llvm.insertvalue %[[T12]], %[[T14]][1] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T16:.*]] = llvm.insertvalue %[[T12]], %[[T15]][2] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T17:.*]] = llvm.insertvalue %[[T16]], %[[T10]][1] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T18:.*]] = llvm.mlir.cast %[[A]] : vector<4x1x2xf32> to !llvm.array<4 x array<1 x vector<2xf32>>>
+// CHECK:       %[[T19:.*]] = llvm.extractvalue %[[T18]][2, 0] : !llvm.array<4 x array<1 x vector<2xf32>>>
+// CHECK:       %[[T20:.*]] = llvm.mlir.cast %[[T2]] : vector<3x2xf32> to !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T21:.*]] = llvm.insertvalue %[[T19]], %[[T20]][0] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T22:.*]] = llvm.insertvalue %[[T19]], %[[T21]][1] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T23:.*]] = llvm.insertvalue %[[T19]], %[[T22]][2] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T24:.*]] = llvm.insertvalue %[[T23]], %[[T17]][2] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T25:.*]] = llvm.mlir.cast %[[A]] : vector<4x1x2xf32> to !llvm.array<4 x array<1 x vector<2xf32>>>
+// CHECK:       %[[T26:.*]] = llvm.extractvalue %[[T25]][3, 0] : !llvm.array<4 x array<1 x vector<2xf32>>>
+// CHECK:       %[[T27:.*]] = llvm.mlir.cast %[[T2]] : vector<3x2xf32> to !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T28:.*]] = llvm.insertvalue %[[T26]], %[[T27]][0] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T29:.*]] = llvm.insertvalue %[[T26]], %[[T28]][1] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T30:.*]] = llvm.insertvalue %[[T26]], %[[T29]][2] : !llvm.array<3 x vector<2xf32>>
+// CHECK:       %[[T31:.*]] = llvm.insertvalue %[[T30]], %[[T24]][3] : !llvm.array<4 x array<3 x vector<2xf32>>>
+// CHECK:       %[[T32:.*]] = llvm.mlir.cast %[[T31]] : !llvm.array<4 x array<3 x vector<2xf32>>> to vector<4x3x2xf32>
+// CHECK:       return %[[T32]] : vector<4x3x2xf32>
 
 // -----
 
 func @outerproduct(%arg0: vector<2xf32>, %arg1: vector<3xf32>) -> vector<2x3xf32> {
   %2 = vector.outerproduct %arg0, %arg1 : vector<2xf32>, vector<3xf32>
   return %2 : vector<2x3xf32>
 }
-// CHECK-LABEL: llvm.func @outerproduct(
-// CHECK-SAME: %[[A:.*]]: vector<2xf32>,
-// CHECK-SAME: %[[B:.*]]: vector<3xf32>)
-//      CHECK: %[[T0:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<2x3xf32>)
-//      CHECK: %[[T1:.*]] = llvm.mlir.constant(0 : i64) : i64
-//      CHECK: %[[T2:.*]] = llvm.extractelement %[[A]][%[[T1]] : i64] : vector<2xf32>
-//      CHECK: %[[T3:.*]] = llvm.mlir.undef : vector<3xf32>
-//      CHECK: %[[T4:.*]] = llvm.mlir.constant(0 : i32) : i32
-//      CHECK: %[[T5:.*]] = llvm.insertelement %[[T2]], %[[T3]][%4 : i32] : vector<3xf32>
-//      CHECK: %[[T6:.*]] = llvm.shufflevector %[[T5]], %[[T3]] [0 : i32, 0 : i32, 0 : i32] : vector<3xf32>, vector<3xf32>
-//      CHECK: %[[T7:.*]] = llvm.fmul %[[T6]], %[[B]] : vector<3xf32>
-//      CHECK: %[[T8:.*]] = llvm.insertvalue %[[T7]], %[[T0]][0] : !llvm.array<2 x vector<3xf32>>
-//      CHECK: %[[T9:.*]] = llvm.mlir.constant(1 : i64) : i64
-//      CHECK: %[[T10:.*]] = llvm.extractelement %[[A]][%9 : i64] : vector<2xf32>
-//      CHECK: %[[T11:.*]] = llvm.mlir.undef : vector<3xf32>
-//      CHECK: %[[T12:.*]] = llvm.mlir.constant(0 : i32) : i32
-//      CHECK: %[[T13:.*]] = llvm.insertelement %[[T10]], %[[T11]][%12 : i32] : vector<3xf32>
-//      CHECK: %[[T14:.*]] = llvm.shufflevector %[[T13]], %[[T11]] [0 : i32, 0 : i32, 0 : i32] : vector<3xf32>, vector<3xf32>
-//      CHECK: %[[T15:.*]] = llvm.fmul %[[T14]], %[[B]] : vector<3xf32>
-//      CHECK: %[[T16:.*]] = llvm.insertvalue %[[T15]], %[[T8]][1] : !llvm.array<2 x vector<3xf32>>
-//      CHECK: llvm.return %[[T16]] : !llvm.array<2 x vector<3xf32>>
+// CHECK-LABEL: @outerproduct(
+// CHECK-SAME:  %[[A:.*]]: vector<2xf32>,
+// CHECK-SAME:  %[[B:.*]]: vector<3xf32>)
+// CHECK:       %[[T2:.*]] = constant dense<0.000000e+00> : vector<2x3xf32>
+// CHECK:       %[[T3:.*]] = llvm.mlir.constant(0 : i64) : i64
+// CHECK:       %[[T4:.*]] = llvm.extractelement %[[A]]{{\[}}%[[T3]] : i64] : vector<2xf32>
+// CHECK:       %[[T5:.*]] = splat %[[T4]] : vector<3xf32>
+// CHECK:       %[[T6:.*]] = mulf %[[T5]], %[[B]] : vector<3xf32>
+// CHECK:       %[[T7:.*]] = llvm.mlir.cast %[[T2]] : vector<2x3xf32> to !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T8:.*]] = llvm.insertvalue %[[T6]], %[[T7]][0] : !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T9:.*]] = llvm.mlir.constant(1 : i64) : i64
+// CHECK:       %[[T10:.*]] = llvm.extractelement %[[A]]{{\[}}%[[T9]] : i64] : vector<2xf32>
+// CHECK:       %[[T11:.*]] = splat %[[T10]] : vector<3xf32>
+// CHECK:       %[[T12:.*]] = mulf %[[T11]], %[[B]] : vector<3xf32>
+// CHECK:       %[[T13:.*]] = llvm.insertvalue %[[T12]], %[[T8]][1] : !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T14:.*]] = llvm.mlir.cast %[[T13]] : !llvm.array<2 x vector<3xf32>> to vector<2x3xf32>
+// CHECK:       return %[[T14]] : vector<2x3xf32>
 
 // -----
 
 func @outerproduct_add(%arg0: vector<2xf32>, %arg1: vector<3xf32>, %arg2: vector<2x3xf32>) -> vector<2x3xf32> {
   %2 = vector.outerproduct %arg0, %arg1, %arg2 : vector<2xf32>, vector<3xf32>
   return %2 : vector<2x3xf32>
 }
-// CHECK-LABEL: llvm.func @outerproduct_add(
-// CHECK-SAME: %[[A:.*]]: vector<2xf32>,
-// CHECK-SAME: %[[B:.*]]: vector<3xf32>,
-// CHECK-SAME: %[[C:.*]]: !llvm.array<2 x vector<3xf32>>)
-//      CHECK: %[[T0:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<2x3xf32>)
-//      CHECK: %[[T1:.*]] = llvm.mlir.constant(0 : i64) : i64
-//      CHECK: %[[T2:.*]] = llvm.extractelement %[[A]][%[[T1]] : i64] : vector<2xf32>
-//      CHECK: %[[T3:.*]] = llvm.mlir.undef : vector<3xf32>
-//      CHECK: %[[T4:.*]] = llvm.mlir.constant(0 : i32) : i32
-//      CHECK: %[[T5:.*]] = llvm.insertelement %[[T2]], %[[T3]][%[[T4]] : i32] : vector<3xf32>
-//      CHECK: %[[T6:.*]] = llvm.shufflevector %[[T5]], %[[T3]] [0 : i32, 0 : i32, 0 : i32] : vector<3xf32>, vector<3xf32>
-//      CHECK: %[[T7:.*]] = llvm.extractvalue %[[C]][0] : !llvm.array<2 x vector<3xf32>>
-//      CHECK: %[[T8:.*]] = "llvm.intr.fmuladd"(%[[T6]], %[[B]], %[[T7]]) : (vector<3xf32>, vector<3xf32>, vector<3xf32>)
-//      CHECK: %[[T9:.*]] = llvm.insertvalue %[[T8]], %[[T0]][0] : !llvm.array<2 x vector<3xf32>>
-//      CHECK: %[[T10:.*]] = llvm.mlir.constant(1 : i64) : i64
-//      CHECK: %[[T11:.*]] = llvm.extractelement %[[A]][%[[T10]] : i64] : vector<2xf32>
-//      CHECK: %[[T12:.*]] = llvm.mlir.undef : vector<3xf32>
-//      CHECK: %[[T13:.*]] = llvm.mlir.constant(0 : i32) : i32
-//      CHECK: %[[T14:.*]] = llvm.insertelement %[[T11]], %[[T12]][%[[T13]] : i32] : vector<3xf32>
-//      CHECK: %[[T15:.*]] = llvm.shufflevector %[[T14]], %[[T12]] [0 : i32, 0 : i32, 0 : i32] : vector<3xf32>, vector<3xf32>
-//      CHECK: %[[T16:.*]] = llvm.extractvalue %[[C]][1] : !llvm.array<2 x vector<3xf32>>
-//      CHECK: %[[T17:.*]] = "llvm.intr.fmuladd"(%[[T15]], %[[B]], %[[T16]]) : (vector<3xf32>, vector<3xf32>, vector<3xf32>)
-//      CHECK: %[[T18:.*]] = llvm.insertvalue %[[T17]], %[[T9]][1] : !llvm.array<2 x vector<3xf32>>
-//      CHECK: llvm.return %[[T18]] : !llvm.array<2 x vector<3xf32>>
+// CHECK-LABEL: @outerproduct_add(
+// CHECK-SAME:  %[[A:.*]]: vector<2xf32>,
+// CHECK-SAME:  %[[B:.*]]: vector<3xf32>,
+// CHECK-SAME:  %[[C:.*]]: vector<2x3xf32>) -> vector<2x3xf32>
+// CHECK:       %[[T3:.*]] = constant dense<0.000000e+00> : vector<2x3xf32>
+// CHECK:       %[[T4:.*]] = llvm.mlir.constant(0 : i64) : i64
+// CHECK:       %[[T5:.*]] = llvm.extractelement %[[A]]{{\[}}%[[T4]] : i64] : vector<2xf32>
+// CHECK:       %[[T6:.*]] = splat %[[T5]] : vector<3xf32>
+// CHECK:       %[[T7:.*]] = llvm.mlir.cast %[[C]] : vector<2x3xf32> to !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T8:.*]] = llvm.extractvalue %[[T7]][0] : !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T9:.*]] = "llvm.intr.fmuladd"(%[[T6]], %[[B]], %[[T8]]) : (vector<3xf32>, vector<3xf32>, vector<3xf32>) -> vector<3xf32>
+// CHECK:       %[[T10:.*]] = llvm.mlir.cast %[[T3]] : vector<2x3xf32> to !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T11:.*]] = llvm.insertvalue %[[T9]], %[[T10]][0] : !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T12:.*]] = llvm.mlir.constant(1 : i64) : i64
+// CHECK:       %[[T13:.*]] = llvm.extractelement %[[A]]{{\[}}%[[T12]] : i64] : vector<2xf32>
+// CHECK:       %[[T14:.*]] = splat %[[T13]] : vector<3xf32>
+// CHECK:       %[[T15:.*]] = llvm.mlir.cast %[[C]] : vector<2x3xf32> to !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T16:.*]] = llvm.extractvalue %[[T15]][1] : !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T17:.*]] = "llvm.intr.fmuladd"(%[[T14]], %[[B]], %[[T16]]) : (vector<3xf32>, vector<3xf32>, vector<3xf32>) -> vector<3xf32>
+// CHECK:       %[[T18:.*]] = llvm.insertvalue %[[T17]], %[[T11]][1] : !llvm.array<2 x vector<3xf32>>
+// CHECK:       %[[T19:.*]] = llvm.mlir.cast %[[T18]] : !llvm.array<2 x vector<3xf32>> to vector<2x3xf32>
+// CHECK:       return %[[T19]] : vector<2x3xf32>
 
 // -----
 
 func @shuffle_1D_direct(%arg0: vector<2xf32>, %arg1: vector<2xf32>) -> vector<2xf32> {
   %1 = vector.shuffle %arg0, %arg1 [0, 1] : vector<2xf32>, vector<2xf32>
   return %1 : vector<2xf32>
 }
-// CHECK-LABEL: llvm.func @shuffle_1D_direct(
+// CHECK-LABEL: @shuffle_1D_direct(
 // CHECK-SAME: %[[A:.*]]: vector<2xf32>,
 // CHECK-SAME: %[[B:.*]]: vector<2xf32>)
 //       CHECK:   %[[s:.*]] = llvm.shufflevector %[[A]], %[[B]] [0, 1] : vector<2xf32>, vector<2xf32>
-//       CHECK:   llvm.return %[[s]] : vector<2xf32>
+//       CHECK:   return %[[s]] : vector<2xf32>
 
