[mlir][Linalg] Retire LinalgStrategyTilePass and filter-based pattern.

[nicolasvasilache]

Summary

Context: https://discourse.llvm.org/t/psa-retire-linalg-filter-based-patterns/63785

Uses of LinalgTilingPattern::returningMatchAndRewrite are replaced by a top-level tileWithLinalgTilingOptions function that is marked obsolete and serves
as a temporary means to transition away from LinalgTilingOptions-based tiling.
LinalgTilingOptions supports too many options that have been orthogonalized with the use of the transform dialect.

Additionally, the revision introduces a transform.structured.tile_to_scf_for structured transform operation that is needed to properly tile tensor.pad
via the TilingInterface. Uses of transform.structured.tile will be deprecated and replaced by this new op.
This will achieve the deprecation of linalg::tileLinalgOp.
Context: https://discourse.llvm.org/t/psa-retire-tileandfuselinalgops-method/63850

In the process of transitioning, tests that were performing tile and distribute on tensors are retired: transformations should be orthogonalized better in the future.
In particular, tiling to specific loop types and tileAndDistribute behavior are not available via the transform ops.
The behavior is still available as part of the tileWithLinalgTilingOptions method to allow downstream clients to transition without breakages but is meant to be retired soon.

As more tests are ported to the transform dialect, it became necessary to introduce a test-transform-dialect-erase-schedule-pass to discard the transform specification
once applied so that e2e lowering and execution is possible.

Lastly, a number of redundant tests that were testing composition of patterns are retired as they are available with a better mechanism via the transform dialect.

[nicolasvasilache]

Update commit message.

[nicolasvasilache]

Update commit message.

[nicolasvasilache]

s/tileAndX/tileWithLinalgTilingOptions

[nicolasvasilache]

Rebase.

[ftynse]

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

1038

There is a bunch of unrelated rewrapping that doesn't seem necessary. The lines were 80 cols wide, is your editor set to 79?

This feels spurious and pollutes the change history.

mlir/test/Dialect/Linalg/transform-patterns.mlir

1

Is the test-patterns part still relevant?

mlir/test/lib/Dialect/Transform/TestTransformDialectInterpreter.cpp

77

::mlir shouldn't be necessary here.

[nicolasvasilache]

mlir/test/Dialect/Linalg/transform-patterns.mlir

1

Yes it is, as it triggers some lowering of linalg to loops that we do not control with the transform dialect yet.

I'll drop that in some future update.

[nicolasvasilache]

Address comments

[nicolasvasilache]

Rebase.

[nicolasvasilache]

Rebase.

mlir/include/mlir/Dialect/Linalg/Passes.h

 std::unique_ptr<Pass> createLinalgFoldUnitExtentDimsPass();
 
 std::unique_ptr<Pass> createLinalgElementwiseOpFusionPass();
 std::unique_ptr<Pass> createFoldReshapeOpsByLinearizationPass();
 
 std::unique_ptr<Pass> createLinalgNamedOpConversionPass();
 
 std::unique_ptr<OperationPass<func::FuncOp>>
-createLinalgTilingPass(ArrayRef<int64_t> tileSizes = {},
-                       linalg::LinalgTilingLoopType loopType =
-                           linalg::LinalgTilingLoopType::Loops);
-
-std::unique_ptr<OperationPass<func::FuncOp>>
 createLinalgInlineScalarOperandsPass();
 
 /// Create a pass to convert Linalg operations to scf.for loops and
 /// memref.load/memref.store accesses.
 std::unique_ptr<OperationPass<func::FuncOp>> createConvertLinalgToLoopsPass();
 
 /// Create a pass to convert Linalg operations to scf.parallel loops and
 /// memref.load/memref.store accesses.

mlir/include/mlir/Dialect/Linalg/Passes.td

def LinalgBufferize : Pass<"linalg-bufferize", "func::FuncOp"> {
   let dependentDialects = [
     "AffineDialect",
     "bufferization::BufferizationDialect",
     "linalg::LinalgDialect",
     "memref::MemRefDialect",
   ];
 }
 
-def LinalgTilingPass : Pass<"linalg-tile", "func::FuncOp"> {
-  let summary = "Tile operations in the linalg dialect";
-  let constructor = "mlir::createLinalgTilingPass()";
-  let dependentDialects = [
-    "AffineDialect",
-    "linalg::LinalgDialect",
-    "memref::MemRefDialect",
-    "scf::SCFDialect"
-  ];
-  let options = [
-    ListOption<"tileSizes", "tile-sizes", "int64_t", "Tile sizes">,
-    Option<"loopType", "loop-type", "std::string", /*default=*/"\"for\"",
-           "Specify the type of loops to generate: for, parallel">
-  ];
-}
-
 def LinalgGeneralization : Pass<"linalg-generalize-named-ops", "func::FuncOp"> {
   let summary = "Convert named ops into generic ops";
   let constructor = "mlir::createLinalgGeneralizationPass()";
   let dependentDialects = ["linalg::LinalgDialect"];
 }
 
 def LinalgDetensorize : Pass<"linalg-detensorize", ""> {
   let summary = "Detensorize linalg ops";
def LinalgDetensorize : Pass<"linalg-detensorize", ""> {
   let options = [
     Option<"aggressiveMode", "aggressive-mode", "bool", /*default=*/"false",
            "Detensorize all ops that qualify for detensoring along with branch"
            " operands and basic-block arguments.">
 
   ];
 }
 
-def LinalgStrategyTilePass
-    : Pass<"linalg-strategy-tile-pass", "func::FuncOp"> {
-  let summary = "Configurable pass to apply pattern-based linalg tiling.";
-  let constructor = "mlir::createLinalgStrategyTilePass()";
-  let dependentDialects = ["linalg::LinalgDialect"];
-  let options = [
-    Option<"anchorFuncName", "anchor-func", "std::string", /*default=*/"",
-      "Which func op is the anchor to latch on.">,
-    Option<"anchorOpName", "anchor-op", "std::string", /*default=*/"",
-      "Which linalg op within the func is the anchor to latch on.">,
-  ];
-}
-
 def LinalgStrategyRemoveMarkersPass
     : Pass<"linalg-strategy-remove-markers-pass", "func::FuncOp"> {
   let summary = "Cleanup pass that drops markers.";
   let constructor = "mlir::createLinalgStrategyRemoveMarkersPass()";
   let dependentDialects = ["linalg::LinalgDialect"];
   let options = [
     Option<"anchorFuncName", "anchor-func", "std::string", /*default=*/"",
       "Which func op is the anchor to latch on.">,
   ];
 }
 
 #endif // MLIR_DIALECT_LINALG_PASSES

mlir/include/mlir/Dialect/Linalg/TransformOps/LinalgTransformOps.td

::mlir::DiagnosedSilenceableFailure apply(
         ::mlir::transform::TransformResults &transformResults,
         ::mlir::transform::TransformState &state);
 
     ::llvm::SmallVector<::mlir::OpFoldResult> getMixedNumThreads();
     ::llvm::SmallVector<::mlir::OpFoldResult> getMixedTileSizes();
   }];
 }
 
