[mlir][Linalg] Drop deprecated convolution vectorization patterns

Differential revision: https://reviews.llvm.org/D117326

Author: nicolasvasilache

mlir/include/mlir/Dialect/Linalg/Transforms/Transforms.h

@@ -49,11 +49,6 @@ using LinalgLoops = SmallVector<Operation *, 4>;
 void populatePadTensorTilingPatterns(RewritePatternSet &patterns,
                                      const LinalgTilingOptions &options);
 
-/// [DEPRECATED] Populate patterns for vectorization of all ConvN-D ops.
-void populateConvVectorizationPatterns(
-    MLIRContext *context, SmallVectorImpl<RewritePatternSet> &patterns,
-    ArrayRef<int64_t> tileSizes);
-
 /// Populate patterns for vectorizing low-D convolution ops. This is a step in
 /// progressive lowering for convolution ops, it assume high-D convolution ops
 /// were decomposed previously.
@@ -1246,54 +1241,6 @@ struct LinalgCopyVTWForwardingPattern
                                 PatternRewriter &rewriter) const override;
 };
 
-/// Converts Convolution op into vector contraction.
-///
-/// Conversion expects ConvOp to have dimensions marked in the *mask* as
-/// false of size 1. This ensures that the ConvOp can be lowered to vector
-/// contraction of dimensions marked in the *mask* as true.
-///
-/// A good example for vectorization is ConvNHWCOp which is 2D Conv op
-/// with channels as the last dimension. Let's vectorize last 3 dimensions.
-/// The initial op definition looks like this:
-/// ```
-/// linalg.conv_2d_nhwc  %arg0, %arg1, %arg2 :
-///   (memref<1x3x3x3xf32>, memref<1x3x3x3xf32>, memref<?x?x?x?xf32>)
-/// ```
-/// This op can be expressed as a dot product between %arg0 (input) and
-/// %arg1 (kernel) which is written into first entry of %arg2 (output). This is
-/// the ConvOp this pass expects and converts into:
-/// ```
-/// #map0 = affine_map<(d0, d1, d2) -> (d0, d1, d2)>
-/// #map1 = affine_map<(d0, d1, d2) -> ()>
-/// .....
-/// %0 = vector.transfer_read %arg0[%c0, %c0, %c0, %c0], %c0_f32
-///   : memref<1x3x3x3xf32>, vector<3x3x3xf32>
-/// %1 = vector.transfer_read %arg1[%c0, %c0, %c0, %c0], %c0_f32
-///   : memref<1x3x3x3xf32>, vector<3x3x3xf32>
-/// %2 = vector.contract {indexing_maps = [#map0, #map0, #map1],
-///   iterator_types = ["reduction", "reduction", "reduction"]} %0, %1,
-///   %c0_f32 : vector<3x3x3xf32>, vector<3x3x3xf32> into f32
-/// store %2, %arg2[%c0, %c0, %c0, %c0] : memref<?x?x?x?xf32>
-/// ```
-/// where first 2 operations read input and kernel memory buffers into vectors.
-/// Subsequently, they are contracted together and the result is written to
-/// the first entry of the output buffer.
-template <typename ConvOp, int N>
-class ConvOpVectorization : public OpRewritePattern<ConvOp> {
-  using OpRewritePattern<ConvOp>::OpRewritePattern;
-  SmallVector<bool, 4> mask;
-
-public:
-  ConvOpVectorization(MLIRContext *context, const SmallVector<bool, 4> &msk)
-      : OpRewritePattern<ConvOp>(context) {
-    assert(msk.size() == N && "Mask size does not match rank");
-    this->mask = msk;
-  }
-
-  LogicalResult matchAndRewrite(ConvOp minOp,
-                                PatternRewriter &rewriter) const override;
-};
-
 /// Rewrite a TiledLoopOp with bounds/step that potentially do not divide evenly
 /// into a TiledLoopOp where the step divides the iteration space evenly,
 /// followed by another TiledLoopOp for the last (partial) iteration (if any).

mlir/lib/Dialect/Linalg/Transforms/Vectorization.cpp

@@ -1098,134 +1098,6 @@ void mlir::linalg::populatePadTensorOpVectorizationPatterns(
       patterns.getContext(), baseBenefit.getBenefit() + 1);
 }
 