 // -----
 
 func @shuffle_1D(%arg0: vector<2xf32>, %arg1: vector<3xf32>) -> vector<5xf32> {
   %1 = vector.shuffle %arg0, %arg1 [4, 3, 2, 1, 0] : vector<2xf32>, vector<3xf32>
   return %1 : vector<5xf32>
 }
-// CHECK-LABEL: llvm.func @shuffle_1D(
+// CHECK-LABEL: @shuffle_1D(
 // CHECK-SAME: %[[A:.*]]: vector<2xf32>,
 // CHECK-SAME: %[[B:.*]]: vector<3xf32>)
 //       CHECK:   %[[u0:.*]] = llvm.mlir.undef : vector<5xf32>
 //       CHECK:   %[[c2:.*]] = llvm.mlir.constant(2 : index) : i64
 //       CHECK:   %[[e1:.*]] = llvm.extractelement %[[B]][%[[c2]] : i64] : vector<3xf32>
 //       CHECK:   %[[c0:.*]] = llvm.mlir.constant(0 : index) : i64
 //       CHECK:   %[[i1:.*]] = llvm.insertelement %[[e1]], %[[u0]][%[[c0]] : i64] : vector<5xf32>
 //       CHECK:   %[[c1:.*]] = llvm.mlir.constant(1 : index) : i64
 //       CHECK:   %[[e2:.*]] = llvm.extractelement %[[B]][%[[c1]] : i64] : vector<3xf32>
 //       CHECK:   %[[c1:.*]] = llvm.mlir.constant(1 : index) : i64
 //       CHECK:   %[[i2:.*]] = llvm.insertelement %[[e2]], %[[i1]][%[[c1]] : i64] : vector<5xf32>
 //       CHECK:   %[[c0:.*]] = llvm.mlir.constant(0 : index) : i64
 //       CHECK:   %[[e3:.*]] = llvm.extractelement %[[B]][%[[c0]] : i64] : vector<3xf32>
 //       CHECK:   %[[c2:.*]] = llvm.mlir.constant(2 : index) : i64
 //       CHECK:   %[[i3:.*]] = llvm.insertelement %[[e3]], %[[i2]][%[[c2]] : i64] : vector<5xf32>
 //       CHECK:   %[[c1:.*]] = llvm.mlir.constant(1 : index) : i64
 //       CHECK:   %[[e4:.*]] = llvm.extractelement %[[A]][%[[c1]] : i64] : vector<2xf32>
 //       CHECK:   %[[c3:.*]] = llvm.mlir.constant(3 : index) : i64
 //       CHECK:   %[[i4:.*]] = llvm.insertelement %[[e4]], %[[i3]][%[[c3]] : i64] : vector<5xf32>
 //       CHECK:   %[[c0:.*]] = llvm.mlir.constant(0 : index) : i64
 //       CHECK:   %[[e5:.*]] = llvm.extractelement %[[A]][%[[c0]] : i64] : vector<2xf32>
 //       CHECK:   %[[c4:.*]] = llvm.mlir.constant(4 : index) : i64
 //       CHECK:   %[[i5:.*]] = llvm.insertelement %[[e5]], %[[i4]][%[[c4]] : i64] : vector<5xf32>
-//       CHECK:   llvm.return %[[i5]] : vector<5xf32>
+//       CHECK:   return %[[i5]] : vector<5xf32>
 
 // -----
 
 func @shuffle_2D(%a: vector<1x4xf32>, %b: vector<2x4xf32>) -> vector<3x4xf32> {
   %1 = vector.shuffle %a, %b[1, 0, 2] : vector<1x4xf32>, vector<2x4xf32>
   return %1 : vector<3x4xf32>
 }
-// CHECK-LABEL: llvm.func @shuffle_2D(
-// CHECK-SAME: %[[A:.*]]: !llvm.array<1 x vector<4xf32>>,
-// CHECK-SAME: %[[B:.*]]: !llvm.array<2 x vector<4xf32>>)
+// CHECK-LABEL: @shuffle_2D(
+// CHECK-SAME: %[[A:.*]]: vector<1x4xf32>,
+// CHECK-SAME: %[[B:.*]]: vector<2x4xf32>)
+//       CHECK:   %[[VAL_0:.*]] = llvm.mlir.cast %[[A]] : vector<1x4xf32> to !llvm.array<1 x vector<4xf32>>
+//       CHECK:   %[[VAL_1:.*]] = llvm.mlir.cast %[[B]] : vector<2x4xf32> to !llvm.array<2 x vector<4xf32>>
 //       CHECK:   %[[u0:.*]] = llvm.mlir.undef : !llvm.array<3 x vector<4xf32>>
-//       CHECK:   %[[e1:.*]] = llvm.extractvalue %[[B]][0] : !llvm.array<2 x vector<4xf32>>
+//       CHECK:   %[[e1:.*]] = llvm.extractvalue %[[VAL_1]][0] : !llvm.array<2 x vector<4xf32>>
 //       CHECK:   %[[i1:.*]] = llvm.insertvalue %[[e1]], %[[u0]][0] : !llvm.array<3 x vector<4xf32>>
-//       CHECK:   %[[e2:.*]] = llvm.extractvalue %[[A]][0] : !llvm.array<1 x vector<4xf32>>
+//       CHECK:   %[[e2:.*]] = llvm.extractvalue %[[VAL_0]][0] : !llvm.array<1 x vector<4xf32>>
 //       CHECK:   %[[i2:.*]] = llvm.insertvalue %[[e2]], %[[i1]][1] : !llvm.array<3 x vector<4xf32>>
-//       CHECK:   %[[e3:.*]] = llvm.extractvalue %[[B]][1] : !llvm.array<2 x vector<4xf32>>
+//       CHECK:   %[[e3:.*]] = llvm.extractvalue %[[VAL_1]][1] : !llvm.array<2 x vector<4xf32>>
 //       CHECK:   %[[i3:.*]] = llvm.insertvalue %[[e3]], %[[i2]][2] : !llvm.array<3 x vector<4xf32>>
-//       CHECK:   llvm.return %[[i3]] : !llvm.array<3 x vector<4xf32>>
+//       CHECK:   %[[VAL_3:.*]] = llvm.mlir.cast %[[i3]] : !llvm.array<3 x vector<4xf32>> to vector<3x4xf32>
+//       CHECK:   return %[[VAL_3]] : vector<3x4xf32>
 
 // -----
 
 func @extract_element(%arg0: vector<16xf32>) -> f32 {
   %0 = constant 15 : i32
   %1 = vector.extractelement %arg0[%0 : i32]: vector<16xf32>
   return %1 : f32
 }
-// CHECK-LABEL: llvm.func @extract_element(
+// CHECK-LABEL: @extract_element(
 // CHECK-SAME: %[[A:.*]]: vector<16xf32>)
-//       CHECK:   %[[c:.*]] = llvm.mlir.constant(15 : i32) : i32
+//       CHECK:   %[[c:.*]] = constant 15 : i32
 //       CHECK:   %[[x:.*]] = llvm.extractelement %[[A]][%[[c]] : i32] : vector<16xf32>
-//       CHECK:   llvm.return %[[x]] : f32
+//       CHECK:   return %[[x]] : f32
 
 // -----
 
 func @extract_element_from_vec_1d(%arg0: vector<16xf32>) -> f32 {
   %0 = vector.extract %arg0[15]: vector<16xf32>
   return %0 : f32
 }
-// CHECK-LABEL: llvm.func @extract_element_from_vec_1d
+// CHECK-LABEL: @extract_element_from_vec_1d
 //       CHECK:   llvm.mlir.constant(15 : i64) : i64
 //       CHECK:   llvm.extractelement {{.*}}[{{.*}} : i64] : vector<16xf32>
-//       CHECK:   llvm.return {{.*}} : f32
+//       CHECK:   return {{.*}} : f32
 
 // -----
 
 func @extract_vec_2d_from_vec_3d(%arg0: vector<4x3x16xf32>) -> vector<3x16xf32> {
   %0 = vector.extract %arg0[0]: vector<4x3x16xf32>
   return %0 : vector<3x16xf32>
 }
-// CHECK-LABEL: llvm.func @extract_vec_2d_from_vec_3d
+// CHECK-LABEL: @extract_vec_2d_from_vec_3d
 //       CHECK:   llvm.extractvalue {{.*}}[0] : !llvm.array<4 x array<3 x vector<16xf32>>>
-//       CHECK:   llvm.return {{.*}} : !llvm.array<3 x vector<16xf32>>
+//       CHECK:   return {{.*}} : vector<3x16xf32>
 
 // -----
 
 func @extract_vec_1d_from_vec_3d(%arg0: vector<4x3x16xf32>) -> vector<16xf32> {
   %0 = vector.extract %arg0[0, 0]: vector<4x3x16xf32>
   return %0 : vector<16xf32>
 }
-// CHECK-LABEL: llvm.func @extract_vec_1d_from_vec_3d
+// CHECK-LABEL: @extract_vec_1d_from_vec_3d
 //       CHECK:   llvm.extractvalue {{.*}}[0, 0] : !llvm.array<4 x array<3 x vector<16xf32>>>
-//       CHECK:   llvm.return {{.*}} : vector<16xf32>
+//       CHECK:   return {{.*}} : vector<16xf32>
 
 // -----
 
 func @extract_element_from_vec_3d(%arg0: vector<4x3x16xf32>) -> f32 {
   %0 = vector.extract %arg0[0, 0, 0]: vector<4x3x16xf32>
   return %0 : f32
 }
-// CHECK-LABEL: llvm.func @extract_element_from_vec_3d
+// CHECK-LABEL: @extract_element_from_vec_3d
 //       CHECK:   llvm.extractvalue {{.*}}[0, 0] : !llvm.array<4 x array<3 x vector<16xf32>>>
 //       CHECK:   llvm.mlir.constant(0 : i64) : i64
 //       CHECK:   llvm.extractelement {{.*}}[{{.*}} : i64] : vector<16xf32>
-//       CHECK:   llvm.return {{.*}} : f32
+//       CHECK:   return {{.*}} : f32
 
 // -----
 
 func @insert_element(%arg0: f32, %arg1: vector<4xf32>) -> vector<4xf32> {
   %0 = constant 3 : i32
   %1 = vector.insertelement %arg0, %arg1[%0 : i32] : vector<4xf32>
   return %1 : vector<4xf32>
 }
-// CHECK-LABEL: llvm.func @insert_element(
+// CHECK-LABEL: @insert_element(
 // CHECK-SAME: %[[A:.*]]: f32,
 // CHECK-SAME: %[[B:.*]]: vector<4xf32>)
-//       CHECK:   %[[c:.*]] = llvm.mlir.constant(3 : i32) : i32
+//       CHECK:   %[[c:.*]] = constant 3 : i32
 //       CHECK:   %[[x:.*]] = llvm.insertelement %[[A]], %[[B]][%[[c]] : i32] : vector<4xf32>
-//       CHECK:   llvm.return %[[x]] : vector<4xf32>
+//       CHECK:   return %[[x]] : vector<4xf32>
 
 // -----
 
 func @insert_element_into_vec_1d(%arg0: f32, %arg1: vector<4xf32>) -> vector<4xf32> {
   %0 = vector.insert %arg0, %arg1[3] : f32 into vector<4xf32>
   return %0 : vector<4xf32>
 }
-// CHECK-LABEL: llvm.func @insert_element_into_vec_1d
+// CHECK-LABEL: @insert_element_into_vec_1d
 //       CHECK:   llvm.mlir.constant(3 : i64) : i64
 //       CHECK:   llvm.insertelement {{.*}}, {{.*}}[{{.*}} : i64] : vector<4xf32>
-//       CHECK:   llvm.return {{.*}} : vector<4xf32>
+//       CHECK:   return {{.*}} : vector<4xf32>
 
 // -----
 
 func @insert_vec_2d_into_vec_3d(%arg0: vector<8x16xf32>, %arg1: vector<4x8x16xf32>) -> vector<4x8x16xf32> {
   %0 = vector.insert %arg0, %arg1[3] : vector<8x16xf32> into vector<4x8x16xf32>
   return %0 : vector<4x8x16xf32>
 }
-// CHECK-LABEL: llvm.func @insert_vec_2d_into_vec_3d
+// CHECK-LABEL: @insert_vec_2d_into_vec_3d
 //       CHECK:   llvm.insertvalue {{.*}}, {{.*}}[3] : !llvm.array<4 x array<8 x vector<16xf32>>>
-//       CHECK:   llvm.return {{.*}} : !llvm.array<4 x array<8 x vector<16xf32>>>
+//       CHECK:   return {{.*}} : vector<4x8x16xf32>
 
 // -----
 
 func @insert_vec_1d_into_vec_3d(%arg0: vector<16xf32>, %arg1: vector<4x8x16xf32>) -> vector<4x8x16xf32> {
   %0 = vector.insert %arg0, %arg1[3, 7] : vector<16xf32> into vector<4x8x16xf32>
   return %0 : vector<4x8x16xf32>
 }
-// CHECK-LABEL: llvm.func @insert_vec_1d_into_vec_3d
+// CHECK-LABEL: @insert_vec_1d_into_vec_3d
 //       CHECK:   llvm.insertvalue {{.*}}, {{.*}}[3, 7] : !llvm.array<4 x array<8 x vector<16xf32>>>
-//       CHECK:   llvm.return {{.*}} : !llvm.array<4 x array<8 x vector<16xf32>>>
+//       CHECK:   return {{.*}} : vector<4x8x16xf32>
 
 // -----
 
 func @insert_element_into_vec_3d(%arg0: f32, %arg1: vector<4x8x16xf32>) -> vector<4x8x16xf32> {
   %0 = vector.insert %arg0, %arg1[3, 7, 15] : f32 into vector<4x8x16xf32>
   return %0 : vector<4x8x16xf32>
 }
-// CHECK-LABEL: llvm.func @insert_element_into_vec_3d
+// CHECK-LABEL: @insert_element_into_vec_3d
 //       CHECK:   llvm.extractvalue {{.*}}[3, 7] : !llvm.array<4 x array<8 x vector<16xf32>>>
 //       CHECK:   llvm.mlir.constant(15 : i64) : i64
 //       CHECK:   llvm.insertelement {{.*}}, {{.*}}[{{.*}} : i64] : vector<16xf32>
 //       CHECK:   llvm.insertvalue {{.*}}, {{.*}}[3, 7] : !llvm.array<4 x array<8 x vector<16xf32>>>
-//       CHECK:   llvm.return {{.*}} : !llvm.array<4 x array<8 x vector<16xf32>>>
+//       CHECK:   return {{.*}} : vector<4x8x16xf32>
 