+def TileToScfForOp : Op<Transform_Dialect, "structured.tile_to_scf_for",
+       [DeclareOpInterfaceMethods<TransformOpInterface>,
+        DeclareOpInterfaceMethods<MemoryEffectsOpInterface>]> {
+  let description = [{
+    Indicates that the given `target` op should be tiled with the given sizes.
+    This transform generates a loop nest with a smaller ("tiled") target
+    operation in its body. The target must implement TilingInterface.
+
+    Tile sizes may be known at transformation time, in which case they are
+    expected to be provided in the `static_size` attribute, or not, in which
+    case the tile value must be computed by the payload IR and the handle to the
+    operation computing it must be provided through `dynamic_sizes`. When the
+    sizes are not known statically, the corresponding entry in the
+    `static_sizes` attribute must be set to `ShapedType::kDynamicSize`. Only
+    the dynamic sizes must be provided in `dynamic_sizes`, i.e., there should
+    be as many handles as `ShapedType::kDynamicSize` values in the
+    `static_sizes` attribute. A static size of `0` indicates that the dimension
+    should not be tiled. No loop will be generated for such dimensions. If all
+    tile sizes are `0`, this transform is effectively a no-op.
+
+    This op returns handles to the tiled op (in the generated loop nest) and the
+    generated loops. The number of loops is the number of tile sizes that are
+    statically known to be non-zero.
+
+    #### Return modes
+
+    On success, the resulting handles are associated with co-indexed lists of
+    tiled operations and loops around them.
+
+    This operation only supports TilingInterface ops and produces a silenceable
+    failure if the input contains any non-TilingInterface ops. The ops preceding
+    it in the list associated with the `target` handle will have been tiled.
+
+    This operation produces a silenceable failure if the `dynamic_sizes` handles
+    are associated with lists of payload operations of a size different than
+    that of the list associated with the `target` handle.
+
+    If the internal implementation of tiling for any of the operations fails,
+    produces a definite failure.
+  }];
+
+  let arguments = (ins PDL_Operation:$target,
+                   Variadic<PDL_Operation>:$dynamic_sizes,
+                   DefaultValuedAttr<I64ArrayAttr, "{}">:$static_sizes,
+                   DefaultValuedAttr<I64ArrayAttr, "{}">:$interchange);
+  let results = (outs PDL_Operation:$tiled_linalg_op,
+                      Variadic<PDL_Operation>:$loops);
+
+  let hasCustomAssemblyFormat = 1;
+
+  let extraClassDeclaration = [{
+    /// Returns the list of tile sizes, which may be static (Attribute) or
+    /// dynamic (Value).
+    SmallVector<OpFoldResult> getMixedSizes();
+  }];
+}
+
 def VectorizeOp : Op<Transform_Dialect, "structured.vectorize",
     [FunctionalStyleTransformOpTrait, MemoryEffectsOpInterface,
      TransformEachOpTrait, TransformOpInterface]> {
   let description = [{
     Indicates that the given `target` op all the ops it contains should be
     vectorized with the configuration specified by the attributes of this op.
     This vectorization only handles structured ops that operate on shaped types
     and does not vectorize loops or straight-line. Internally, it applies a

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

struct Transformation {
   explicit Transformation(LinalgTransformationFilter::FilterFunction f)
       : filter(std::move(f)) {}
   virtual ~Transformation() = default;
   virtual void addToPassPipeline(OpPassManager &pm,
                                  LinalgTransformationFilter m) const = 0;
   LinalgTransformationFilter::FilterFunction filter = nullptr;
 };
 
-/// Represent one application of LinalgStrategyTilePass.
-struct Tile : public Transformation {
-  Tile(StringRef name, linalg::LinalgTilingOptions options,
-       LinalgTransformationFilter::FilterFunction f = nullptr)
-      : Transformation(std::move(f)), opName(name),
-        options(std::move(options)) {}
-
-  void addToPassPipeline(OpPassManager &pm,
-                         LinalgTransformationFilter m) const override {
-    pm.addPass(createLinalgStrategyTilePass(opName, options, m));
-  }
-
-private:
-  std::string opName;
-  linalg::LinalgTilingOptions options;
-};
-
 /// Codegen strategy controls how a Linalg op is progressively lowered.
 struct CodegenStrategy {
-  /// Append a pattern to add a level of tiling for Op `opName` with tiling
-  /// `options`.
-  CodegenStrategy &
-  tile(StringRef opName, const linalg::LinalgTilingOptions &options,
-       const LinalgTransformationFilter::FilterFunction &f = nullptr) {
-    transformationSequence.emplace_back(
-        std::make_unique<Tile>(opName, options, f));
-    return *this;
-  }
-  /// Conditionally append a pattern to add a level of tiling for
-  /// `LinalgOpType` with tiling `options`.
-  CodegenStrategy &
-  tileIf(bool b, StringRef opName, linalg::LinalgTilingOptions options,
-         LinalgTransformationFilter::FilterFunction f = nullptr) {
-    return b ? tile(opName, std::move(options), std::move(f)) : *this;
-  }
   /// Configure the post staged-patterns global enabling passes options.
   CodegenStrategy &
   setVectorTransferToSCFOptions(LinalgEnablingOptions options) {
     linalgEnablingOptions = options;
     return *this;
   }
 
   /// Apply the transformation patterns in sequence with cleanup

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

 };
 
 /// Canonicalization patterns relevant to apply after tiling patterns. These are
 /// applied automatically by the tiling pass but need to be applied manually
 /// when tiling is called programmatically.
 RewritePatternSet getLinalgTilingCanonicalizationPatterns(MLIRContext *ctx);
 void populateLinalgTilingCanonicalizationPatterns(RewritePatternSet &patterns);
 
-///
-/// Linalg tiling pattern.
-///
-/// Apply the `tiling` transformation as a pattern.
-/// `filter` controls LinalgTransformMarker matching and update when specified.
-/// See `tiling` for more details.
-// TODO: TiledOpInterface
-struct LinalgTilingPattern : public OpInterfaceRewritePattern<LinalgOp> {
-  /// Construct a generic pattern applied to all LinalgOp that verify `filter`.
-  LinalgTilingPattern(
-      MLIRContext *context, LinalgTilingOptions options,
-      LinalgTransformationFilter f = LinalgTransformationFilter(),
-      PatternBenefit benefit = 1);
-
-  /// Construct a pattern specifically applied to `opName`.
-  LinalgTilingPattern(
-      StringRef opName, MLIRContext *context, LinalgTilingOptions options,
-      LinalgTransformationFilter f = LinalgTransformationFilter(),
-      PatternBenefit benefit = 1);
-
-  /// `matchAndRewrite` implementation that returns the significant transformed
-  /// pieces of IR.
-  FailureOr<TiledLinalgOp>
-  returningMatchAndRewrite(LinalgOp op, PatternRewriter &rewriter) const;
-
-  LogicalResult matchAndRewrite(LinalgOp op,
-                                PatternRewriter &rewriter) const override {
-    return returningMatchAndRewrite(op, rewriter);
-  }
-
-private:
-  /// LinalgTransformMarker handles special attribute manipulations.
-  LinalgTransformationFilter filter;
-  /// Options to control tiling;
-  LinalgTilingOptions options;
-};
+/// Perform tiling using LinalgTilingOptions.
+/// Note: this is on a path to deprecation that only works on LinalgOp.
+/// Clients should favor using `tileUsingSCFForOp`  that more generally works on
+/// TilingInterface.
+tileWithLinalgTilingOptions(RewriterBase &rewriter, LinalgOp op,
+                            const LinalgTilingOptions &options);
 
 ///
 /// Linalg padding pattern.
 ///
 /// Apply the `padding` transformation as a pattern.
-/// `filter` controls LinalgTransformMarker matching and update when specified.
 /// See `padding` for more details.
 struct LinalgPaddingPattern : public OpInterfaceRewritePattern<LinalgOp> {
-  /// Construct a generic pattern applied to all LinalgOp that verify `filter`.
   LinalgPaddingPattern(MLIRContext *context,
                        LinalgPaddingOptions options = LinalgPaddingOptions(),
                        PatternBenefit benefit = 1);
 
-  /// `matchAndRewrite` implementation that returns the significant transformed
-  /// pieces of IR.
+  /// `matchAndRewrite` implementation that returns the significant
+  /// transformed pieces of IR.
   FailureOr<LinalgOp> returningMatchAndRewrite(LinalgOp op,
                                                PatternRewriter &rewriter) const;
 
   LogicalResult matchAndRewrite(LinalgOp op,
                                 PatternRewriter &rewriter) const override {
     return returningMatchAndRewrite(op, rewriter);
   }
 

 /// Populates `patterns` with patterns to convert spec-generated named ops to
 /// linalg.generic ops.
 void populateLinalgNamedOpsGeneralizationPatterns(
     RewritePatternSet &patterns,
     const LinalgTransformationFilter &filter = LinalgTransformationFilter());
 