-// TODO: cleanup all the convolution vectorization patterns.
-template <class ConvOp, int N>
-LogicalResult ConvOpVectorization<ConvOp, N>::matchAndRewrite(
-    ConvOp op, PatternRewriter &rewriter) const {
-  Location loc = op.getLoc();
-  MLIRContext *context = op.getContext();
-
-  OpOperand *input = op.getInputOperand(0);
-  OpOperand *kernel = op.getInputOperand(1);
-  OpOperand *output = op.getOutputOperand(0);
-  ArrayRef<int64_t> inShape = op.getShape(input);
-  ArrayRef<int64_t> kShape = op.getShape(kernel);
-
-  if (llvm::any_of(inShape, ShapedType::isDynamic) ||
-      llvm::any_of(kShape, ShapedType::isDynamic))
-    return failure();
-
-  SmallVector<AffineExpr, 4> mapping;
-  SmallVector<int64_t, 4> vectorDims;
-  // Fail to apply when the size of not vectorized dimension is not 1.
-  for (unsigned i = 0; i < N; i++) {
-    if (!mask[i] && (inShape[i] != 1 || kShape[i] != 1))
-      return failure();
-
-    if (mask[i] && inShape[i] != kShape[i])
-      return failure();
-
-    if (mask[i]) {
-      mapping.push_back(getAffineDimExpr(i, context));
-      vectorDims.push_back(inShape[i]);
-    }
-  }
-
-  int64_t rank = op.getRank(input);
-  int64_t numDims = mapping.size();
-  Type elemType = getElementTypeOrSelf(input->get());
-
-  auto map = AffineMap::get(rank, 0, mapping, context);
-  SmallVector<Value, 4> zeros(rank,
-                              rewriter.create<arith::ConstantIndexOp>(loc, 0));
-  auto vecType = VectorType::get(vectorDims, elemType);
-
-  auto inputVec = rewriter.create<vector::TransferReadOp>(
-      loc, vecType, input->get(), zeros, map);
-  auto kernelVec = rewriter.create<vector::TransferReadOp>(
-      loc, vecType, kernel->get(), zeros, map);
-
-  auto acc = rewriter.create<arith::ConstantOp>(loc, elemType,
-                                                rewriter.getZeroAttr(elemType));
-
-  std::array<AffineMap, 3> indexingMaps{
-      AffineMap::getMultiDimIdentityMap(numDims, context),
-      AffineMap::getMultiDimIdentityMap(numDims, context),
-      AffineMap::get(numDims, 0, {}, context)};
-
-  std::vector<StringRef> iteratorTypes(numDims, "reduction");
-
-  auto result = rewriter.create<vector::ContractionOp>(
-      loc, inputVec, kernelVec, acc,
-      rewriter.getAffineMapArrayAttr(indexingMaps),
-      rewriter.getStrArrayAttr(iteratorTypes));
-
-  rewriter.create<memref::StoreOp>(loc, result, output->get(),
-                                   ValueRange(zeros));
-  rewriter.eraseOp(op);
-  return success();
-}
-
-/// Inserts tiling, promotion and vectorization pattern for ConvOp
-/// conversion into corresponding pattern lists.
-template <typename ConvOp, unsigned N>
-static void populateVectorizationPatterns(
-    RewritePatternSet &tilingPatterns, RewritePatternSet &promotionPatterns,
-    RewritePatternSet &vectorizationPatterns, ArrayRef<int64_t> tileSizes) {
-  auto *context = tilingPatterns.getContext();
-  if (tileSizes.size() < N)
-    return;
-
-  constexpr static StringRef kTiledMarker = "TILED";
-  constexpr static StringRef kPromotedMarker = "PROMOTED";
-  tilingPatterns.add<LinalgTilingPattern>(
-      ConvOp::getOperationName(), context,
-      LinalgTilingOptions().setTileSizes(tileSizes),
-      LinalgTransformationFilter(ArrayRef<StringAttr>{},
-                                 StringAttr::get(context, kTiledMarker)));
-
-  promotionPatterns.add<LinalgPromotionPattern<ConvOp>>(
-      context, LinalgPromotionOptions().setUseFullTileBuffersByDefault(true),
-      LinalgTransformationFilter(StringAttr::get(context, kTiledMarker),
-                                 StringAttr::get(context, kPromotedMarker)));
-
-  SmallVector<bool, 4> mask(N);
-  int offset = tileSizes.size() - N;
-  std::transform(tileSizes.begin() + offset, tileSizes.end(), mask.begin(),
-                 [](int64_t i) -> bool { return i > 1; });
-
-  vectorizationPatterns.add<ConvOpVectorization<ConvOp, N>>(context, mask);
-}
-
-void mlir::linalg::populateConvVectorizationPatterns(
-    MLIRContext *context, SmallVectorImpl<RewritePatternSet> &patterns,
-    ArrayRef<int64_t> tileSizes) {
-  RewritePatternSet tiling(context);
-  RewritePatternSet promotion(context);
-  RewritePatternSet vectorization(context);
-  populateVectorizationPatterns<Conv1DOp, 1>(tiling, promotion, vectorization,
-                                             tileSizes);
-
-  populateVectorizationPatterns<Conv2DOp, 2>(tiling, promotion, vectorization,
-                                             tileSizes);
-
-  populateVectorizationPatterns<Conv3DOp, 3>(tiling, promotion, vectorization,
-                                             tileSizes);
-
-  populateVectorizationPatterns<Conv1DNwcWcfOp, 3>(tiling, promotion,
-                                                   vectorization, tileSizes);
-
-  populateVectorizationPatterns<Conv2DNhwcHwcfOp, 4>(tiling, promotion,
-                                                     vectorization, tileSizes);
-
-  populateVectorizationPatterns<Conv3DNdhwcDhwcfOp, 5>(
-      tiling, promotion, vectorization, tileSizes);
-
-  patterns.push_back(std::move(tiling));
-  patterns.push_back(std::move(promotion));
-  patterns.push_back(std::move(vectorization));
-}
-
 //----------------------------------------------------------------------------//
 // Forwarding patterns
 //----------------------------------------------------------------------------//