 // -----
 
 func @vector_type_cast(%arg0: memref<8x8x8xf32>) -> memref<vector<8x8x8xf32>> {
   %0 = vector.type_cast %arg0: memref<8x8x8xf32> to memref<vector<8x8x8xf32>>
   return %0 : memref<vector<8x8x8xf32>>
 }
-// CHECK-LABEL: llvm.func @vector_type_cast
+// CHECK-LABEL: @vector_type_cast
 //       CHECK:   llvm.mlir.undef : !llvm.struct<(ptr<array<8 x array<8 x vector<8xf32>>>>, ptr<array<8 x array<8 x vector<8xf32>>>>, i64)>
 //       CHECK:   %[[allocated:.*]] = llvm.extractvalue {{.*}}[0] : !llvm.struct<(ptr<f32>, ptr<f32>, i64, array<3 x i64>, array<3 x i64>)>
 //       CHECK:   %[[allocatedBit:.*]] = llvm.bitcast %[[allocated]] : !llvm.ptr<f32> to !llvm.ptr<array<8 x array<8 x vector<8xf32>>>>
 //       CHECK:   llvm.insertvalue %[[allocatedBit]], {{.*}}[0] : !llvm.struct<(ptr<array<8 x array<8 x vector<8xf32>>>>, ptr<array<8 x array<8 x vector<8xf32>>>>, i64)>
 //       CHECK:   %[[aligned:.*]] = llvm.extractvalue {{.*}}[1] : !llvm.struct<(ptr<f32>, ptr<f32>, i64, array<3 x i64>, array<3 x i64>)>
 //       CHECK:   %[[alignedBit:.*]] = llvm.bitcast %[[aligned]] : !llvm.ptr<f32> to !llvm.ptr<array<8 x array<8 x vector<8xf32>>>>
 //       CHECK:   llvm.insertvalue %[[alignedBit]], {{.*}}[1] : !llvm.struct<(ptr<array<8 x array<8 x vector<8xf32>>>>, ptr<array<8 x array<8 x vector<8xf32>>>>, i64)>
 //       CHECK:   llvm.mlir.constant(0 : index
 //       CHECK:   llvm.insertvalue {{.*}}[2] : !llvm.struct<(ptr<array<8 x array<8 x vector<8xf32>>>>, ptr<array<8 x array<8 x vector<8xf32>>>>, i64)>
 
 // -----
 
 func @vector_type_cast_non_zero_addrspace(%arg0: memref<8x8x8xf32, 3>) -> memref<vector<8x8x8xf32>, 3> {
   %0 = vector.type_cast %arg0: memref<8x8x8xf32, 3> to memref<vector<8x8x8xf32>, 3>
   return %0 : memref<vector<8x8x8xf32>, 3>
 }
-// CHECK-LABEL: llvm.func @vector_type_cast_non_zero_addrspace
+// CHECK-LABEL: @vector_type_cast_non_zero_addrspace
 //       CHECK:   llvm.mlir.undef : !llvm.struct<(ptr<array<8 x array<8 x vector<8xf32>>>, 3>, ptr<array<8 x array<8 x vector<8xf32>>>, 3>, i64)>
 //       CHECK:   %[[allocated:.*]] = llvm.extractvalue {{.*}}[0] : !llvm.struct<(ptr<f32, 3>, ptr<f32, 3>, i64, array<3 x i64>, array<3 x i64>)>
 //       CHECK:   %[[allocatedBit:.*]] = llvm.bitcast %[[allocated]] : !llvm.ptr<f32, 3> to !llvm.ptr<array<8 x array<8 x vector<8xf32>>>, 3>
 //       CHECK:   llvm.insertvalue %[[allocatedBit]], {{.*}}[0] : !llvm.struct<(ptr<array<8 x array<8 x vector<8xf32>>>, 3>, ptr<array<8 x array<8 x vector<8xf32>>>, 3>, i64)>
 //       CHECK:   %[[aligned:.*]] = llvm.extractvalue {{.*}}[1] : !llvm.struct<(ptr<f32, 3>, ptr<f32, 3>, i64, array<3 x i64>, array<3 x i64>)>
 //       CHECK:   %[[alignedBit:.*]] = llvm.bitcast %[[aligned]] : !llvm.ptr<f32, 3> to !llvm.ptr<array<8 x array<8 x vector<8xf32>>>, 3>
 //       CHECK:   llvm.insertvalue %[[alignedBit]], {{.*}}[1] : !llvm.struct<(ptr<array<8 x array<8 x vector<8xf32>>>, 3>, ptr<array<8 x array<8 x vector<8xf32>>>, 3>, i64)>
 //       CHECK:   llvm.mlir.constant(0 : index
 //       CHECK:   llvm.insertvalue {{.*}}[2] : !llvm.struct<(ptr<array<8 x array<8 x vector<8xf32>>>, 3>, ptr<array<8 x array<8 x vector<8xf32>>>, 3>, i64)>
 
 // -----
 
 func @vector_print_scalar_i1(%arg0: i1) {
   vector.print %arg0 : i1
   return
 }
 //
 // Type "boolean" always uses zero extension.
 //
-// CHECK-LABEL: llvm.func @vector_print_scalar_i1(
+// CHECK-LABEL: @vector_print_scalar_i1(
 // CHECK-SAME: %[[A:.*]]: i1)
-//       CHECK: %[[S:.*]] = llvm.zext %[[A]] : i1 to i64
+//       CHECK: %[[S:.*]] = zexti %[[A]] : i1 to i64
 //       CHECK: llvm.call @printI64(%[[S]]) : (i64) -> ()
 //       CHECK: llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_i4(%arg0: i4) {
   vector.print %arg0 : i4
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_i4(
+// CHECK-LABEL: @vector_print_scalar_i4(
 // CHECK-SAME: %[[A:.*]]: i4)
-//       CHECK: %[[S:.*]] = llvm.sext %[[A]] : i4 to i64
+//       CHECK: %[[S:.*]] = sexti %[[A]] : i4 to i64
 //       CHECK: llvm.call @printI64(%[[S]]) : (i64) -> ()
 //       CHECK: llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_si4(%arg0: si4) {
   vector.print %arg0 : si4
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_si4(
-// CHECK-SAME: %[[A:.*]]: i4)
-//       CHECK: %[[S:.*]] = llvm.sext %[[A]] : i4 to i64
+// CHECK-LABEL: @vector_print_scalar_si4(
+// CHECK-SAME: %[[A:.*]]: si4)
+//       CHECK: %[[C:.*]] = llvm.mlir.cast %[[A]] : si4 to i4
+//       CHECK: %[[S:.*]] = sexti %[[C]] : i4 to i64
 //       CHECK: llvm.call @printI64(%[[S]]) : (i64) -> ()
 //       CHECK: llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_ui4(%arg0: ui4) {
   vector.print %arg0 : ui4
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_ui4(
-// CHECK-SAME: %[[A:.*]]: i4)
-//       CHECK: %[[S:.*]] = llvm.zext %[[A]] : i4 to i64
+// CHECK-LABEL: @vector_print_scalar_ui4(
+// CHECK-SAME: %[[A:.*]]: ui4)
+//       CHECK: %[[C:.*]] = llvm.mlir.cast %[[A]] : ui4 to i4
+//       CHECK: %[[S:.*]] = zexti %[[C]] : i4 to i64
 //       CHECK: llvm.call @printU64(%[[S]]) : (i64) -> ()
 //       CHECK: llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_i32(%arg0: i32) {
   vector.print %arg0 : i32
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_i32(
+// CHECK-LABEL: @vector_print_scalar_i32(
 // CHECK-SAME: %[[A:.*]]: i32)
-//       CHECK: %[[S:.*]] = llvm.sext %[[A]] : i32 to i64
+//       CHECK: %[[S:.*]] = sexti %[[A]] : i32 to i64
 //       CHECK: llvm.call @printI64(%[[S]]) : (i64) -> ()
 //       CHECK: llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_ui32(%arg0: ui32) {
   vector.print %arg0 : ui32
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_ui32(
-// CHECK-SAME: %[[A:.*]]: i32)
-//       CHECK: %[[S:.*]] = llvm.zext %[[A]] : i32 to i64
+// CHECK-LABEL: @vector_print_scalar_ui32(
+// CHECK-SAME: %[[A:.*]]: ui32)
+//       CHECK: %[[C:.*]] = llvm.mlir.cast %[[A]] : ui32 to i32
+//       CHECK: %[[S:.*]] = zexti %[[C]] : i32 to i64
 //       CHECK: llvm.call @printU64(%[[S]]) : (i64) -> ()
 
 // -----
 
 func @vector_print_scalar_i40(%arg0: i40) {
   vector.print %arg0 : i40
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_i40(
+// CHECK-LABEL: @vector_print_scalar_i40(
 // CHECK-SAME: %[[A:.*]]: i40)
-//       CHECK: %[[S:.*]] = llvm.sext %[[A]] : i40 to i64
+//       CHECK: %[[S:.*]] = sexti %[[A]] : i40 to i64
 //       CHECK: llvm.call @printI64(%[[S]]) : (i64) -> ()
 //       CHECK: llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_si40(%arg0: si40) {
   vector.print %arg0 : si40
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_si40(
-// CHECK-SAME: %[[A:.*]]: i40)
-//       CHECK: %[[S:.*]] = llvm.sext %[[A]] : i40 to i64
+// CHECK-LABEL: @vector_print_scalar_si40(
+// CHECK-SAME: %[[A:.*]]: si40)
+//       CHECK: %[[C:.*]] = llvm.mlir.cast %[[A]] : si40 to i40
+//       CHECK: %[[S:.*]] = sexti %[[C]] : i40 to i64
 //       CHECK: llvm.call @printI64(%[[S]]) : (i64) -> ()
 //       CHECK: llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_ui40(%arg0: ui40) {
   vector.print %arg0 : ui40
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_ui40(
-// CHECK-SAME: %[[A:.*]]: i40)
-//       CHECK: %[[S:.*]] = llvm.zext %[[A]] : i40 to i64
+// CHECK-LABEL: @vector_print_scalar_ui40(
+// CHECK-SAME: %[[A:.*]]: ui40)
+//       CHECK: %[[C:.*]] = llvm.mlir.cast %[[A]] : ui40 to i40
+//       CHECK: %[[S:.*]] = zexti %[[C]] : i40 to i64
 //       CHECK: llvm.call @printU64(%[[S]]) : (i64) -> ()
 //       CHECK: llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_i64(%arg0: i64) {
   vector.print %arg0 : i64
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_i64(
+// CHECK-LABEL: @vector_print_scalar_i64(
 // CHECK-SAME: %[[A:.*]]: i64)
 //       CHECK:    llvm.call @printI64(%[[A]]) : (i64) -> ()
 //       CHECK:    llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_ui64(%arg0: ui64) {
   vector.print %arg0 : ui64
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_ui64(
-// CHECK-SAME: %[[A:.*]]: i64)
-//       CHECK:    llvm.call @printU64(%[[A]]) : (i64) -> ()
+// CHECK-LABEL: @vector_print_scalar_ui64(
+// CHECK-SAME: %[[A:.*]]: ui64)
+//       CHECK:    %[[C:.*]] = llvm.mlir.cast %[[A]] : ui64 to i64
+//       CHECK:    llvm.call @printU64(%[[C]]) : (i64) -> ()
 //       CHECK:    llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_index(%arg0: index) {
   vector.print %arg0 : index
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_index(
-// CHECK-SAME: %[[A:.*]]: i64)
-//       CHECK:    llvm.call @printU64(%[[A]]) : (i64) -> ()
+// CHECK-LABEL: @vector_print_scalar_index(
+// CHECK-SAME: %[[A:.*]]: index)
+//       CHECK:    %[[C:.*]] = llvm.mlir.cast %[[A]] : index to i64
+//       CHECK:    llvm.call @printU64(%[[C]]) : (i64) -> ()
 //       CHECK:    llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_f32(%arg0: f32) {
   vector.print %arg0 : f32
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_f32(
+// CHECK-LABEL: @vector_print_scalar_f32(
 // CHECK-SAME: %[[A:.*]]: f32)
 //       CHECK:    llvm.call @printF32(%[[A]]) : (f32) -> ()
 //       CHECK:    llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_scalar_f64(%arg0: f64) {
   vector.print %arg0 : f64
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_scalar_f64(
+// CHECK-LABEL: @vector_print_scalar_f64(
 // CHECK-SAME: %[[A:.*]]: f64)
 //       CHECK:    llvm.call @printF64(%[[A]]) : (f64) -> ()
 //       CHECK:    llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @vector_print_vector(%arg0: vector<2x2xf32>) {
   vector.print %arg0 : vector<2x2xf32>
   return
 }
-// CHECK-LABEL: llvm.func @vector_print_vector(
-// CHECK-SAME: %[[A:.*]]: !llvm.array<2 x vector<2xf32>>)
+// CHECK-LABEL: @vector_print_vector(
+// CHECK-SAME: %[[A:.*]]: vector<2x2xf32>)
+//       CHECK:    %[[VAL_1:.*]] = llvm.mlir.cast %[[A]] : vector<2x2xf32> to !llvm.array<2 x vector<2xf32>>
 //       CHECK:    llvm.call @printOpen() : () -> ()
-//       CHECK:    %[[x0:.*]] = llvm.extractvalue %[[A]][0] : !llvm.array<2 x vector<2xf32>>
+//       CHECK:    %[[x0:.*]] = llvm.extractvalue %[[VAL_1]][0] : !llvm.array<2 x vector<2xf32>>
 //       CHECK:    llvm.call @printOpen() : () -> ()
 //       CHECK:    %[[x1:.*]] = llvm.mlir.constant(0 : index) : i64
 //       CHECK:    %[[x2:.*]] = llvm.extractelement %[[x0]][%[[x1]] : i64] : vector<2xf32>
 //       CHECK:    llvm.call @printF32(%[[x2]]) : (f32) -> ()
 //       CHECK:    llvm.call @printComma() : () -> ()
 //       CHECK:    %[[x3:.*]] = llvm.mlir.constant(1 : index) : i64
 //       CHECK:    %[[x4:.*]] = llvm.extractelement %[[x0]][%[[x3]] : i64] : vector<2xf32>
 //       CHECK:    llvm.call @printF32(%[[x4]]) : (f32) -> ()
 //       CHECK:    llvm.call @printClose() : () -> ()
 //       CHECK:    llvm.call @printComma() : () -> ()
-//       CHECK:    %[[x5:.*]] = llvm.extractvalue %[[A]][1] : !llvm.array<2 x vector<2xf32>>
+//       CHECK:    %[[x5:.*]] = llvm.extractvalue %[[VAL_1]][1] : !llvm.array<2 x vector<2xf32>>
 //       CHECK:    llvm.call @printOpen() : () -> ()
 //       CHECK:    %[[x6:.*]] = llvm.mlir.constant(0 : index) : i64
 //       CHECK:    %[[x7:.*]] = llvm.extractelement %[[x5]][%[[x6]] : i64] : vector<2xf32>
 //       CHECK:    llvm.call @printF32(%[[x7]]) : (f32) -> ()
 //       CHECK:    llvm.call @printComma() : () -> ()
 //       CHECK:    %[[x8:.*]] = llvm.mlir.constant(1 : index) : i64
 //       CHECK:    %[[x9:.*]] = llvm.extractelement %[[x5]][%[[x8]] : i64] : vector<2xf32>
 //       CHECK:    llvm.call @printF32(%[[x9]]) : (f32) -> ()
 //       CHECK:    llvm.call @printClose() : () -> ()
 //       CHECK:    llvm.call @printClose() : () -> ()
 //       CHECK:    llvm.call @printNewline() : () -> ()
 