 /// Linalg decompose convolutions patterns
 
-/// Populates patterns to decompose high-D convolution ops into low-D ones. This
-/// is a step in progressive lowering for convolution ops, afterwards we can
-/// vectorize the low-D convolution ops.
+/// Populates patterns to decompose high-D convolution ops into low-D ones.
+/// This is a step in progressive lowering for convolution ops, afterwards we
+/// can vectorize the low-D convolution ops.
 void populateDecomposeConvolutionPatterns(RewritePatternSet &patterns,
                                           PatternBenefit benefit = 1);
 
 //===----------------------------------------------------------------------===//
 // Op-specific patterns.
 //===----------------------------------------------------------------------===//
 
 /// tensor::PadOp is not canonicalized away yet, so we provide a transformation
 /// to `linalg.generic`.
 struct PadOpTransformationPattern : public OpRewritePattern<tensor::PadOp> {
   using OpRewritePattern<tensor::PadOp>::OpRewritePattern;
 
   LogicalResult matchAndRewrite(tensor::PadOp padOp,
                                 PatternRewriter &rewriter) const override;
 };
 
 /// Pad the iterator dimensions `paddingDimensions` of all `opToPad` operands to
 /// a static bounding box. Use `paddingValues` and `packPaddings` to set padding
 /// value and nofold attribute of the created tensor::PadOps, respectively.
 /// Update `paddedOp` to the cloned operation with statically shaped
-/// `paddingDimensions` and return the extracted dynamically shaped results. If
-/// padding fails, return failure.
+/// `paddingDimensions` and return the extracted dynamically shaped results.
+/// If padding fails, return failure.
 FailureOr<SmallVector<Value>>
 rewriteAsPaddedOp(OpBuilder &b, LinalgOp opToPad,
                   ArrayRef<int64_t> paddingDimensions,
                   ArrayRef<Attribute> paddingValues,
                   ArrayRef<bool> packPaddings, LinalgOp &paddedOp);
 
 using OptimizeCopyFn =
     std::function<LogicalResult(PatternRewriter &, tensor::PadOp, Value)>;

 ///    [unchanged] %alloc = ...
 ///    [unchanged] [optional] %view = memref.view %alloc ...
 ///    [unchanged] %subView = subview %allocOrView...
 ///    ...
 ///    vector.transfer_write %..., %out[...]
 /// ```
 /// Where there is no interleaved use between transfer_write and memref.copy.
 /// This is a custom rewrite to forward partial writes to vector.transfer_write.
 struct LinalgCopyVTWForwardingPattern
     : public OpRewritePattern<vector::TransferWriteOp> {
   using OpRewritePattern<vector::TransferWriteOp>::OpRewritePattern;
 
   LogicalResult matchAndRewrite(vector::TransferWriteOp xferOp,
                                 PatternRewriter &rewriter) const override;
 };
 
 //===----------------------------------------------------------------------===//

 template <>
 class VectorizationPatterns<> {
 public:
   static void insert(RewritePatternSet &patterns,
                      const LinalgVectorizationOptions &options,
                      const LinalgTransformationFilter &f) {}
 };
 
-template <typename... OpTypes>
-class TilingPatterns;
-
-template <>
-class TilingPatterns<> {
-public:
-  static void insert(RewritePatternSet &patterns,
-                     const LinalgTilingOptions &options,
-                     const LinalgTransformationFilter &f) {}
-};
-
-template <typename OpTy, typename... OpTypes>
-class TilingPatterns<OpTy, OpTypes...> {
-public:
-  static void insert(RewritePatternSet &patterns,
-                     const LinalgTilingOptions &options,
-                     const LinalgTransformationFilter &f) {
-    patterns.add<LinalgTilingPattern>(OpTy::getOperationName(),
-                                      patterns.getContext(), options, f);
-    TilingPatterns<OpTypes...>::insert(patterns, options, f);
-  }
-};
-
 /// Split Reduction options.
 struct SplitReductionOptions {
   // Ratio used to split the reduction dimension.  If the ratio is <= 1, nothing
   // will be done.
   int64_t ratio = 0;
   // Index where the extra dimension is added to the intermediate tensor shape.
   unsigned index = 0;
   // If the inner dimension after splitting is parallel or reduction.

 void populateSplitReductionPattern(
     RewritePatternSet &patterns,
     const ControlSplitReductionFn &controlSplitReductionFn,
     const LinalgTransformationFilter &f = LinalgTransformationFilter(),
     bool useAlloc = false);
 
 /// Apply transformation to split the single linalg op reduction into a parallel
 /// and reduction dimension. Then create a new linalg.generic op doing the rest
-/// of the reduction. Return the new linalg op with an extra parallel dimension
-/// or failure if the transformation didn't happen.
+/// of the reduction.
+/// Return the new linalg op with an extra parallel dimension or failure if the
+/// transformation didn't happen.
+///
 /// Example:
 /// ```
 ///  %r = linalg.generic {indexing_maps = [affine_map<(d0) -> (d0)>,
 ///                                        affine_map<(d0) -> ()>],
 ///       iterator_types = ["reduction"]}
 ///  ins(%in : tensor<32xf32>)
 ///  outs(%out : tensor<f32>) {
 ///  ^bb0(%arg1: f32, %arg2: f32):

 ///  %cst = arith.constant 0.000000e+00 : f32
 ///  %1 = linalg.fill ins(%cst : f32) outs(%0 : tensor<16x32x64xf32>) ->
 ///     tensor<16x32x64xf32>
 ///  %2 = tensor.empty [64, 4] : tensor<64x4xi1>
 ///
 ///  %3 = linalg.generic {indexing_maps = [#map0, #map1, #map2, #map3],
 ///    iterator_types = ["parallel", "parallel", "parallel", "reduction"]}
 ///    ins(%A, %B, %2 : tensor<16x256xf32>, tensor<256x32xf32>, tensor<64x4xi1>)
-///    outs(%1 : tensor<16x32x64xf32>) {
+///   outs(%1 : tensor<16x32x64xf32>) {
 ///      ^bb0(%arg3: f32, %arg4: f32, %arg5: i1, %arg6: f32):
 ///        %5 = arith.mulf %arg3, %arg4 : f32
 ///        %6 = arith.addf %arg6, %5 : f32
 ///        linalg.yield %6 : f32
 ///  } -> tensor<16x32x64xf32>
 ///
 ///  %4 = linalg.generic {indexing_maps = [#map4, #map5],
 ///    iterator_types = ["parallel", "parallel", "reduction"]}

mlir/lib/Dialect/Linalg/TransformOps/LinalgTransformOps.cpp

 static SmallVector<unsigned> extractUIntArray(ArrayAttr attr) {
   SmallVector<unsigned> result;
   result.reserve(attr.size());
   for (APInt value : attr.getAsValueRange<IntegerAttr>())
     result.push_back(value.getZExtValue());
   return result;
 }
 
+/// Extracts a vector of int64_t from an array attribute. Asserts if the
+/// attribute contains values other than integers.
+static SmallVector<int64_t> extractI64Array(ArrayAttr attr) {
+  SmallVector<int64_t> result;
+  result.reserve(attr.size());
+  for (APInt value : attr.getAsValueRange<IntegerAttr>())
+    result.push_back(value.getSExtValue());
+  return result;
+}
+
 namespace {
 /// A simple pattern rewriter that implements no special logic.
 class SimpleRewriter : public PatternRewriter {
 public:
   SimpleRewriter(MLIRContext *context) : PatternRewriter(context) {}
 };
 } // namespace
 

 transform::ScalarizeOp::applyToOne(linalg::LinalgOp target,
                                    SmallVectorImpl<Operation *> &results,
                                    transform::TransformState &state) {
   LinalgTilingOptions tilingOptions;
   tilingOptions.scalarizeDynamicDims();
   // Tiling with "scalarize_dyn_dims" actually sets the same lambda as the
   // tile sizes and asserts that it is not already set.
   SmallVector<int64_t> emptyTileSizes;
-  LinalgTilingPattern pattern(getContext(), tilingOptions);
   SimpleRewriter rewriter(getContext());
   rewriter.setInsertionPoint(target);
   FailureOr<TiledLinalgOp> result =
-      pattern.returningMatchAndRewrite(target, rewriter);
+      tileWithLinalgTilingOptions(rewriter, target, tilingOptions);
   if (failed(result))
     return DiagnosedSilenceableFailure(reportUnknownTransformError(target));
 
   results.push_back(result->op);
   return DiagnosedSilenceableFailure(success());
 }
 