mlir/test/Conversion/LinalgToVector/linalg-to-vector.mlir

@@ -9,17 +9,6 @@
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -test-conv-vectorization="tile-sizes=1,3" -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// 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 -linalg-tile="tile-sizes=4" \
-// RUN:   -test-conv-vectorization="tile-sizes=1,3" -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// RUN: mlir-cpu-runner -e main -entry-point-result=void \
-// RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
-// RUN: | FileCheck %s
-
 func private @print_memref_f32(memref<*xf32>)
 
 // Creates and returns a 1-D buffer of size %s1 filled with the value %f

mlir/test/Integration/Dialect/Linalg/CPU/test-conv-1d-call.mlir

@@ -9,17 +9,6 @@
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -test-conv-vectorization="tile-sizes=1,1,1,3,3" -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// 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 -linalg-tile="tile-sizes=2,4" \
-// RUN:   -test-conv-vectorization="tile-sizes=1,1,1,3,3" -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// RUN: mlir-cpu-runner -e main -entry-point-result=void \
-// RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
-// RUN: | FileCheck %s
-
 func private @print_memref_f32(memref<*xf32>)
 
 // Creates and returns 3-D buffer of size (%s1, %s2, %s3) filled with the value %f

mlir/test/Integration/Dialect/Linalg/CPU/test-conv-1d-nwc-wcf-call.mlir

@@ -9,17 +9,6 @@
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -test-conv-vectorization="tile-sizes=1,1,3,3"  -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// 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 -linalg-tile="tile-sizes=2,2" \
-// RUN:   -test-conv-vectorization="tile-sizes=1,1,3,3"  -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// RUN: mlir-cpu-runner -e main -entry-point-result=void \
-// RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
-// RUN: | FileCheck %s
-
 func private @print_memref_f32(memref<*xf32>)
 
 // Creates and returns a 2-D buffer of size (%s1, %s2) filled with the value %f

mlir/test/Integration/Dialect/Linalg/CPU/test-conv-2d-call.mlir

@@ -9,17 +9,6 @@
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -test-conv-vectorization="tile-sizes=1,1,1,1,3,3,3" -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// 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 -linalg-tile="tile-sizes=2,3,3,2" \
-// RUN:   -test-conv-vectorization="tile-sizes=1,1,1,1,3,3,3" -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// RUN: mlir-cpu-runner -e main -entry-point-result=void \
-// RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
-// RUN: | FileCheck %s
-
 func private @print_memref_f32(memref<*xf32>)
 
 // Creates and returns 4-D buffer of size (%s1, %s2, %s3, %s4) filled with the value %f

mlir/test/Integration/Dialect/Linalg/CPU/test-conv-2d-nhwc-hwcf-call.mlir

@@ -9,17 +9,6 @@
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -test-conv-vectorization="tile-sizes=1,1,1,3,3,3" -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// 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 -linalg-tile="tile-sizes=2,2,2" \
-// RUN:   -test-conv-vectorization="tile-sizes=1,1,1,3,3,3" -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// RUN: mlir-cpu-runner -e main -entry-point-result=void \
-// RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
-// RUN: | FileCheck %s
-
 func private @print_memref_f32(memref<*xf32>)
 
 // Creates and returns 3-D buffer of size (%s1, %s2, %s3) filled with the value %f

mlir/test/Integration/Dialect/Linalg/CPU/test-conv-3d-call.mlir

@@ -9,17 +9,6 @@
 // RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -test-conv-vectorization="tile-sizes=1,1,1,1,1,3,3,3,3" -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// 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 -linalg-tile="tile-sizes=0,5,5,5" \
-// RUN:   -test-conv-vectorization="tile-sizes=1,1,1,1,1,3,3,3,3" -convert-linalg-to-llvm -lower-affine -convert-scf-to-std -convert-vector-to-llvm --convert-memref-to-llvm -convert-std-to-llvm -reconcile-unrealized-casts | \
-// RUN: mlir-cpu-runner -e main -entry-point-result=void \
-// RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_runner_utils%shlibext \
-// RUN: | FileCheck %s
-
 func private @print_memref_f32(memref<*xf32>)
 
 // Creates and returns 5-D buffer of size (%s1, %s2, %s3, %s4, %s5) filled with the value %f

mlir/test/Integration/Dialect/Linalg/CPU/test-conv-3d-ndhwc-dhwcf-call.mlir

@@ -1,7 +1,6 @@
 # Exclude tests from libMLIR.so
 add_mlir_library(MLIRLinalgTestPasses
   TestComprehensiveBufferize.cpp
-  TestConvVectorization.cpp
   TestLinalgCodegenStrategy.cpp
   TestLinalgDistribution.cpp
   TestLinalgElementwiseFusion.cpp