 // -----
 
 func @extract_strided_slice1(%arg0: vector<4xf32>) -> vector<2xf32> {
   %0 = vector.extract_strided_slice %arg0 {offsets = [2], sizes = [2], strides = [1]} : vector<4xf32> to vector<2xf32>
   return %0 : vector<2xf32>
 }
-// CHECK-LABEL: llvm.func @extract_strided_slice1(
+// CHECK-LABEL: @extract_strided_slice1(
 //  CHECK-SAME:    %[[A:.*]]: vector<4xf32>)
 //       CHECK:    %[[T0:.*]] = llvm.shufflevector %[[A]], %[[A]] [2, 3] : vector<4xf32>, vector<4xf32>
-//       CHECK:    llvm.return %[[T0]] : vector<2xf32>
+//       CHECK:    return %[[T0]] : vector<2xf32>
 
 // -----
 
 func @extract_strided_slice2(%arg0: vector<4x8xf32>) -> vector<2x8xf32> {
   %0 = vector.extract_strided_slice %arg0 {offsets = [2], sizes = [2], strides = [1]} : vector<4x8xf32> to vector<2x8xf32>
   return %0 : vector<2x8xf32>
 }
-// CHECK-LABEL: llvm.func @extract_strided_slice2(
-//  CHECK-SAME:    %[[A:.*]]: !llvm.array<4 x vector<8xf32>>)
+// CHECK-LABEL: @extract_strided_slice2(
+//  CHECK-SAME:    %[[ARG:.*]]: vector<4x8xf32>)
+//       CHECK:    %[[A:.*]] = llvm.mlir.cast %[[ARG]] : vector<4x8xf32> to !llvm.array<4 x vector<8xf32>>
 //       CHECK:    %[[T0:.*]] = llvm.mlir.undef : !llvm.array<2 x vector<8xf32>>
 //       CHECK:    %[[T1:.*]] = llvm.extractvalue %[[A]][2] : !llvm.array<4 x vector<8xf32>>
 //       CHECK:    %[[T2:.*]] = llvm.insertvalue %[[T1]], %[[T0]][0] : !llvm.array<2 x vector<8xf32>>
 //       CHECK:    %[[T3:.*]] = llvm.extractvalue %[[A]][3] : !llvm.array<4 x vector<8xf32>>
 //       CHECK:    %[[T4:.*]] = llvm.insertvalue %[[T3]], %[[T2]][1] : !llvm.array<2 x vector<8xf32>>
-//       CHECK:    llvm.return %[[T4]] : !llvm.array<2 x vector<8xf32>>
+//       CHECK:    %[[T5:.*]] = llvm.mlir.cast %[[T4]] : !llvm.array<2 x vector<8xf32>> to vector<2x8xf32>
+//       CHECK:    return %[[T5]]
 
 // -----
 
 func @extract_strided_slice3(%arg0: vector<4x8xf32>) -> vector<2x2xf32> {
   %0 = vector.extract_strided_slice %arg0 {offsets = [2, 2], sizes = [2, 2], strides = [1, 1]} : vector<4x8xf32> to vector<2x2xf32>
   return %0 : vector<2x2xf32>
 }
-// CHECK-LABEL: llvm.func @extract_strided_slice3(
-//  CHECK-SAME:    %[[A:.*]]: !llvm.array<4 x vector<8xf32>>)
-//       CHECK:    %[[T1:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<2x2xf32>) : !llvm.array<2 x vector<2xf32>>
+// CHECK-LABEL: @extract_strided_slice3(
+//  CHECK-SAME:    %[[ARG:.*]]: vector<4x8xf32>)
+//       CHECK:    %[[VAL_1:.*]] = constant 0.000000e+00 : f32
+//       CHECK:    %[[VAL_2:.*]] = splat %[[VAL_1]] : vector<2x2xf32>
+//       CHECK:    %[[A:.*]] = llvm.mlir.cast %[[ARG]] : vector<4x8xf32> to !llvm.array<4 x vector<8xf32>>
 //       CHECK:    %[[T2:.*]] = llvm.extractvalue %[[A]][2] : !llvm.array<4 x vector<8xf32>>
 //       CHECK:    %[[T3:.*]] = llvm.shufflevector %[[T2]], %[[T2]] [2, 3] : vector<8xf32>, vector<8xf32>
-//       CHECK:    %[[T4:.*]] = llvm.insertvalue %[[T3]], %[[T1]][0] : !llvm.array<2 x vector<2xf32>>
+//       CHECK:    %[[VAL_6:.*]] = llvm.mlir.cast %[[VAL_2]] : vector<2x2xf32> to !llvm.array<2 x vector<2xf32>>
+//       CHECK:    %[[T4:.*]] = llvm.insertvalue %[[T3]], %[[VAL_6]][0] : !llvm.array<2 x vector<2xf32>>
+//       CHECK:    %[[A:.*]] = llvm.mlir.cast %[[ARG]] : vector<4x8xf32> to !llvm.array<4 x vector<8xf32>>
 //       CHECK:    %[[T5:.*]] = llvm.extractvalue %[[A]][3] : !llvm.array<4 x vector<8xf32>>
 //       CHECK:    %[[T6:.*]] = llvm.shufflevector %[[T5]], %[[T5]] [2, 3] : vector<8xf32>, vector<8xf32>
 //       CHECK:    %[[T7:.*]] = llvm.insertvalue %[[T6]], %[[T4]][1] : !llvm.array<2 x vector<2xf32>>
-//       CHECK:    llvm.return %[[T7]] : !llvm.array<2 x vector<2xf32>>
+//       CHECK:    %[[VAL_12:.*]] = llvm.mlir.cast %[[T7]] : !llvm.array<2 x vector<2xf32>> to vector<2x2xf32>
+//       CHECK:    return %[[VAL_12]] : vector<2x2xf32>
 
 // -----
 
 func @insert_strided_slice1(%b: vector<4x4xf32>, %c: vector<4x4x4xf32>) -> vector<4x4x4xf32> {
   %0 = vector.insert_strided_slice %b, %c {offsets = [2, 0, 0], strides = [1, 1]} : vector<4x4xf32> into vector<4x4x4xf32>
   return %0 : vector<4x4x4xf32>
 }
-// CHECK-LABEL: llvm.func @insert_strided_slice1
+// CHECK-LABEL: @insert_strided_slice1
 //       CHECK:    llvm.extractvalue {{.*}}[2] : !llvm.array<4 x array<4 x vector<4xf32>>>
-//  CHECK-NEXT:    llvm.insertvalue {{.*}}, {{.*}}[2] : !llvm.array<4 x array<4 x vector<4xf32>>>
+//       CHECK:    llvm.insertvalue {{.*}}, {{.*}}[2] : !llvm.array<4 x array<4 x vector<4xf32>>>
 
 // -----
 
 func @insert_strided_slice2(%a: vector<2x2xf32>, %b: vector<4x4xf32>) -> vector<4x4xf32> {
   %0 = vector.insert_strided_slice %a, %b {offsets = [2, 2], strides = [1, 1]} : vector<2x2xf32> into vector<4x4xf32>
   return %0 : vector<4x4xf32>
 }
-// CHECK-LABEL: llvm.func @insert_strided_slice2
+// CHECK-LABEL: @insert_strided_slice2
 //
 // Subvector vector<2xf32> @0 into vector<4xf32> @2
 //       CHECK:    llvm.extractvalue {{.*}}[0] : !llvm.array<2 x vector<2xf32>>
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : vector<4x4xf32> to !llvm.array<4 x vector<4xf32>>
 //  CHECK-NEXT:    llvm.extractvalue {{.*}}[2] : !llvm.array<4 x vector<4xf32>>
 // Element @0 -> element @2
-//  CHECK-NEXT:    llvm.mlir.constant(0 : index) : i64
+//  CHECK-NEXT:    constant 0 : index
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : index to i64
 //  CHECK-NEXT:    llvm.extractelement {{.*}}[{{.*}} : i64] : vector<2xf32>
-//  CHECK-NEXT:    llvm.mlir.constant(2 : index) : i64
+//  CHECK-NEXT:    constant 2 : index
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : index to i64
 //  CHECK-NEXT:    llvm.insertelement {{.*}}, {{.*}}[{{.*}} : i64] : vector<4xf32>
 // Element @1 -> element @3
-//  CHECK-NEXT:    llvm.mlir.constant(1 : index) : i64
+//  CHECK-NEXT:    constant 1 : index
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : index to i64
 //  CHECK-NEXT:    llvm.extractelement {{.*}}[{{.*}} : i64] : vector<2xf32>
-//  CHECK-NEXT:    llvm.mlir.constant(3 : index) : i64
+//  CHECK-NEXT:    constant 3 : index
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : index to i64
 //  CHECK-NEXT:    llvm.insertelement {{.*}}, {{.*}}[{{.*}} : i64] : vector<4xf32>
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : vector<4x4xf32> to !llvm.array<4 x vector<4xf32>>
 //  CHECK-NEXT:    llvm.insertvalue {{.*}}, {{.*}}[2] : !llvm.array<4 x vector<4xf32>>
 //
 // Subvector vector<2xf32> @1 into vector<4xf32> @3
 //       CHECK:    llvm.extractvalue {{.*}}[1] : !llvm.array<2 x vector<2xf32>>
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : vector<4x4xf32> to !llvm.array<4 x vector<4xf32>>
 //  CHECK-NEXT:    llvm.extractvalue {{.*}}[3] : !llvm.array<4 x vector<4xf32>>
 // Element @0 -> element @2
-//  CHECK-NEXT:    llvm.mlir.constant(0 : index) : i64
+//  CHECK-NEXT:    constant 0 : index
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : index to i64
 //  CHECK-NEXT:    llvm.extractelement {{.*}}[{{.*}} : i64] : vector<2xf32>
-//  CHECK-NEXT:    llvm.mlir.constant(2 : index) : i64
+//  CHECK-NEXT:    constant 2 : index
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : index to i64
 //  CHECK-NEXT:    llvm.insertelement {{.*}}, {{.*}}[{{.*}} : i64] : vector<4xf32>
 // Element @1 -> element @3
-//  CHECK-NEXT:    llvm.mlir.constant(1 : index) : i64
+//  CHECK-NEXT:    constant 1 : index
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : index to i64
 //  CHECK-NEXT:    llvm.extractelement {{.*}}[{{.*}} : i64] : vector<2xf32>
-//  CHECK-NEXT:    llvm.mlir.constant(3 : index) : i64
+//  CHECK-NEXT:    constant 3 : index
+//  CHECK-NEXT:    llvm.mlir.cast %{{.*}} : index to i64
 //  CHECK-NEXT:    llvm.insertelement {{.*}}, {{.*}}[{{.*}} : i64] : vector<4xf32>
 //  CHECK-NEXT:    llvm.insertvalue {{.*}}, {{.*}}[3] : !llvm.array<4 x vector<4xf32>>
 