 //===----------------------------------------------------------------------===//

 
 //===----------------------------------------------------------------------===//
 // TileOp
 //===----------------------------------------------------------------------===//
 
 DiagnosedSilenceableFailure
 transform::TileOp::apply(TransformResults &transformResults,
                          TransformState &state) {
-  LinalgTilingOptions tilingOptions;
   SmallVector<int64_t> tileSizes = extractFromI64ArrayAttr(getStaticSizes());
 
   ArrayRef<Operation *> targets = state.getPayloadOps(getTarget());
   SmallVector<ArrayRef<Operation *>> dynamicSizeProducers;
   dynamicSizeProducers.reserve(getDynamicSizes().size());
   for (Value dynamicSizeProducerHandle : getDynamicSizes()) {
     dynamicSizeProducers.push_back(
         state.getPayloadOps(dynamicSizeProducerHandle));
for (auto &en : llvm::enumerate(targets)) {
     auto linalgOp = dyn_cast<LinalgOp>(en.value());
     if (!linalgOp) {
       DiagnosedSilenceableFailure diag = emitSilenceableError()
                                          << "only linalg ops are supported";
       diag.attachNote(en.value()->getLoc()) << "target op";
       return diag;
     }
 
+    LinalgTilingOptions tilingOptions;
     unsigned index = en.index();
     if (!tileSizes.empty()) {
       tilingOptions.setTileSizeComputationFunction(
           [&, index](OpBuilder &b, Operation *) {
             SmallVector<Value, 4> sizes;
             sizes.reserve(tileSizes.size());
             unsigned dynamicIdx = 0;
             for (OpFoldResult ofr : getMixedSizes()) {
               if (auto attr = ofr.dyn_cast<Attribute>()) {
                 sizes.push_back(b.create<arith::ConstantIndexOp>(
                     getLoc(), attr.cast<IntegerAttr>().getInt()));
               } else {
                 sizes.push_back(
                     dynamicSizeProducers[dynamicIdx++][index]->getResult(0));
               }
             }
             return sizes;
           });
     }
 
     tilingOptions.setInterchange(extractUIntArray(getInterchange()));
-    LinalgTilingPattern pattern(getContext(), tilingOptions);
     SimpleRewriter rewriter(linalgOp.getContext());
     FailureOr<TiledLinalgOp> tiledOp =
-        pattern.returningMatchAndRewrite(linalgOp, rewriter);
+        tileWithLinalgTilingOptions(rewriter, linalgOp, tilingOptions);
     if (failed(tiledOp))
       return DiagnosedSilenceableFailure::definiteFailure();
 
     tiled.push_back(tiledOp->op);
     for (const auto &en2 : llvm::enumerate(tiledOp->loops))
       loops[en2.index()].push_back(en2.value());
   }
 

 
 LogicalResult TileToForeachThreadOp::verify() {
   if (getMixedNumThreads().empty() == getMixedTileSizes().empty())
     return emitOpError("either num_threads or tile_sizes must be specified");
   return success();
 }
 
 //===----------------------------------------------------------------------===//
+// TileToScfForOp
+//===----------------------------------------------------------------------===//
+
+DiagnosedSilenceableFailure
+transform::TileToScfForOp::apply(TransformResults &transformResults,
+                                 TransformState &state) {
+  SmallVector<int64_t> tileSizes = extractFromI64ArrayAttr(getStaticSizes());
+
+  ArrayRef<Operation *> targets = state.getPayloadOps(getTarget());
+  SmallVector<ArrayRef<Operation *>> dynamicSizeProducers;
+  dynamicSizeProducers.reserve(getDynamicSizes().size());
+  for (Value dynamicSizeProducerHandle : getDynamicSizes()) {
+    dynamicSizeProducers.push_back(
+        state.getPayloadOps(dynamicSizeProducerHandle));
+
+    if (dynamicSizeProducers.back().size() != targets.size()) {
+      DiagnosedSilenceableFailure diag =
+          emitSilenceableError()
+          << "expected as many dynamic size-producing operations ("
+          << dynamicSizeProducers.back().size() << ") as target ops ("
+          << targets.size() << ")";
+      diag.attachNote(dynamicSizeProducerHandle.getLoc()) << "for this handle";
+      return diag;
+    }
+
+    for (Operation *op : dynamicSizeProducers.back()) {
+      if (op->getNumResults() == 1 &&
+          op->getResult(0).getType().isa<IndexType>())
+        continue;
+      DiagnosedSilenceableFailure diag =
+          emitSilenceableError() << "expected sizes to be produced by ops "
+                                    "with a single index-type result";
+      diag.attachNote(op->getLoc()) << "size producer op";
+      diag.attachNote(dynamicSizeProducerHandle.getLoc()) << "for this handle";
+      return diag;
+    }
+  }
+
+  SmallVector<Operation *> tiled;
+  SmallVector<SmallVector<Operation *, 4>, 4> loops;
+  loops.resize(getLoops().size());
+  for (auto &en : llvm::enumerate(targets)) {
+    auto tilingInterfaceOp = dyn_cast<TilingInterface>(en.value());
+    if (!tilingInterfaceOp) {
+      DiagnosedSilenceableFailure diag =
+          emitSilenceableError() << "only TilingInterface ops are supported";
+      diag.attachNote(en.value()->getLoc()) << "target op";
+      return diag;
+    }
+
+    scf::SCFTilingOptions tilingOptions;
+    unsigned index = en.index();
+    if (!tileSizes.empty()) {
+      tilingOptions.setTileSizeComputationFunction(
+          [&, index](OpBuilder &b, Operation *) {
+            SmallVector<Value, 4> sizes;
+            sizes.reserve(tileSizes.size());
+            unsigned dynamicIdx = 0;
+            for (OpFoldResult ofr : getMixedSizes()) {
+              if (auto attr = ofr.dyn_cast<Attribute>()) {
+                sizes.push_back(b.create<arith::ConstantIndexOp>(
+                    getLoc(), attr.cast<IntegerAttr>().getInt()));
+              } else {
+                sizes.push_back(
+                    dynamicSizeProducers[dynamicIdx++][index]->getResult(0));
+              }
+            }
+            return sizes;
+          });
+    }
+
+    tilingOptions.setInterchange(extractI64Array(getInterchange()));
+    SimpleRewriter rewriter(tilingInterfaceOp.getContext());
+    FailureOr<scf::SCFTilingResult> tilingResult =
+        tileUsingSCFForOp(rewriter, tilingInterfaceOp, tilingOptions);
+    if (failed(tilingResult))
+      return DiagnosedSilenceableFailure::definiteFailure();
+
+    rewriter.replaceOp(tilingInterfaceOp, tilingResult->replacements);
+
+    tiled.push_back(tilingResult->tiledOp);
+    for (const auto &en2 : llvm::enumerate(tilingResult->loops))
+      loops[en2.index()].push_back(en2.value());
+  }
+
+  transformResults.set(getTiledLinalgOp().cast<OpResult>(), tiled);
+  for (const auto &en : llvm::enumerate(loops))
+    transformResults.set(getLoops()[en.index()].cast<OpResult>(), en.value());
+
+  return DiagnosedSilenceableFailure::success();
+}
+
+SmallVector<OpFoldResult> transform::TileToScfForOp::getMixedSizes() {
+  ValueRange dynamic = getDynamicSizes();
+  SmallVector<int64_t> tileSizes = extractFromI64ArrayAttr(getStaticSizes());
+  SmallVector<OpFoldResult> results;
+  results.reserve(tileSizes.size());
+  unsigned dynamicPos = 0;
+  Builder builder(getContext());
+  for (int64_t size : tileSizes) {
+    if (size == ShapedType::kDynamicSize) {
+      results.push_back(dynamic[dynamicPos++]);
+    } else {
+      results.push_back(builder.getIndexAttr(size));
+    }
+  }
+  return results;
+}
+
+ParseResult transform::TileToScfForOp::parse(OpAsmParser &parser,
+                                             OperationState &result) {
+  OpAsmParser::UnresolvedOperand target;
+  SmallVector<OpAsmParser::UnresolvedOperand> dynamicSizes;
+  ArrayAttr staticSizes;
+  auto pdlOperationType = pdl::OperationType::get(parser.getContext());
+  if (parser.parseOperand(target) ||
+      parser.resolveOperand(target, pdlOperationType, result.operands) ||
+      parseDynamicIndexList(parser, dynamicSizes, staticSizes,
+                            ShapedType::kDynamicSize) ||
+      parser.resolveOperands(dynamicSizes, pdlOperationType, result.operands) ||
+      parser.parseOptionalAttrDict(result.attributes))
+    return ParseResult::failure();
+
+  result.addAttribute(getStaticSizesAttrName(result.name), staticSizes);
+  size_t numExpectedLoops =
+      staticSizes.size() - llvm::count(extractFromI64ArrayAttr(staticSizes), 0);
+  result.addTypes(SmallVector<Type>(numExpectedLoops + 1, pdlOperationType));
+  return success();
+}
+
+void TileToScfForOp::print(OpAsmPrinter &p) {
+  p << ' ' << getTarget();
+  printDynamicIndexList(p, getOperation(), getDynamicSizes(), getStaticSizes(),
+                        ShapedType::kDynamicSize);
+  p.printOptionalAttrDict((*this)->getAttrs(), {getStaticSizesAttrName()});
+}
+
+void transform::TileToScfForOp::getEffects(
+    SmallVectorImpl<MemoryEffects::EffectInstance> &effects) {
+  consumesHandle(getTarget(), effects);
+  onlyReadsHandle(getDynamicSizes(), effects);
+  producesHandle(getTiledLinalgOp(), effects);
+  producesHandle(getLoops(), effects);
+  modifiesPayload(effects);
+}
+
+//===----------------------------------------------------------------------===//
 // VectorizeOp
 //===----------------------------------------------------------------------===//
 
 namespace {
 /// This is an helper only to call vectorize via a pattern inside of
 /// VectorizeOp::applyToOne.
 struct VectorizationPattern : public RewritePattern {
   explicit VectorizationPattern(MLIRContext *context)

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

 } // namespace mlir
 
 using namespace mlir;
 using namespace mlir::vector;
 using namespace linalg;
 
 namespace {
 
-/// Configurable pass to apply pattern-based linalg tiling.
-struct LinalgStrategyTilePass
-    : public impl::LinalgStrategyTilePassBase<LinalgStrategyTilePass> {
-
-  LinalgStrategyTilePass() = default;
-
-  LinalgStrategyTilePass(StringRef opName,
-                         mlir::linalg::LinalgTilingOptions opt,
-                         LinalgTransformationFilter filt)
-      : options(std::move(opt)), filter(std::move(filt)) {
-    this->anchorOpName.setValue(opName.str());
-  }
-
-  void runOnOperation() override {
-    auto funcOp = getOperation();
-    if (!anchorFuncName.empty() && funcOp.getName() != anchorFuncName)
-      return;
-
-    MLIRContext *ctx = funcOp.getContext();
-    RewritePatternSet tilingPattern(ctx);
-    if (!anchorOpName.empty())
-      tilingPattern.add<LinalgTilingPattern>(anchorOpName, ctx, options,
-                                             filter);
-    else
-      tilingPattern.add<LinalgTilingPattern>(ctx, options, filter);
-    if (anchorOpName == tensor::PadOp::getOperationName())
-      populatePadTensorTilingPatterns(tilingPattern, options);
-    (void)applyPatternsAndFoldGreedily(funcOp, std::move(tilingPattern));
-  }
-
-  mlir::linalg::LinalgTilingOptions options;
-  LinalgTransformationFilter filter;
-};
-
 /// Configurable pass to lower vector operations.
 struct LinalgStrategyRemoveMarkersPass
     : public impl::LinalgStrategyRemoveMarkersPassBase<
           LinalgStrategyRemoveMarkersPass> {
 
   void runOnOperation() override {
     auto funcOp = getOperation();
     if (!anchorFuncName.empty() && funcOp.getName() != anchorFuncName)
       return;
     funcOp.walk([](LinalgOp op) {
       op->removeAttr(LinalgTransforms::kLinalgTransformMarker);
     });
   }
 };
 } // namespace
 
-/// Create a LinalgStrategyTilePass.
-std::unique_ptr<OperationPass<func::FuncOp>>
-mlir::createLinalgStrategyTilePass(StringRef opName,
-                                   const LinalgTilingOptions &opt,
-                                   const LinalgTransformationFilter &filter) {
-  return std::make_unique<LinalgStrategyTilePass>(opName, opt, filter);
-}
-
 /// Create a LinalgStrategyRemoveMarkersPass.
 std::unique_ptr<OperationPass<func::FuncOp>>
 mlir::createLinalgStrategyRemoveMarkersPass() {
   return std::make_unique<LinalgStrategyRemoveMarkersPass>();
 }

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

void mlir::linalg::populateLinalgTilingCanonicalizationPatterns(
   ctx->getLoadedDialect<LinalgDialect>()->getCanonicalizationPatterns(patterns);
 
   CanonicalizationPatternList<
 #define GET_OP_LIST
 #include "mlir/Dialect/Linalg/IR/LinalgStructuredOps.cpp.inc"
       >::insert(patterns);
 }
 
-/// Populate the given list with patterns that apply Linalg tiling.
-static void insertTilingPatterns(RewritePatternSet &patterns,
-                                 const LinalgTilingOptions &options) {
-  auto *ctx = patterns.getContext();
-  LinalgTransformationFilter f(ArrayRef<StringAttr>{},
-                               StringAttr::get(ctx, "tiled"));
-  TilingPatterns<GenericOp,
-#define GET_OP_LIST
-#include "mlir/Dialect/Linalg/IR/LinalgStructuredOps.cpp.inc"
-                 >::insert(patterns, options, f);
-  patterns.add<PadOpTilingPattern>(ctx, options);
-}
-
 void mlir::linalg::populatePadTensorTilingPatterns(
     RewritePatternSet &patterns, const LinalgTilingOptions &options) {
   auto *ctx = patterns.getContext();
   patterns.add<PadOpTilingPattern>(ctx, options);
 }
-
-static void applyExtractSliceOfPadTensorSwapPattern(func::FuncOp funcOp) {
-  MLIRContext *ctx = funcOp.getContext();
-  RewritePatternSet patterns(ctx);
-  patterns.add<ExtractSliceOfPadTensorSwapPattern>(patterns.getContext());
-  (void)applyPatternsAndFoldGreedily(funcOp, std::move(patterns));
-  (void)applyPatternsAndFoldGreedily(
-      funcOp, getLinalgTilingCanonicalizationPatterns(ctx));
-}
-
-namespace {
-struct LinalgTilingPass : public impl::LinalgTilingPassBase<LinalgTilingPass> {
-  LinalgTilingPass() = default;
-  LinalgTilingPass(ArrayRef<int64_t> tileSizes, LinalgTilingLoopType loopType) {
-    this->tileSizes = tileSizes;
-    this->loopType = "";
-    this->loopTypeEnum = loopType;
-  }
-
-  void runOnOperation() override {
-    func::FuncOp funcOp = getOperation();
-    LinalgTilingLoopType type =
-        llvm::StringSwitch<LinalgTilingLoopType>(loopType)
-            .Case("for", LinalgTilingLoopType::Loops)
-            .Case("affine", LinalgTilingLoopType::AffineLoops)
-            .Case("parallel", LinalgTilingLoopType::ParallelLoops)
-            .Default(loopTypeEnum);
-    auto options =
-        LinalgTilingOptions().setTileSizes(tileSizes).setLoopType(type);
-    MLIRContext *ctx = funcOp.getContext();
-    RewritePatternSet patterns(ctx);
-    insertTilingPatterns(patterns, options);
-    scf::populateSCFForLoopCanonicalizationPatterns(patterns);
-    (void)applyPatternsAndFoldGreedily(funcOp, std::move(patterns));
-    (void)applyPatternsAndFoldGreedily(
-        funcOp, getLinalgTilingCanonicalizationPatterns(ctx));
-    // Drop the marker.
-    funcOp.walk([](LinalgOp op) {
-      op->removeAttr(LinalgTransforms::kLinalgTransformMarker);
-    });
-
-    // Apply swap pattern after generating loop nest and running
-    // canonicalizations.
-    applyExtractSliceOfPadTensorSwapPattern(funcOp);
-  }
-
-  LinalgTilingLoopType loopTypeEnum;
-};
-
-} // namespace
-
-std::unique_ptr<OperationPass<func::FuncOp>>
-mlir::createLinalgTilingPass(ArrayRef<int64_t> tileSizes,
-                             linalg::LinalgTilingLoopType loopType) {
-  return std::make_unique<LinalgTilingPass>(tileSizes, loopType);
-}