 // -----
 
 func @insert_strided_slice3(%arg0: vector<2x4xf32>, %arg1: vector<16x4x8xf32>) -> vector<16x4x8xf32> {
   %0 = vector.insert_strided_slice %arg0, %arg1 {offsets = [0, 0, 2], strides = [1, 1]}:
         vector<2x4xf32> into vector<16x4x8xf32>
   return %0 : vector<16x4x8xf32>
 }
-// CHECK-LABEL: llvm.func @insert_strided_slice3(
-// CHECK-SAME: %[[A:.*]]: !llvm.array<2 x vector<4xf32>>,
-// CHECK-SAME: %[[B:.*]]: !llvm.array<16 x array<4 x vector<8xf32>>>)
-//      CHECK: %[[s0:.*]] = llvm.extractvalue %[[B]][0] : !llvm.array<16 x array<4 x vector<8xf32>>>
-//      CHECK: %[[s1:.*]] = llvm.extractvalue %[[A]][0] : !llvm.array<2 x vector<4xf32>>
-//      CHECK: %[[s2:.*]] = llvm.extractvalue %[[B]][0, 0] : !llvm.array<16 x array<4 x vector<8xf32>>>
-//      CHECK: %[[s3:.*]] = llvm.mlir.constant(0 : index) : i64
-//      CHECK: %[[s4:.*]] = llvm.extractelement %[[s1]][%[[s3]] : i64] : vector<4xf32>
-//      CHECK: %[[s5:.*]] = llvm.mlir.constant(2 : index) : i64
-//      CHECK: %[[s6:.*]] = llvm.insertelement %[[s4]], %[[s2]][%[[s5]] : i64] : vector<8xf32>
-//      CHECK: %[[s7:.*]] = llvm.mlir.constant(1 : index) : i64
-//      CHECK: %[[s8:.*]] = llvm.extractelement %[[s1]][%[[s7]] : i64] : vector<4xf32>
-//      CHECK: %[[s9:.*]] = llvm.mlir.constant(3 : index) : i64
-//      CHECK: %[[s10:.*]] = llvm.insertelement %[[s8]], %[[s6]][%[[s9]] : i64] : vector<8xf32>
-//      CHECK: %[[s11:.*]] = llvm.mlir.constant(2 : index) : i64
-//      CHECK: %[[s12:.*]] = llvm.extractelement %[[s1]][%[[s11]] : i64] : vector<4xf32>
-//      CHECK: %[[s13:.*]] = llvm.mlir.constant(4 : index) : i64
-//      CHECK: %[[s14:.*]] = llvm.insertelement %[[s12]], %[[s10]][%[[s13]] : i64] : vector<8xf32>
-//      CHECK: %[[s15:.*]] = llvm.mlir.constant(3 : index) : i64
-//      CHECK: %[[s16:.*]] = llvm.extractelement %[[s1]][%[[s15]] : i64] : vector<4xf32>
-//      CHECK: %[[s17:.*]] = llvm.mlir.constant(5 : index) : i64
-//      CHECK: %[[s18:.*]] = llvm.insertelement %[[s16]], %[[s14]][%[[s17]] : i64] : vector<8xf32>
-//      CHECK: %[[s19:.*]] = llvm.insertvalue %[[s18]], %[[s0]][0] : !llvm.array<4 x vector<8xf32>>
-//      CHECK: %[[s20:.*]] = llvm.extractvalue %[[A]][1] : !llvm.array<2 x vector<4xf32>>
-//      CHECK: %[[s21:.*]] = llvm.extractvalue %[[B]][0, 1] : !llvm.array<16 x array<4 x vector<8xf32>>>
-//      CHECK: %[[s22:.*]] = llvm.mlir.constant(0 : index) : i64
-//      CHECK: %[[s23:.*]] = llvm.extractelement %[[s20]][%[[s22]] : i64] : vector<4xf32>
-//      CHECK: %[[s24:.*]] = llvm.mlir.constant(2 : index) : i64
-//      CHECK: %[[s25:.*]] = llvm.insertelement %[[s23]], %[[s21]][%[[s24]] : i64] : vector<8xf32>
-//      CHECK: %[[s26:.*]] = llvm.mlir.constant(1 : index) : i64
-//      CHECK: %[[s27:.*]] = llvm.extractelement %[[s20]][%[[s26]] : i64] : vector<4xf32>
-//      CHECK: %[[s28:.*]] = llvm.mlir.constant(3 : index) : i64
-//      CHECK: %[[s29:.*]] = llvm.insertelement %[[s27]], %[[s25]][%[[s28]] : i64] : vector<8xf32>
-//      CHECK: %[[s30:.*]] = llvm.mlir.constant(2 : index) : i64
-//      CHECK: %[[s31:.*]] = llvm.extractelement %[[s20]][%[[s30]] : i64] : vector<4xf32>
-//      CHECK: %[[s32:.*]] = llvm.mlir.constant(4 : index) : i64
-//      CHECK: %[[s33:.*]] = llvm.insertelement %[[s31]], %[[s29]][%[[s32]] : i64] : vector<8xf32>
-//      CHECK: %[[s34:.*]] = llvm.mlir.constant(3 : index) : i64
-//      CHECK: %[[s35:.*]] = llvm.extractelement %[[s20]][%[[s34]] : i64] : vector<4xf32>
-//      CHECK: %[[s36:.*]] = llvm.mlir.constant(5 : index) : i64
-//      CHECK: %[[s37:.*]] = llvm.insertelement %[[s35]], %[[s33]][%[[s36]] : i64] : vector<8xf32>
-//      CHECK: %[[s38:.*]] = llvm.insertvalue %[[s37]], %[[s19]][1] : !llvm.array<4 x vector<8xf32>>
-//      CHECK: %[[s39:.*]] = llvm.insertvalue %[[s38]], %[[B]][0] : !llvm.array<16 x array<4 x vector<8xf32>>>
-//      CHECK: llvm.return %[[s39]] : !llvm.array<16 x array<4 x vector<8xf32>>>
+// CHECK-LABEL: @insert_strided_slice3(
+// CHECK-SAME: %[[A:.*]]: vector<2x4xf32>,
+// CHECK-SAME: %[[B:.*]]: vector<16x4x8xf32>)
+//      CHECK: %[[s2:.*]] = llvm.mlir.cast %[[B]] : vector<16x4x8xf32> to !llvm.array<16 x array<4 x vector<8xf32>>>
+//      CHECK: %[[s3:.*]] = llvm.extractvalue %[[s2]][0] : !llvm.array<16 x array<4 x vector<8xf32>>>
+//      CHECK: %[[s4:.*]] = llvm.mlir.cast %[[A]] : vector<2x4xf32> to !llvm.array<2 x vector<4xf32>>
+//      CHECK: %[[s5:.*]] = llvm.extractvalue %[[s4]][0] : !llvm.array<2 x vector<4xf32>>
+//      CHECK: %[[s6:.*]] = llvm.mlir.cast %[[B]] : vector<16x4x8xf32> to !llvm.array<16 x array<4 x vector<8xf32>>>
+//      CHECK: %[[s7:.*]] = llvm.extractvalue %[[s6]][0, 0] : !llvm.array<16 x array<4 x vector<8xf32>>>
+//      CHECK: %[[s8:.*]] = constant 0 : index
+//      CHECK: %[[s9:.*]] = llvm.mlir.cast %[[s8]] : index to i64
+//      CHECK: %[[s10:.*]] = llvm.extractelement %[[s5]]{{\[}}%[[s9]] : i64] : vector<4xf32>
+//      CHECK: %[[s11:.*]] = constant 2 : index
+//      CHECK: %[[s12:.*]] = llvm.mlir.cast %[[s11]] : index to i64
+//      CHECK: %[[s13:.*]] = llvm.insertelement %[[s10]], %[[s7]]{{\[}}%[[s12]] : i64] : vector<8xf32>
+//      CHECK: %[[s14:.*]] = constant 1 : index
+//      CHECK: %[[s15:.*]] = llvm.mlir.cast %[[s14]] : index to i64
+//      CHECK: %[[s16:.*]] = llvm.extractelement %[[s5]]{{\[}}%[[s15]] : i64] : vector<4xf32>
+//      CHECK: %[[s17:.*]] = constant 3 : index
+//      CHECK: %[[s18:.*]] = llvm.mlir.cast %[[s17]] : index to i64
+//      CHECK: %[[s19:.*]] = llvm.insertelement %[[s16]], %[[s13]]{{\[}}%[[s18]] : i64] : vector<8xf32>
+//      CHECK: %[[s20:.*]] = constant 2 : index
+//      CHECK: %[[s21:.*]] = llvm.mlir.cast %[[s20]] : index to i64
+//      CHECK: %[[s22:.*]] = llvm.extractelement %[[s5]]{{\[}}%[[s21]] : i64] : vector<4xf32>
+//      CHECK: %[[s23:.*]] = constant 4 : index
+//      CHECK: %[[s24:.*]] = llvm.mlir.cast %[[s23]] : index to i64
+//      CHECK: %[[s25:.*]] = llvm.insertelement %[[s22]], %[[s19]]{{\[}}%[[s24]] : i64] : vector<8xf32>
+//      CHECK: %[[s26:.*]] = constant 3 : index
+//      CHECK: %[[s27:.*]] = llvm.mlir.cast %[[s26]] : index to i64
+//      CHECK: %[[s28:.*]] = llvm.extractelement %[[s5]]{{\[}}%[[s27]] : i64] : vector<4xf32>
+//      CHECK: %[[s29:.*]] = constant 5 : index
+//      CHECK: %[[s30:.*]] = llvm.mlir.cast %[[s29]] : index to i64
+//      CHECK: %[[s31:.*]] = llvm.insertelement %[[s28]], %[[s25]]{{\[}}%[[s30]] : i64] : vector<8xf32>
+//      CHECK: %[[s32:.*]] = llvm.insertvalue %[[s31]], %[[s3]][0] : !llvm.array<4 x vector<8xf32>>
+//      CHECK: %[[s33:.*]] = llvm.mlir.cast %[[A]] : vector<2x4xf32> to !llvm.array<2 x vector<4xf32>>
+//      CHECK: %[[s34:.*]] = llvm.extractvalue %[[s33]][1] : !llvm.array<2 x vector<4xf32>>
+//      CHECK: %[[s35:.*]] = llvm.mlir.cast %[[B]] : vector<16x4x8xf32> to !llvm.array<16 x array<4 x vector<8xf32>>>
+//      CHECK: %[[s36:.*]] = llvm.extractvalue %[[s35]][0, 1] : !llvm.array<16 x array<4 x vector<8xf32>>>
+//      CHECK: %[[s37:.*]] = constant 0 : index
+//      CHECK: %[[s38:.*]] = llvm.mlir.cast %[[s37]] : index to i64
+//      CHECK: %[[s39:.*]] = llvm.extractelement %[[s34]]{{\[}}%[[s38]] : i64] : vector<4xf32>
+//      CHECK: %[[s40:.*]] = constant 2 : index
+//      CHECK: %[[s41:.*]] = llvm.mlir.cast %[[s40]] : index to i64
+//      CHECK: %[[s42:.*]] = llvm.insertelement %[[s39]], %[[s36]]{{\[}}%[[s41]] : i64] : vector<8xf32>
+//      CHECK: %[[s43:.*]] = constant 1 : index
+//      CHECK: %[[s44:.*]] = llvm.mlir.cast %[[s43]] : index to i64
+//      CHECK: %[[s45:.*]] = llvm.extractelement %[[s34]]{{\[}}%[[s44]] : i64] : vector<4xf32>
+//      CHECK: %[[s46:.*]] = constant 3 : index
+//      CHECK: %[[s47:.*]] = llvm.mlir.cast %[[s46]] : index to i64
+//      CHECK: %[[s48:.*]] = llvm.insertelement %[[s45]], %[[s42]]{{\[}}%[[s47]] : i64] : vector<8xf32>
+//      CHECK: %[[s49:.*]] = constant 2 : index
+//      CHECK: %[[s50:.*]] = llvm.mlir.cast %[[s49]] : index to i64
+//      CHECK: %[[s51:.*]] = llvm.extractelement %[[s34]]{{\[}}%[[s50]] : i64] : vector<4xf32>
+//      CHECK: %[[s52:.*]] = constant 4 : index
+//      CHECK: %[[s53:.*]] = llvm.mlir.cast %[[s52]] : index to i64
+//      CHECK: %[[s54:.*]] = llvm.insertelement %[[s51]], %[[s48]]{{\[}}%[[s53]] : i64] : vector<8xf32>
+//      CHECK: %[[s55:.*]] = constant 3 : index
+//      CHECK: %[[s56:.*]] = llvm.mlir.cast %[[s55]] : index to i64
+//      CHECK: %[[s57:.*]] = llvm.extractelement %[[s34]]{{\[}}%[[s56]] : i64] : vector<4xf32>
+//      CHECK: %[[s58:.*]] = constant 5 : index
+//      CHECK: %[[s59:.*]] = llvm.mlir.cast %[[s58]] : index to i64
+//      CHECK: %[[s60:.*]] = llvm.insertelement %[[s57]], %[[s54]]{{\[}}%[[s59]] : i64] : vector<8xf32>
+//      CHECK: %[[s61:.*]] = llvm.insertvalue %[[s60]], %[[s32]][1] : !llvm.array<4 x vector<8xf32>>
+//      CHECK: %[[s62:.*]] = llvm.mlir.cast %[[B]] : vector<16x4x8xf32> to !llvm.array<16 x array<4 x vector<8xf32>>>
+//      CHECK: %[[s63:.*]] = llvm.insertvalue %[[s61]], %[[s62]][0] : !llvm.array<16 x array<4 x vector<8xf32>>>
+//      CHECK: %[[s64:.*]] = llvm.mlir.cast %[[s63]] : !llvm.array<16 x array<4 x vector<8xf32>>> to vector<16x4x8xf32>
+//      CHECK: return %[[s64]] : vector<16x4x8xf32>
 
 // -----
 
 func @extract_strides(%arg0: vector<3x3xf32>) -> vector<1x1xf32> {
   %0 = vector.extract_slices %arg0, [2, 2], [1, 1]
     : vector<3x3xf32> into tuple<vector<2x2xf32>, vector<2x1xf32>, vector<1x2xf32>, vector<1x1xf32>>
   %1 = vector.tuple_get %0, 3 : tuple<vector<2x2xf32>, vector<2x1xf32>, vector<1x2xf32>, vector<1x1xf32>>
   return %1 : vector<1x1xf32>
 }
-// CHECK-LABEL: llvm.func @extract_strides(
-// CHECK-SAME: %[[A:.*]]: !llvm.array<3 x vector<3xf32>>)
-//      CHECK: %[[T1:.*]] = llvm.mlir.constant(dense<0.000000e+00> : vector<1x1xf32>) : !llvm.array<1 x vector<1xf32>>
+// CHECK-LABEL: @extract_strides(
+// CHECK-SAME: %[[ARG:.*]]: vector<3x3xf32>)
+//      CHECK: %[[VAL_1:.*]] = constant 0.000000e+00 : f32
+//      CHECK: %[[VAL_2:.*]] = splat %[[VAL_1]] : vector<1x1xf32>
+//      CHECK: %[[A:.*]] = llvm.mlir.cast %[[ARG]] : vector<3x3xf32> to !llvm.array<3 x vector<3xf32>>
 //      CHECK: %[[T2:.*]] = llvm.extractvalue %[[A]][2] : !llvm.array<3 x vector<3xf32>>
 //      CHECK: %[[T3:.*]] = llvm.shufflevector %[[T2]], %[[T2]] [2] : vector<3xf32>, vector<3xf32>
-//      CHECK: %[[T4:.*]] = llvm.insertvalue %[[T3]], %[[T1]][0] : !llvm.array<1 x vector<1xf32>>
-//      CHECK: llvm.return %[[T4]] : !llvm.array<1 x vector<1xf32>>
+//      CHECK: %[[VAL_6:.*]] = llvm.mlir.cast %[[VAL_2]] : vector<1x1xf32> to !llvm.array<1 x vector<1xf32>>
+//      CHECK: %[[T4:.*]] = llvm.insertvalue %[[T3]], %[[VAL_6]][0] : !llvm.array<1 x vector<1xf32>>
+//      CHECK: %[[VAL_8:.*]] = llvm.mlir.cast %[[T4]] : !llvm.array<1 x vector<1xf32>> to vector<1x1xf32>
+//      CHECK: return %[[VAL_8]] : vector<1x1xf32>
 
-// CHECK-LABEL: llvm.func @vector_fma(
-//  CHECK-SAME: %[[A:.*]]: vector<8xf32>, %[[B:.*]]: !llvm.array<2 x vector<4xf32>>)
-//  CHECK-SAME: -> !llvm.struct<(vector<8xf32>, array<2 x vector<4xf32>>)> {
 // -----
 
 func @vector_fma(%a: vector<8xf32>, %b: vector<2x4xf32>) -> (vector<8xf32>, vector<2x4xf32>) {
-  //         CHECK: "llvm.intr.fmuladd"(%[[A]], %[[A]], %[[A]]) :
-  //    CHECK-SAME:   (vector<8xf32>, vector<8xf32>, vector<8xf32>) -> vector<8xf32>
+  // CHECK-LABEL: @vector_fma
+  //  CHECK-SAME: %[[A:.*]]: vector<8xf32>
+  //  CHECK-SAME: %[[B:.*]]: vector<2x4xf32>
+  //       CHECK: "llvm.intr.fmuladd"
+  //  CHECK-SAME:   (vector<8xf32>, vector<8xf32>, vector<8xf32>) -> vector<8xf32>
   %0 = vector.fma %a, %a, %a : vector<8xf32>
 
-  //       CHECK: %[[b00:.*]] = llvm.extractvalue %[[B]][0] : !llvm.array<2 x vector<4xf32>>
-  //       CHECK: %[[b01:.*]] = llvm.extractvalue %[[B]][0] : !llvm.array<2 x vector<4xf32>>
-  //       CHECK: %[[b02:.*]] = llvm.extractvalue %[[B]][0] : !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[BL:.*]] = llvm.mlir.cast %[[B]] : vector<2x4xf32> to !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[b00:.*]] = llvm.extractvalue %[[BL]][0] : !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[BL:.*]] = llvm.mlir.cast %[[B]] : vector<2x4xf32> to !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[b01:.*]] = llvm.extractvalue %[[BL]][0] : !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[BL:.*]] = llvm.mlir.cast %[[B]] : vector<2x4xf32> to !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[b02:.*]] = llvm.extractvalue %[[BL]][0] : !llvm.array<2 x vector<4xf32>>
   //       CHECK: %[[B0:.*]] = "llvm.intr.fmuladd"(%[[b00]], %[[b01]], %[[b02]]) :
   //  CHECK-SAME: (vector<4xf32>, vector<4xf32>, vector<4xf32>) -> vector<4xf32>
   //       CHECK: llvm.insertvalue %[[B0]], {{.*}}[0] : !llvm.array<2 x vector<4xf32>>
-  //       CHECK: %[[b10:.*]] = llvm.extractvalue %[[B]][1] : !llvm.array<2 x vector<4xf32>>
-  //       CHECK: %[[b11:.*]] = llvm.extractvalue %[[B]][1] : !llvm.array<2 x vector<4xf32>>
-  //       CHECK: %[[b12:.*]] = llvm.extractvalue %[[B]][1] : !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[BL:.*]] = llvm.mlir.cast %[[B]] : vector<2x4xf32> to !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[b10:.*]] = llvm.extractvalue %[[BL]][1] : !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[BL:.*]] = llvm.mlir.cast %[[B]] : vector<2x4xf32> to !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[b11:.*]] = llvm.extractvalue %[[BL]][1] : !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[BL:.*]] = llvm.mlir.cast %[[B]] : vector<2x4xf32> to !llvm.array<2 x vector<4xf32>>
+  //       CHECK: %[[b12:.*]] = llvm.extractvalue %[[BL]][1] : !llvm.array<2 x vector<4xf32>>
   //       CHECK: %[[B1:.*]] = "llvm.intr.fmuladd"(%[[b10]], %[[b11]], %[[b12]]) :
   //  CHECK-SAME: (vector<4xf32>, vector<4xf32>, vector<4xf32>) -> vector<4xf32>
   //       CHECK: llvm.insertvalue %[[B1]], {{.*}}[1] : !llvm.array<2 x vector<4xf32>>
   %1 = vector.fma %b, %b, %b : vector<2x4xf32>
 
   return %0, %1: vector<8xf32>, vector<2x4xf32>
 }
 
 // -----
 
 func @reduce_f16(%arg0: vector<16xf16>) -> f16 {
   %0 = vector.reduction "add", %arg0 : vector<16xf16> into f16
   return %0 : f16
 }
-// CHECK-LABEL: llvm.func @reduce_f16(
+// CHECK-LABEL: @reduce_f16(
 // CHECK-SAME: %[[A:.*]]: vector<16xf16>)
 //      CHECK: %[[C:.*]] = llvm.mlir.constant(0.000000e+00 : f16) : f16
 //      CHECK: %[[V:.*]] = "llvm.intr.vector.reduce.fadd"(%[[C]], %[[A]])
 // CHECK-SAME: {reassoc = false} : (f16, vector<16xf16>) -> f16
-//      CHECK: llvm.return %[[V]] : f16
+//      CHECK: return %[[V]] : f16
 
 // -----
 
 func @reduce_f32(%arg0: vector<16xf32>) -> f32 {
   %0 = vector.reduction "add", %arg0 : vector<16xf32> into f32
   return %0 : f32
 }
-// CHECK-LABEL: llvm.func @reduce_f32(
+// CHECK-LABEL: @reduce_f32(
 // CHECK-SAME: %[[A:.*]]: vector<16xf32>)
 //      CHECK: %[[C:.*]] = llvm.mlir.constant(0.000000e+00 : f32) : f32
 //      CHECK: %[[V:.*]] = "llvm.intr.vector.reduce.fadd"(%[[C]], %[[A]])
 // CHECK-SAME: {reassoc = false} : (f32, vector<16xf32>) -> f32
-//      CHECK: llvm.return %[[V]] : f32
+//      CHECK: return %[[V]] : f32
 
 // -----
 
 func @reduce_f64(%arg0: vector<16xf64>) -> f64 {
   %0 = vector.reduction "add", %arg0 : vector<16xf64> into f64
   return %0 : f64
 }
-// CHECK-LABEL: llvm.func @reduce_f64(
+// CHECK-LABEL: @reduce_f64(
 // CHECK-SAME: %[[A:.*]]: vector<16xf64>)
 //      CHECK: %[[C:.*]] = llvm.mlir.constant(0.000000e+00 : f64) : f64
 //      CHECK: %[[V:.*]] = "llvm.intr.vector.reduce.fadd"(%[[C]], %[[A]])
 // CHECK-SAME: {reassoc = false} : (f64, vector<16xf64>) -> f64
-//      CHECK: llvm.return %[[V]] : f64
+//      CHECK: return %[[V]] : f64
 
 // -----
 
 func @reduce_i8(%arg0: vector<16xi8>) -> i8 {
   %0 = vector.reduction "add", %arg0 : vector<16xi8> into i8
   return %0 : i8
 }
-// CHECK-LABEL: llvm.func @reduce_i8(
+// CHECK-LABEL: @reduce_i8(
 // CHECK-SAME: %[[A:.*]]: vector<16xi8>)
 //      CHECK: %[[V:.*]] = "llvm.intr.vector.reduce.add"(%[[A]])
-//      CHECK: llvm.return %[[V]] : i8
+//      CHECK: return %[[V]] : i8
 
 // -----
 
 func @reduce_i32(%arg0: vector<16xi32>) -> i32 {
   %0 = vector.reduction "add", %arg0 : vector<16xi32> into i32
   return %0 : i32
 }
-// CHECK-LABEL: llvm.func @reduce_i32(
+// CHECK-LABEL: @reduce_i32(
 // CHECK-SAME: %[[A:.*]]: vector<16xi32>)
 //      CHECK: %[[V:.*]] = "llvm.intr.vector.reduce.add"(%[[A]])
-//      CHECK: llvm.return %[[V]] : i32
+//      CHECK: return %[[V]] : i32
 
 // -----
 
 func @reduce_i64(%arg0: vector<16xi64>) -> i64 {
   %0 = vector.reduction "add", %arg0 : vector<16xi64> into i64
   return %0 : i64
 }
-// CHECK-LABEL: llvm.func @reduce_i64(
+// CHECK-LABEL: @reduce_i64(
 // CHECK-SAME: %[[A:.*]]: vector<16xi64>)
 //      CHECK: %[[V:.*]] = "llvm.intr.vector.reduce.add"(%[[A]])
-//      CHECK: llvm.return %[[V]] : i64
+//      CHECK: return %[[V]] : i64
 
 
 //                          4x16                16x3               4x3
 // -----
 
 func @matrix_ops(%A: vector<64xf64>, %B: vector<48xf64>) -> vector<12xf64> {
   %C = vector.matrix_multiply %A, %B
     { lhs_rows = 4: i32, lhs_columns = 16: i32 , rhs_columns = 3: i32 } :
     (vector<64xf64>, vector<48xf64>) -> vector<12xf64>
   return %C: vector<12xf64>
 }
-// CHECK-LABEL: llvm.func @matrix_ops
+// CHECK-LABEL: @matrix_ops
 //       CHECK:   llvm.intr.matrix.multiply %{{.*}}, %{{.*}} {
 //  CHECK-SAME: lhs_columns = 16 : i32, lhs_rows = 4 : i32, rhs_columns = 3 : i32
 //  CHECK-SAME: } : (vector<64xf64>, vector<48xf64>) -> vector<12xf64>
 
 // -----
 
 func @transfer_read_1d(%A : memref<?xf32>, %base: index) -> vector<17xf32> {
   %f7 = constant 7.0: f32
   %f = vector.transfer_read %A[%base], %f7
       {permutation_map = affine_map<(d0) -> (d0)>} :
     memref<?xf32>, vector<17xf32>
   vector.transfer_write %f, %A[%base]
       {permutation_map = affine_map<(d0) -> (d0)>} :
     vector<17xf32>, memref<?xf32>
   return %f: vector<17xf32>
 }
 // CHECK-LABEL: func @transfer_read_1d
-//  CHECK-SAME: %[[BASE:[a-zA-Z0-9]*]]: i64) -> vector<17xf32>
+//  CHECK-SAME: %[[BASE:[a-zA-Z0-9]*]]: index) -> vector<17xf32>
+//       CHECK: %[[c7:.*]] = constant 7.0
 //
 // 1. Bitcast to vector form.
 //       CHECK: %[[gep:.*]] = llvm.getelementptr {{.*}} :
 //  CHECK-SAME: (!llvm.ptr<f32>, i64) -> !llvm.ptr<f32>
 //       CHECK: %[[vecPtr:.*]] = llvm.bitcast %[[gep]] :
 //  CHECK-SAME: !llvm.ptr<f32> to !llvm.ptr<vector<17xf32>>
-//       CHECK: %[[DIM:.*]] = llvm.extractvalue %{{.*}}[3, 0] :
-//  CHECK-SAME: !llvm.struct<(ptr<f32>, ptr<f32>, i64, array<1 x i64>, array<1 x i64>)>
+//       CHECK: %[[C0:.*]] = constant 0 : index
+//       CHECK: %[[DIM:.*]] = dim %{{.*}}, %[[C0]] : memref<?xf32>
 //
 // 2. Create a vector with linear indices [ 0 .. vector_length - 1 ].
-//       CHECK: %[[linearIndex:.*]] = llvm.mlir.constant(dense
+//       CHECK: %[[linearIndex:.*]] = constant dense
 //  CHECK-SAME: <[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16]> :
-//  CHECK-SAME: vector<17xi32>) : vector<17xi32>
+//  CHECK-SAME: vector<17xi32>
 //
 // 3. Create offsetVector = [ offset + 0 .. offset + vector_length - 1 ].
-//       CHECK: %[[otrunc:.*]] = llvm.trunc %[[BASE]] : i64 to i32
-//       CHECK: %[[offsetVec:.*]] = llvm.mlir.undef : vector<17xi32>
-//       CHECK: %[[c0:.*]] = llvm.mlir.constant(0 : i32) : i32
-//       CHECK: %[[offsetVec2:.*]] = llvm.insertelement %[[otrunc]], %[[offsetVec]][%[[c0]] :
-//  CHECK-SAME: i32] : vector<17xi32>
-//       CHECK: %[[offsetVec3:.*]] = llvm.shufflevector %[[offsetVec2]], %{{.*}} [
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32] :
-//  CHECK-SAME: vector<17xi32>, vector<17xi32>
-//       CHECK: %[[offsetVec4:.*]] = llvm.add %[[offsetVec3]], %[[linearIndex]] :
-//  CHECK-SAME: vector<17xi32>
+//       CHECK: %[[otrunc:.*]] = index_cast %[[BASE]] : index to i32
+//       CHECK: %[[offsetVec:.*]] = splat %[[otrunc]] : vector<17xi32>
+//       CHECK: %[[offsetVec2:.*]] = addi %[[offsetVec]], %[[linearIndex]] : vector<17xi32>
 //
 // 4. Let dim the memref dimension, compute the vector comparison mask:
 //    [ offset + 0 .. offset + vector_length - 1 ] < [ dim .. dim ]
-//       CHECK: %[[dtrunc:.*]] = llvm.trunc %[[DIM]] : i64 to i32
-//       CHECK: %[[dimVec:.*]] = llvm.mlir.undef : vector<17xi32>
-//       CHECK: %[[c01:.*]] = llvm.mlir.constant(0 : i32) : i32
-//       CHECK: %[[dimVec2:.*]] = llvm.insertelement %[[dtrunc]], %[[dimVec]][%[[c01]] :
-//  CHECK-SAME:  i32] : vector<17xi32>
-//       CHECK: %[[dimVec3:.*]] = llvm.shufflevector %[[dimVec2]], %{{.*}} [
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32] :
-//  CHECK-SAME: vector<17xi32>, vector<17xi32>
-//       CHECK: %[[mask:.*]] = llvm.icmp "slt" %[[offsetVec4]], %[[dimVec3]] :
-//  CHECK-SAME: vector<17xi32>
+//       CHECK: %[[dtrunc:.*]] = index_cast %[[DIM]] : index to i32
+//       CHECK: %[[dimVec:.*]] = splat %[[dtrunc]] : vector<17xi32>
+//       CHECK: %[[mask:.*]] = cmpi slt, %[[offsetVec2]], %[[dimVec]] : vector<17xi32>
 //
 // 5. Rewrite as a masked read.
-//       CHECK: %[[PASS_THROUGH:.*]] =  llvm.mlir.constant(dense<7.000000e+00> :
-//  CHECK-SAME:  vector<17xf32>) : vector<17xf32>
+//       CHECK: %[[PASS_THROUGH:.*]] = splat %[[c7]] : vector<17xf32>
 //       CHECK: %[[loaded:.*]] = llvm.intr.masked.load %[[vecPtr]], %[[mask]],
 //  CHECK-SAME: %[[PASS_THROUGH]] {alignment = 4 : i32} :
 //  CHECK-SAME: (!llvm.ptr<vector<17xf32>>, vector<17xi1>, vector<17xf32>) -> vector<17xf32>
 