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

for (int64_t loop : peeledLoops) {
     // The result of the loop nest may change with peeling.
     if (res.tensorResults.size() == loopOp->getNumResults() &&
         std::equal(res.tensorResults.begin(), res.tensorResults.end(),
                    loopOp->getResults().begin()))
       res.tensorResults = loopResults;
   }
 }
 
-/// Linalg tiling pattern.
-mlir::linalg::LinalgTilingPattern::LinalgTilingPattern(
-    MLIRContext *context, LinalgTilingOptions options,
-    LinalgTransformationFilter f, PatternBenefit benefit)
-    : OpInterfaceRewritePattern<LinalgOp>(context, benefit),
-      filter(std::move(f)), options(std::move(options)) {}
-
-mlir::linalg::LinalgTilingPattern::LinalgTilingPattern(
-    StringRef opName, MLIRContext *context, LinalgTilingOptions options,
-    LinalgTransformationFilter f, PatternBenefit benefit)
-    : OpInterfaceRewritePattern<LinalgOp>(context, benefit),
-      filter(f.addOpNameFilter(opName)), options(std::move(options)) {}
-
 FailureOr<TiledLinalgOp>
-mlir::linalg::LinalgTilingPattern::returningMatchAndRewrite(
-    LinalgOp op, PatternRewriter &rewriter) const {
-  if (failed(filter.checkAndNotify(rewriter, op)))
-    return failure();
-
+mlir::linalg::tileWithLinalgTilingOptions(RewriterBase &rewriter, LinalgOp op,
+                                          const LinalgTilingOptions &options) {
   FailureOr<TiledLinalgOp> res = tileLinalgOp(rewriter, op, options);
   if (failed(res))
     return failure();
 
-  // Clear filter to stop recursive pattern application.
-  // This must be done here to properly propagate to peeling branches.
-  filter.replaceLinalgTransformationFilter(rewriter, res->op);
-
   // Peel the loops of the TiledLinalgOp.
   peelTiledLinalgOp(rewriter, *res, options.peeledLoops, options.loopType);
 
   if (res->tensorResults.empty())
     rewriter.eraseOp(op);
   else
     rewriter.replaceOp(op, res->tensorResults);
 