 //
 // 1. Bitcast to vector form.
 //       CHECK: %[[gep_b:.*]] = llvm.getelementptr {{.*}} :
 //  CHECK-SAME: (!llvm.ptr<f32>, i64) -> !llvm.ptr<f32>
 //       CHECK: %[[vecPtr_b:.*]] = llvm.bitcast %[[gep_b]] :
 //  CHECK-SAME: !llvm.ptr<f32> to !llvm.ptr<vector<17xf32>>
 //
 // 2. Create a vector with linear indices [ 0 .. vector_length - 1 ].
-//       CHECK: %[[linearIndex_b:.*]] = llvm.mlir.constant(dense
+//       CHECK: %[[linearIndex_b:.*]] = constant dense
 //  CHECK-SAME: <[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16]> :
-//  CHECK-SAME: vector<17xi32>) : vector<17xi32>
+//  CHECK-SAME: vector<17xi32>
 //
 // 3. Create offsetVector = [ offset + 0 .. offset + vector_length - 1 ].
-//       CHECK: llvm.shufflevector {{.*}} [0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32] :
-//  CHECK-SAME: vector<17xi32>, vector<17xi32>
-//       CHECK: llvm.add
+//       CHECK: splat %{{.*}} : vector<17xi32>
+//       CHECK: addi
 //
 // 4. Let dim the memref dimension, compute the vector comparison mask:
 //    [ offset + 0 .. offset + vector_length - 1 ] < [ dim .. dim ]
-//       CHECK: llvm.shufflevector {{.*}} [0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32] :
-//  CHECK-SAME: vector<17xi32>, vector<17xi32>
-//       CHECK: %[[mask_b:.*]] = llvm.icmp "slt" {{.*}} : vector<17xi32>
+//       CHECK: splat %{{.*}} : vector<17xi32>
+//       CHECK: %[[mask_b:.*]] = cmpi slt, {{.*}} : vector<17xi32>
 //
 // 5. Rewrite as a masked write.
 //       CHECK: llvm.intr.masked.store %[[loaded]], %[[vecPtr_b]], %[[mask_b]]
 //  CHECK-SAME: {alignment = 4 : i32} :
 //  CHECK-SAME: vector<17xf32>, vector<17xi1> into !llvm.ptr<vector<17xf32>>
 
 // -----
 
 func @transfer_read_2d_to_1d(%A : memref<?x?xf32>, %base0: index, %base1: index) -> vector<17xf32> {
   %f7 = constant 7.0: f32
   %f = vector.transfer_read %A[%base0, %base1], %f7
       {permutation_map = affine_map<(d0, d1) -> (d1)>} :
     memref<?x?xf32>, vector<17xf32>
   return %f: vector<17xf32>
 }
 // CHECK-LABEL: func @transfer_read_2d_to_1d
-//  CHECK-SAME: %[[BASE_0:[a-zA-Z0-9]*]]: i64, %[[BASE_1:[a-zA-Z0-9]*]]: i64) -> vector<17xf32>
-//       CHECK: %[[DIM:.*]] = llvm.extractvalue %{{.*}}[3, 1] :
-//  CHECK-SAME: !llvm.struct<(ptr<f32>, ptr<f32>, i64, array<2 x i64>, array<2 x i64>)>
+//  CHECK-SAME: %[[BASE_0:[a-zA-Z0-9]*]]: index, %[[BASE_1:[a-zA-Z0-9]*]]: index) -> vector<17xf32>
+//       CHECK: %[[c1:.*]] = constant 1 : index
+//       CHECK: %[[DIM:.*]] = dim %{{.*}}, %[[c1]] : memref<?x?xf32>
 //
 // Create offsetVector = [ offset + 0 .. offset + vector_length - 1 ].
-//       CHECK: %[[trunc:.*]] = llvm.trunc %[[BASE_1]] : i64 to i32
-//       CHECK: %[[offsetVec:.*]] = llvm.mlir.undef : vector<17xi32>
-//       CHECK: %[[c0:.*]] = llvm.mlir.constant(0 : i32) : i32
-//       CHECK: %[[offsetVec2:.*]] = llvm.insertelement %[[trunc]], %[[offsetVec]][%[[c0]] :
-//  CHECK-SAME: i32] : vector<17xi32>
-//       CHECK: %[[offsetVec3:.*]] = llvm.shufflevector %[[offsetVec2]], %{{.*}} [
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32] :
-//  CHECK-SAME: vector<17xi32>, vector<17xi32>
+//       CHECK: %[[trunc:.*]] = index_cast %[[BASE_1]] : index to i32
+//       CHECK: %[[offsetVec:.*]] = splat %[[trunc]] : vector<17xi32>
 //
 // Let dim the memref dimension, compute the vector comparison mask:
 //    [ offset + 0 .. offset + vector_length - 1 ] < [ dim .. dim ]
-//       CHECK: %[[dimtrunc:.*]] = llvm.trunc %[[DIM]] : i64 to i32
-//       CHECK: %[[dimVec:.*]] = llvm.mlir.undef : vector<17xi32>
-//       CHECK: %[[c01:.*]] = llvm.mlir.constant(0 : i32) : i32
-//       CHECK: %[[dimVec2:.*]] = llvm.insertelement %[[dimtrunc]], %[[dimVec]][%[[c01]] :
-//  CHECK-SAME:  i32] : vector<17xi32>
-//       CHECK: %[[dimVec3:.*]] = llvm.shufflevector %[[dimVec2]], %{{.*}} [
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32, 0 : i32,
-//  CHECK-SAME:  0 : i32, 0 : i32, 0 : i32] :
-//  CHECK-SAME: vector<17xi32>, vector<17xi32>
+//       CHECK: %[[dimtrunc:.*]] = index_cast %[[DIM]] : index to i32
+//       CHECK: splat %[[dimtrunc]] : vector<17xi32>
 
 // -----
 
 func @transfer_read_1d_non_zero_addrspace(%A : memref<?xf32, 3>, %base: index) -> vector<17xf32> {
   %f7 = constant 7.0: f32
   %f = vector.transfer_read %A[%base], %f7
       {permutation_map = affine_map<(d0) -> (d0)>} :
     memref<?xf32, 3>, vector<17xf32>
   vector.transfer_write %f, %A[%base]
       {permutation_map = affine_map<(d0) -> (d0)>} :
     vector<17xf32>, memref<?xf32, 3>
   return %f: vector<17xf32>
 }
 // CHECK-LABEL: func @transfer_read_1d_non_zero_addrspace
-//  CHECK-SAME: %[[BASE:[a-zA-Z0-9]*]]: i64) -> vector<17xf32>
+//  CHECK-SAME: %[[BASE:[a-zA-Z0-9]*]]: index) -> vector<17xf32>
 //
 // 1. Check address space for GEP is correct.
 //       CHECK: %[[gep:.*]] = llvm.getelementptr {{.*}} :
 //  CHECK-SAME: (!llvm.ptr<f32, 3>, i64) -> !llvm.ptr<f32, 3>
 //       CHECK: %[[vecPtr:.*]] = llvm.addrspacecast %[[gep]] :
 //  CHECK-SAME: !llvm.ptr<f32, 3> to !llvm.ptr<vector<17xf32>>
 //
 // 2. Check address space of the memref is correct.
-//       CHECK: %[[DIM:.*]] = llvm.extractvalue %{{.*}}[3, 0] :
-//  CHECK-SAME: !llvm.struct<(ptr<f32, 3>, ptr<f32, 3>, i64, array<1 x i64>, array<1 x i64>)>
+//       CHECK: %[[c0:.*]] = constant 0 : index
+//       CHECK: %[[DIM:.*]] = dim %{{.*}}, %[[c0]] : memref<?xf32, 3>
 //
 // 3. Check address apce for GEP is correct.
 //       CHECK: %[[gep_b:.*]] = llvm.getelementptr {{.*}} :
 //  CHECK-SAME: (!llvm.ptr<f32, 3>, i64) -> !llvm.ptr<f32, 3>
 //       CHECK: %[[vecPtr_b:.*]] = llvm.addrspacecast %[[gep_b]] :
 //  CHECK-SAME: !llvm.ptr<f32, 3> to !llvm.ptr<vector<17xf32>>
 
 // -----
 
 func @transfer_read_1d_not_masked(%A : memref<?xf32>, %base: index) -> vector<17xf32> {
   %f7 = constant 7.0: f32
   %f = vector.transfer_read %A[%base], %f7 {masked = [false]} :
     memref<?xf32>, vector<17xf32>
   return %f: vector<17xf32>
 }
 // CHECK-LABEL: func @transfer_read_1d_not_masked
-//  CHECK-SAME: %[[BASE:[a-zA-Z0-9]*]]: i64) -> vector<17xf32>
+//  CHECK-SAME: %[[BASE:[a-zA-Z0-9]*]]: index) -> vector<17xf32>
 //
 // 1. Bitcast to vector form.
 //       CHECK: %[[gep:.*]] = llvm.getelementptr {{.*}} :
 //  CHECK-SAME: (!llvm.ptr<f32>, i64) -> !llvm.ptr<f32>
 //       CHECK: %[[vecPtr:.*]] = llvm.bitcast %[[gep]] :
 //  CHECK-SAME: !llvm.ptr<f32> to !llvm.ptr<vector<17xf32>>
 //
 // 2. Rewrite as a load.
 //       CHECK: %[[loaded:.*]] = llvm.load %[[vecPtr]] {alignment = 4 : i64} : !llvm.ptr<vector<17xf32>>
 
 // -----
 
 func @transfer_read_1d_cast(%A : memref<?xi32>, %base: index) -> vector<12xi8> {
   %c0 = constant 0: i32
   %v = vector.transfer_read %A[%base], %c0 {masked = [false]} :
     memref<?xi32>, vector<12xi8>
   return %v: vector<12xi8>
 }
 // CHECK-LABEL: func @transfer_read_1d_cast
-//  CHECK-SAME: %[[BASE:[a-zA-Z0-9]*]]: i64) -> vector<12xi8>
+//  CHECK-SAME: %[[BASE:[a-zA-Z0-9]*]]: index) -> vector<12xi8>
 //
 // 1. Bitcast to vector form.
 //       CHECK: %[[gep:.*]] = llvm.getelementptr {{.*}} :
 //  CHECK-SAME: (!llvm.ptr<i32>, i64) -> !llvm.ptr<i32>
 //       CHECK: %[[vecPtr:.*]] = llvm.bitcast %[[gep]] :
 //  CHECK-SAME: !llvm.ptr<i32> to !llvm.ptr<vector<12xi8>>
 //
 // 2. Rewrite as a load.
 //       CHECK: %[[loaded:.*]] = llvm.load %[[vecPtr]] {alignment = 4 : i64} : !llvm.ptr<vector<12xi8>>
 
 // -----
 
 func @genbool_1d() -> vector<8xi1> {
   %0 = vector.constant_mask [4] : vector<8xi1>
   return %0 : vector<8xi1>
 }
 // CHECK-LABEL: func @genbool_1d
-// CHECK: %[[C1:.*]] = llvm.mlir.constant(dense<[true, true, true, true, false, false, false, false]> : vector<8xi1>) : vector<8xi1>
-// CHECK: llvm.return %[[C1]] : vector<8xi1>
+// CHECK: %[[VAL_0:.*]] = constant dense<[true, true, true, true, false, false, false, false]> : vector<8xi1>
+// CHECK: return %[[VAL_0]] : vector<8xi1>
 