mlir/test/Dialect/Linalg/tile-and-distribute.mlir

-// RUN: mlir-opt %s -test-linalg-transform-patterns=test-tile-and-distribute-options -split-input-file | FileCheck %s
-
-func.func @gemm1(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
-{
-  linalg.matmul {__internal_linalg_transform__ = "distribute1"}
-    ins(%a, %b: memref<?x?xf32>, memref<?x?xf32>)
-   outs(%c: memref<?x?xf32>)
-  return
-}
-//  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
-//      CHECK: func @gemm1(
-// CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
-//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
-//      CHECK: scf.for %[[ARG3:.*]] =
-//      CHECK:   %[[OFFSETY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK:   %[[OFFSETX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK:   %[[OFFSETY_2:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK:   %[[OFFSETX_2:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK:   %[[SV1:.*]] = memref.subview %[[ARG0]][%[[OFFSETY]], %[[ARG3]]]
-//      CHECK:   %[[SV2:.*]] = memref.subview %[[ARG1]][%[[ARG3]], %[[OFFSETX]]]
-//      CHECK:   %[[SV3:.*]] = memref.subview %[[ARG2]][%[[OFFSETY_2]], %[[OFFSETX_2]]]
-//      CHECK:   linalg.matmul ins(%[[SV1]], %[[SV2]]{{.*}} outs(%[[SV3]]
-
-// -----
-
-func.func @gemm2(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
-{
-  linalg.matmul  {__internal_linalg_transform__ = "distribute2"}
-    ins(%a, %b: memref<?x?xf32>, memref<?x?xf32>)
-   outs(%c:memref<?x?xf32>)
-  return
-}
-//  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
-//      CHECK: func @gemm2(
-// CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
-//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
-//      CHECK: %[[ITERY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK: %[[ITERX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK: %[[INBOUNDSY:.*]] = arith.cmpi slt, %[[ITERY]], %{{.*}}
-//      CHECK: %[[INBOUNDSX:.*]] = arith.cmpi slt, %[[ITERX]], %{{.*}}
-//      CHECK: %[[INBOUNDS:.*]] = arith.andi %[[INBOUNDSY]], %[[INBOUNDSX]]
-//      CHECK: scf.if %[[INBOUNDS]]
-//      CHECK:   scf.for %[[ARG3:.*]] =
-//      CHECK:     %[[OFFSETY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK:     %[[OFFSETX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK:     %[[OFFSETY_2:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK:     %[[OFFSETX_2:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK:     %[[SV1:.*]] = memref.subview %[[ARG0]][%[[OFFSETY]], %[[ARG3]]]
-//      CHECK:     %[[SV2:.*]] = memref.subview %[[ARG1]][%[[ARG3]], %[[OFFSETX]]]
-//      CHECK:     %[[SV3:.*]] = memref.subview %[[ARG2]][%[[OFFSETY_2]], %[[OFFSETX_2]]]
-//      CHECK:     linalg.matmul ins(%[[SV1]], %[[SV2]]{{.*}} outs(%[[SV3]]
-
-// -----
-
-func.func @gemm3(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
-{
-  linalg.matmul {__internal_linalg_transform__ = "distribute3"}
-    ins(%a, %b: memref<?x?xf32>, memref<?x?xf32>)
-   outs(%c: memref<?x?xf32>)
-  return
-}
-//  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
-//      CHECK: func @gemm3(
-// CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
-//  CHECK-DAG: %[[NBLOCKSY:.*]] = gpu.grid_dim y
-//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
-//  CHECK-DAG: %[[NBLOCKSX:.*]] = gpu.grid_dim x
-//      CHECK: %[[LBY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK: %[[STEPY:.*]] = affine.apply #[[MAP0]]()[%[[NBLOCKSY]]]
-//      CHECK: %[[LBX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK: %[[STEPX:.*]] = affine.apply #[[MAP0]]()[%[[NBLOCKSX]]]
-//      CHECK: scf.parallel (%[[ARG3:.*]], %[[ARG4:.*]]) = (%[[LBY]], %[[LBX]]) to (%{{.*}}, %{{.*}}) step (%[[STEPY]], %[[STEPX]])
-//      CHECK:   scf.for %[[ARG5:.*]] =
-//      CHECK:     %[[SV1:.*]] = memref.subview %[[ARG0]][%[[ARG3]], %[[ARG5]]]
-//      CHECK:     %[[SV2:.*]] = memref.subview %[[ARG1]][%[[ARG5]], %[[ARG4]]]
-//      CHECK:     %[[SV3:.*]] = memref.subview %[[ARG2]][%[[ARG3]], %[[ARG4]]]
-//      CHECK:     linalg.matmul ins(%[[SV1]], %[[SV2]]{{.*}} outs(%[[SV3]]
-
-// -----
-
-func.func @gemm4(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
-{
-  linalg.matmul {__internal_linalg_transform__ = "distribute4"}
-    ins(%a, %b: memref<?x?xf32>, memref<?x?xf32>)
-   outs(%c: memref<?x?xf32>)
-  return
-}
-//  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
-//      CHECK: func @gemm4(
-// CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
-//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
-//      CHECK: %[[LBX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK: %[[INBOUNDS:.*]] = arith.cmpi slt, %[[LBX]], %{{.*}}
-//      CHECK: scf.if %[[INBOUNDS]]
-//      CHECK:   scf.for %[[ARG3:.*]] =
-//      CHECK:     %[[OFFSETY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK:     %[[OFFSETX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK:     %[[OFFSETY_2:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK:     %[[OFFSETX_2:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK:     %[[SV1:.*]] = memref.subview %[[ARG0]][%[[OFFSETY]], %[[ARG3]]]
-//      CHECK:     %[[SV2:.*]] = memref.subview %[[ARG1]][%[[ARG3]], %[[OFFSETX]]]
-//      CHECK:     %[[SV3:.*]] = memref.subview %[[ARG2]][%[[OFFSETY_2]], %[[OFFSETX_2]]]
-//      CHECK:     linalg.matmul ins(%[[SV1]], %[[SV2]]{{.*}} outs(%[[SV3]]
-
-// -----
-
-func.func @gemm5(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
-{
-  linalg.matmul {__internal_linalg_transform__ = "distribute5"}
-    ins(%a, %b: memref<?x?xf32>, memref<?x?xf32>)
-   outs(%c: memref<?x?xf32>)
-  return
-}
-//  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
-//      CHECK: func @gemm5(
-// CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
-//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
-//  CHECK-DAG: %[[NBLOCKSX:.*]] = gpu.grid_dim x
-//      CHECK: %[[LBY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK: %[[LBX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK: %[[STEPX:.*]] = affine.apply #[[MAP0]]()[%[[NBLOCKSX]]]
-//      CHECK: %[[INBOUNDS:.*]] = arith.cmpi slt, %[[LBY]], %{{.*}}
-//      CHECK: scf.if %[[INBOUNDS]]
-//      CHECK:   scf.parallel (%[[ARG3:.*]]) = (%[[LBX]]) to (%{{.*}}) step (%[[STEPX]])
-//      CHECK:     scf.for %[[ARG4:.*]] =
-//      CHECK:      %[[OFFSETY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK:       %[[OFFSETY_2:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK:       %[[SV1:.*]] = memref.subview %[[ARG0]][%[[OFFSETY]], %[[ARG4]]]
-//      CHECK:       %[[SV2:.*]] = memref.subview %[[ARG1]][%[[ARG4]], %[[ARG3]]]
-//      CHECK:       %[[SV3:.*]] = memref.subview %[[ARG2]][%[[OFFSETY_2]], %[[ARG3]]]
-//      CHECK:       linalg.matmul ins(%[[SV1]], %[[SV2]]{{.*}} outs(%[[SV3]]
-
-// -----
-
-func.func @gemm6(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
-{
-  linalg.matmul {__internal_linalg_transform__ = "distribute6"}
-    ins(%a, %b: memref<?x?xf32>, memref<?x?xf32>)
-   outs(%c: memref<?x?xf32>)
-  return
-}
-//  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
-//      CHECK: func @gemm6(
-// CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-// CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
-//  CHECK-DAG: %[[NBLOCKSY:.*]] = gpu.grid_dim y
-//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
-//      CHECK: %[[LBY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
-//      CHECK: %[[STEPY:.*]] = affine.apply #[[MAP0]]()[%[[NBLOCKSY]]]
-//      CHECK: scf.parallel (%[[ARG3:.*]]) = (%[[LBY]]) to (%{{.*}}) step (%[[STEPY]])
-//      CHECK:   scf.for %[[ARG4:.*]] =
-//      CHECK:     %[[OFFSETX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK:     %[[OFFSETX_2:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
-//      CHECK:     %[[SV1:.*]] = memref.subview %[[ARG0]][%[[ARG3]], %[[ARG4]]]
-//      CHECK:     %[[SV2:.*]] = memref.subview %[[ARG1]][%[[ARG4]], %[[OFFSETX]]]
-//      CHECK:     %[[SV3:.*]] = memref.subview %[[ARG2]][%[[ARG3]], %[[OFFSETX_2]]]
-//      CHECK:     linalg.matmul ins(%[[SV1]], %[[SV2]]{{.*}} outs(%[[SV3]]
-
-// -----
-
-//      CHECK: #[[MULMAP:.+]] = affine_map<()[s0, s1] -> (s0 * s1)>
-//      CHECK: #[[ADDMAP:.+]] = affine_map<()[s0, s1] -> (s0 + s1)>
-//      CHECK: func @matmul_tensors(
-// CHECK-SAME:    %[[TA:[0-9a-z]+]]: tensor<?x?xf32>
-// CHECK-SAME:    %[[TB:[0-9a-z]+]]: tensor<?x?xf32>
-// CHECK-SAME:    %[[TC:[0-9a-z]+]]: tensor<?x?xf32>) -> tensor<?x?xf32> {
-func.func @matmul_tensors(
-  %arg0: tensor<?x?xf32>, %arg1: tensor<?x?xf32>, %arg2: tensor<?x?xf32>)
-    -> tensor<?x?xf32> {
-//  CHECK-DAG: %[[C8:.*]] = arith.constant 8 : index
-//  CHECK-DAG: %[[C0:.*]] = arith.constant 0 : index
-//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
-//  CHECK-DAG: %[[NBLOCKSY:.*]] = gpu.grid_dim y
-//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
-//  CHECK-DAG: %[[NBLOCKSX:.*]] = gpu.grid_dim x
-//      CHECK: %[[MUL:.+]] = affine.apply #[[MULMAP]]()[%[[BIDY]], %[[C8]]]
-//      CHECK: %[[LBY:.+]] = affine.apply #[[ADDMAP]]()[%[[MUL]], %[[C0]]]
-//      CHECK: %[[STEPY:.+]] = affine.apply #[[MULMAP]]()[%[[NBLOCKSY]], %[[C8]]]
-//      CHECK: %[[TD0:.*]] = scf.for {{.*}} to {{.*}} step {{.*}} iter_args(%[[TC0:.*]] = %[[TC]]) -> (tensor<?x?xf32>) {
-//      CHECK: %[[MUL:.+]] = affine.apply #[[MULMAP]]()[%[[BIDX]], %[[C8]]]
-//      CHECK: %[[LBX:.+]] = affine.apply #[[ADDMAP]]()[%[[MUL]], %[[C0]]]
-//      CHECK: %[[STEPX:.+]] = affine.apply #[[MULMAP]]()[%[[NBLOCKSX]], %[[C8]]]
-//      CHECK:   %[[TD1:.*]] = scf.for {{.*}} to {{.*}} step {{.*}} iter_args(%[[TC1:.*]] = %[[TC0]]) -> (tensor<?x?xf32>) {
-//      CHECK:     %[[TD2:.*]] = scf.for {{.*}} to {{.*}} step {{.*}} iter_args(%[[TC2:.*]] = %[[TC1]]) -> (tensor<?x?xf32>) {
-//      CHECK:       %[[sTA:.*]] = tensor.extract_slice %[[TA]][{{.*}}] : tensor<?x?xf32> to tensor<?x?xf32>
-//      CHECK:       %[[sTB:.*]] = tensor.extract_slice %[[TB]][{{.*}}] : tensor<?x?xf32> to tensor<?x?xf32>
-//      CHECK:       %[[sTC:.*]] = tensor.extract_slice %[[TC2]][{{.*}}] : tensor<?x?xf32> to tensor<?x?xf32>
-//      CHECK:       %[[sTD:.*]] = linalg.matmul ins(%[[sTA]], %[[sTB]] : tensor<?x?xf32>, tensor<?x?xf32>)
-// CHECK-SAME:                                  outs(%[[sTC]] : tensor<?x?xf32>)  -> tensor<?x?xf32>
-//      CHECK:       %[[TD:.*]] = tensor.insert_slice %[[sTD]] into %[[TC2]][{{.*}}]  : tensor<?x?xf32> into tensor<?x?xf32>
-//      CHECK:       scf.yield %[[TD]] : tensor<?x?xf32>
-//      CHECK:     scf.yield %[[TD2]] : tensor<?x?xf32>
-//      CHECK:   scf.yield %[[TD1]] : tensor<?x?xf32>
-  %0 = linalg.matmul {__internal_linalg_transform__ = "tensors_distribute1"}
-       ins(%arg0, %arg1: tensor<?x?xf32>, tensor<?x?xf32>)
-      outs(%arg2: tensor<?x?xf32>)
-    -> tensor<?x?xf32>
-
-//      CHECK: return %[[TD0]] : tensor<?x?xf32>
-  return %0 : tensor<?x?xf32>
-}
-