 // -----
 
 func @genbool_2d() -> vector<4x4xi1> {
   %v = vector.constant_mask [2, 2] : vector<4x4xi1>
   return %v: vector<4x4xi1>
 }
 
 // CHECK-LABEL: func @genbool_2d
-// CHECK: %[[C1:.*]] = llvm.mlir.constant(dense<[true, true, false, false]> : vector<4xi1>) : vector<4xi1>
-// CHECK: %[[C2:.*]] = llvm.mlir.constant(dense<false> : vector<4x4xi1>) : !llvm.array<4 x vector<4xi1>>
-// CHECK: %[[T0:.*]] = llvm.insertvalue %[[C1]], %[[C2]][0] : !llvm.array<4 x vector<4xi1>>
-// CHECK: %[[T1:.*]] = llvm.insertvalue %[[C1]], %[[T0]][1] : !llvm.array<4 x vector<4xi1>>
-// CHECK: llvm.return %[[T1]] : !llvm.array<4 x vector<4xi1>>
+// CHECK: %[[VAL_0:.*]] = constant dense<[true, true, false, false]> : vector<4xi1>
+// CHECK: %[[VAL_1:.*]] = constant dense<false> : vector<4x4xi1>
+// CHECK: %[[VAL_2:.*]] = llvm.mlir.cast %[[VAL_1]] : vector<4x4xi1> to !llvm.array<4 x vector<4xi1>>
+// CHECK: %[[VAL_3:.*]] = llvm.insertvalue %[[VAL_0]], %[[VAL_2]][0] : !llvm.array<4 x vector<4xi1>>
+// CHECK: %[[VAL_4:.*]] = llvm.insertvalue %[[VAL_0]], %[[VAL_3]][1] : !llvm.array<4 x vector<4xi1>>
+// CHECK: %[[VAL_5:.*]] = llvm.mlir.cast %[[VAL_4]] : !llvm.array<4 x vector<4xi1>> to vector<4x4xi1>
+// CHECK: return %[[VAL_5]] : vector<4x4xi1>
 
 // -----
 
 func @flat_transpose(%arg0: vector<16xf32>) -> vector<16xf32> {
   %0 = vector.flat_transpose %arg0 { rows = 4: i32, columns = 4: i32 }
      : vector<16xf32> -> vector<16xf32>
   return %0 : vector<16xf32>
 }
 
 // CHECK-LABEL: func @flat_transpose
 // CHECK-SAME:  %[[A:.*]]: vector<16xf32>
 // CHECK:       %[[T:.*]] = llvm.intr.matrix.transpose %[[A]]
 // CHECK-SAME:      {columns = 4 : i32, rows = 4 : i32} :
 // CHECK-SAME:      vector<16xf32> into vector<16xf32>
-// CHECK:       llvm.return %[[T]] : vector<16xf32>
+// CHECK:       return %[[T]] : vector<16xf32>
 
 // -----
 
 func @masked_load_op(%arg0: memref<?xf32>, %arg1: vector<16xi1>, %arg2: vector<16xf32>) -> vector<16xf32> {
   %c0 = constant 0: index
   %0 = vector.maskedload %arg0[%c0], %arg1, %arg2 : memref<?xf32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
   return %0 : vector<16xf32>
 }
 
 // CHECK-LABEL: func @masked_load_op
-// CHECK: %[[C:.*]] = llvm.mlir.constant(0 : index) : i64
+// CHECK: %[[CO:.*]] = constant 0 : index
+// CHECK: %[[C:.*]] = llvm.mlir.cast %[[CO]] : index to i64
 // CHECK: %[[P:.*]] = llvm.getelementptr %{{.*}}[%[[C]]] : (!llvm.ptr<f32>, i64) -> !llvm.ptr<f32>
 // CHECK: %[[B:.*]] = llvm.bitcast %[[P]] : !llvm.ptr<f32> to !llvm.ptr<vector<16xf32>>
 // CHECK: %[[L:.*]] = llvm.intr.masked.load %[[B]], %{{.*}}, %{{.*}} {alignment = 4 : i32} : (!llvm.ptr<vector<16xf32>>, vector<16xi1>, vector<16xf32>) -> vector<16xf32>
-// CHECK: llvm.return %[[L]] : vector<16xf32>
+// CHECK: return %[[L]] : vector<16xf32>
 
 // -----
 
 func @masked_store_op(%arg0: memref<?xf32>, %arg1: vector<16xi1>, %arg2: vector<16xf32>) {
   %c0 = constant 0: index
   vector.maskedstore %arg0[%c0], %arg1, %arg2 : memref<?xf32>, vector<16xi1>, vector<16xf32>
   return
 }
 
 // CHECK-LABEL: func @masked_store_op
-// CHECK: %[[C:.*]] = llvm.mlir.constant(0 : index) : i64
+// CHECK: %[[CO:.*]] = constant 0 : index
+// CHECK: %[[C:.*]] = llvm.mlir.cast %[[CO]] : index to i64
 // CHECK: %[[P:.*]] = llvm.getelementptr %{{.*}}[%[[C]]] : (!llvm.ptr<f32>, i64) -> !llvm.ptr<f32>
 // CHECK: %[[B:.*]] = llvm.bitcast %[[P]] : !llvm.ptr<f32> to !llvm.ptr<vector<16xf32>>
 // CHECK: llvm.intr.masked.store %{{.*}}, %[[B]], %{{.*}} {alignment = 4 : i32} : vector<16xf32>, vector<16xi1> into !llvm.ptr<vector<16xf32>>
-// CHECK: llvm.return
 
 // -----
 
 func @gather_op(%arg0: memref<?xf32>, %arg1: vector<3xi32>, %arg2: vector<3xi1>, %arg3: vector<3xf32>) -> vector<3xf32> {
   %0 = vector.gather %arg0[%arg1], %arg2, %arg3 : memref<?xf32>, vector<3xi32>, vector<3xi1>, vector<3xf32> into vector<3xf32>
   return %0 : vector<3xf32>
 }
 
 // CHECK-LABEL: func @gather_op
 // CHECK: %[[P:.*]] = llvm.getelementptr {{.*}}[%{{.*}}] : (!llvm.ptr<f32>, vector<3xi32>) -> !llvm.vec<3 x ptr<f32>>
 // CHECK: %[[G:.*]] = llvm.intr.masked.gather %[[P]], %{{.*}}, %{{.*}} {alignment = 4 : i32} : (!llvm.vec<3 x ptr<f32>>, vector<3xi1>, vector<3xf32>) -> vector<3xf32>
-// CHECK: llvm.return %[[G]] : vector<3xf32>
+// CHECK: return %[[G]] : vector<3xf32>
 
 // -----
 
 func @scatter_op(%arg0: memref<?xf32>, %arg1: vector<3xi32>, %arg2: vector<3xi1>, %arg3: vector<3xf32>) {
   vector.scatter %arg0[%arg1], %arg2, %arg3 : memref<?xf32>, vector<3xi32>, vector<3xi1>, vector<3xf32>
   return
 }
 
 // CHECK-LABEL: func @scatter_op
 // CHECK: %[[P:.*]] = llvm.getelementptr {{.*}}[%{{.*}}] : (!llvm.ptr<f32>, vector<3xi32>) -> !llvm.vec<3 x ptr<f32>>
 // CHECK: llvm.intr.masked.scatter %{{.*}}, %[[P]], %{{.*}} {alignment = 4 : i32} : vector<3xf32>, vector<3xi1> into !llvm.vec<3 x ptr<f32>>
-// CHECK: llvm.return
 
 // -----
 
 func @expand_load_op(%arg0: memref<?xf32>, %arg1: vector<11xi1>, %arg2: vector<11xf32>) -> vector<11xf32> {
   %c0 = constant 0: index
   %0 = vector.expandload %arg0[%c0], %arg1, %arg2 : memref<?xf32>, vector<11xi1>, vector<11xf32> into vector<11xf32>
   return %0 : vector<11xf32>
 }
 
 // CHECK-LABEL: func @expand_load_op
-// CHECK: %[[C:.*]] = llvm.mlir.constant(0 : index) : i64
+// CHECK: %[[CO:.*]] = constant 0 : index
+// CHECK: %[[C:.*]] = llvm.mlir.cast %[[CO]] : index to i64
 // CHECK: %[[P:.*]] = llvm.getelementptr %{{.*}}[%[[C]]] : (!llvm.ptr<f32>, i64) -> !llvm.ptr<f32>
 // CHECK: %[[E:.*]] = "llvm.intr.masked.expandload"(%[[P]], %{{.*}}, %{{.*}}) : (!llvm.ptr<f32>, vector<11xi1>, vector<11xf32>) -> vector<11xf32>
-// CHECK: llvm.return %[[E]] : vector<11xf32>
+// CHECK: return %[[E]] : vector<11xf32>
 
 // -----
 
 func @compress_store_op(%arg0: memref<?xf32>, %arg1: vector<11xi1>, %arg2: vector<11xf32>) {
   %c0 = constant 0: index
   vector.compressstore %arg0[%c0], %arg1, %arg2 : memref<?xf32>, vector<11xi1>, vector<11xf32>
   return
 }
 
 // CHECK-LABEL: func @compress_store_op
-// CHECK: %[[C:.*]] = llvm.mlir.constant(0 : index) : i64
+// CHECK: %[[CO:.*]] = constant 0 : index
+// CHECK: %[[C:.*]] = llvm.mlir.cast %[[CO]] : index to i64
 // CHECK: %[[P:.*]] = llvm.getelementptr %{{.*}}[%[[C]]] : (!llvm.ptr<f32>, i64) -> !llvm.ptr<f32>
 // CHECK: "llvm.intr.masked.compressstore"(%{{.*}}, %[[P]], %{{.*}}) : (vector<11xf32>, !llvm.ptr<f32>, vector<11xi1>) -> ()
-// CHECK: llvm.return

mlir/test/Dialect/LLVMIR/dialect-cast.mlir

 // RUN: mlir-opt -split-input-file -verify-diagnostics %s
 
 // These are the supported cases, just make sure they don't trigger errors, op
 // syntax is tested elsewhere.
 
-func @mlir_dialect_cast(%0: index, %1: i32, %2: bf16, %3: f16, %4: f32, %5: f64,
+func @mlir_dialect_cast(%0: index, %1: vector<2x2x2xf32>,
                         %6: vector<42xf32>, %7: memref<42xf32>,
                         %8: memref<?xf32>, %9: memref<f32>,
                         %10: memref<*xf32>) {
   llvm.mlir.cast %0 : index to i64
   llvm.mlir.cast %0 : index to i32
+  llvm.mlir.cast %1 : vector<2x2x2xf32> to !llvm.array<2 x array<2 x vector<2xf32>>>
   llvm.mlir.cast %7 : memref<42xf32> to !llvm.ptr<f32>
   llvm.mlir.cast %7 : memref<42xf32> to !llvm.struct<(ptr<f32>, ptr<f32>, i64, array<1xi64>, array<1xi64>)>
   llvm.mlir.cast %8 : memref<?xf32> to !llvm.struct<(ptr<f32>, ptr<f32>, i64, array<1xi64>, array<1xi64>)>
   llvm.mlir.cast %9 : memref<f32> to !llvm.struct<(ptr<f32>, ptr<f32>, i64)>
   llvm.mlir.cast %10 : memref<*xf32> to !llvm.struct<(i64, ptr<i8>)>
   return
 }
 

 func @mlir_dialect_cast_f64(%0 : f64) {
   // expected-error@+1 {{unsupported cast}}
   llvm.mlir.cast %0 : f64 to f32
 }
 
 // -----
 
 func @mlir_dialect_cast_integer_non_integer(%0 : i16) {
-  // expected-error@+1 {{unsupported cast}}
+  // expected-error@+1 {{invalid cast between integer and non-integer}}
   llvm.mlir.cast %0 : i16 to f16
 }
 
 // -----
 
 func @mlir_dialect_cast_scalable_vector(%0 : vector<2xf32>) {
-  // expected-error@+1 {{vector types should not be casted}}
+  // expected-error@+1 {{invalid cast for vector types}}
   llvm.mlir.cast %0 : vector<2xf32> to !llvm.vec<?x2xf32>
 }
 
 // -----
 
+func @mlir_dialect_cast_vector_to_self(%0 : vector<2xf32>) {
+  // expected-error@+1 {{vector types should not be casted}}
+  llvm.mlir.cast %0 : vector<2xf32> to vector<2xf32>
+}
+
+// -----
+
+func @mlir_dialect_cast_nd_vector(%0 : vector<2x2xf32>) {
+  // expected-error@+1 {{invalid cast for vector, expected array}}
+  llvm.mlir.cast %0 : vector<2x2xf32> to !llvm.struct<()>
+}
+
+// -----
+
 func @mlir_dialect_cast_dynamic_memref_bare_ptr(%0 : memref<?xf32>) {
   // expected-error@+1 {{unexpected bare pointer for dynamically shaped memref}}
   llvm.mlir.cast %0 : memref<?xf32> to !llvm.ptr<f32>
 }
 
 // -----
 
 func @mlir_dialect_cast_memref_bare_ptr_space(%0 : memref<4xf32, 4>) {

mlir/test/Target/vector-to-llvm-ir.mlir

-// RUN: mlir-opt %s -convert-vector-to-llvm | mlir-translate -mlir-to-llvmir | FileCheck %s
+// RUN: mlir-opt %s -convert-vector-to-llvm -convert-std-to-llvm | mlir-translate -mlir-to-llvmir | FileCheck %s
 
 func @genbool_1d() -> vector<8xi1> {
   %0 = vector.constant_mask [4] : vector<8xi1>
   return %0 : vector<8xi1>
 }
 // CHECK-LABEL: @genbool_1d()
 // CHECK-NEXT: ret <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 false, i1 false, i1 false, i1 false>