mlir/test/Dialect/Linalg/tile-and-peel-tensors.mlir

-// RUN: mlir-opt %s -test-linalg-transform-patterns="test-tile-pattern tile-sizes=256,128,512 peeled-loops=0" -canonicalize | \
-// RUN:     FileCheck %s -check-prefix=CHECK-PEEL-0
-
-// RUN: mlir-opt %s -test-linalg-transform-patterns="test-tile-pattern tile-sizes=256,128,512 peeled-loops=1,2" -canonicalize | \
-// RUN:     FileCheck %s -check-prefix=CHECK-PEEL-12
-
-//     CHECK-PEEL-0: func @matmul_static_tensor
-// CHECK-PEEL-0-DAG:   %[[c0:.*]] = arith.constant 0 : index
-// CHECK-PEEL-0-DAG:   %[[c128:.*]] = arith.constant 128 : index
-// CHECK-PEEL-0-DAG:   %[[c256:.*]] = arith.constant 256 : index
-// CHECK-PEEL-0-DAG:   %[[c512:.*]] = arith.constant 512 : index
-// CHECK-PEEL-0-DAG:   %[[c1280:.*]] = arith.constant 1280 : index
-// CHECK-PEEL-0-DAG:   %[[c1600:.*]] = arith.constant 1600 : index
-// CHECK-PEEL-0-DAG:   %[[c1700:.*]] = arith.constant 1700 : index
-//     CHECK-PEEL-0:   scf.for %{{.*}} = %[[c0]] to %[[c1280]] step %[[c256]] {{.*}} {
-//     CHECK-PEEL-0:     scf.for %{{.*}} = %[[c0]] to %[[c1700]] step %[[c128]] {{.*}} {
-//     CHECK-PEEL-0:       scf.for %{{.*}} = %[[c0]] to %[[c1600]] step %[[c512]] {{.*}} {
-//     CHECK-PEEL-0:         linalg.matmul ins({{.*}} : tensor<256x?xf32>, tensor<?x?xf32>) outs({{.*}} : tensor<256x?xf32>)
-//     CHECK-PEEL-0:       }
-//     CHECK-PEEL-0:     }
-//     CHECK-PEEL-0:   }
-//     CHECK-PEEL-0:   scf.for %{{.*}} = %[[c0]] to %[[c1700]] step %[[c128]] {{.*}} {
-//     CHECK-PEEL-0:     scf.for %{{.*}} = %[[c0]] to %[[c1600]] step %[[c512]] {{.*}} {
-//     CHECK-PEEL-0:       linalg.matmul ins({{.*}} : tensor<220x?xf32>, tensor<?x?xf32>) outs({{.*}} : tensor<220x?xf32>)
-//     CHECK-PEEL-0:     }
-//     CHECK-PEEL-0:   }
-
-//     CHECK-PEEL-12: func @matmul_static_tensor
-// CHECK-PEEL-12-DAG:   %[[c0:.*]] = arith.constant 0 : index
-// CHECK-PEEL-12-DAG:   %[[c128:.*]] = arith.constant 128 : index
-// CHECK-PEEL-12-DAG:   %[[c256:.*]] = arith.constant 256 : index
-// CHECK-PEEL-12-DAG:   %[[c512:.*]] = arith.constant 512 : index
-// CHECK-PEEL-12-DAG:   %[[c1500:.*]] = arith.constant 1500 : index
-// CHECK-PEEL-12-DAG:   %[[c1536:.*]] = arith.constant 1536 : index
-// CHECK-PEEL-12-DAG:   %[[c1600:.*]] = arith.constant 1600 : index
-// CHECK-PEEL-12-DAG:   %[[c1664:.*]] = arith.constant 1664 : index
-//     CHECK-PEEL-12:   scf.for %{{.*}} = %[[c0]] to %[[c1500]] step %[[c256]] {{.*}} {
-//     CHECK-PEEL-12:     scf.for %{{.*}} = %[[c0]] to %[[c1664]] step %[[c128]] {{.*}} {
-//     CHECK-PEEL-12:       scf.for %{{.*}} = %[[c0]] to %[[c1536]] step %[[c512]] {{.*}} {
-//     CHECK-PEEL-12:         linalg.matmul ins({{.*}} : tensor<?x512xf32>, tensor<512x128xf32>) outs({{.*}} : tensor<?x128xf32>)
-//     CHECK-PEEL-12:       }
-//     CHECK-PEEL-12:       linalg.matmul ins({{.*}} : tensor<?x64xf32>, tensor<64x128xf32>) outs({{.*}} : tensor<?x128xf32>)
-//     CHECK-PEEL-12:     }
-//     CHECK-PEEL-12:     scf.for %{{.*}} = %[[c0]] to %[[c1600]] step %[[c512]] {{.*}} {
-//     CHECK-PEEL-12:       linalg.matmul ins({{.*}} : tensor<?x?xf32>, tensor<?x36xf32>) outs({{.*}} : tensor<?x36xf32>)
-//     CHECK-PEEL-12:     }
-//     CHECK-PEEL-12:   }
-func.func @matmul_static_tensor(%arg0: tensor<1500x1600xf32>, %arg1: tensor<1600x1700xf32>)
-    -> tensor<1500x1700xf32> {
-  %out = tensor.empty() : tensor<1500x1700xf32>
-  %r = linalg.matmul {__internal_linalg_transform__ = "tile"}
-      ins(%arg0, %arg1: tensor<1500x1600xf32>, tensor<1600x1700xf32>)
-      outs(%out: tensor<1500x1700xf32>) -> tensor<1500x1700xf32>
-  return %r : tensor<1500x1700xf32>
-}
-
-// -----
-
-//     CHECK-PEEL-0: func @matmul_dynamic_tensor
-// CHECK-PEEL-0-DAG:   %[[c0:.*]] = arith.constant 0 : index
-// CHECK-PEEL-0-DAG:   %[[c128:.*]] = arith.constant 128 : index
-// CHECK-PEEL-0-DAG:   %[[c256:.*]] = arith.constant 256 : index
-// CHECK-PEEL-0-DAG:   %[[c512:.*]] = arith.constant 512 : index
-//     CHECK-PEEL-0:   scf.for %{{.*}} = %[[c0]] to %{{.*}} step %[[c256]] {{.*}} {
-//     CHECK-PEEL-0:     scf.for %{{.*}} = %[[c0]] to %{{.*}} step %[[c128]] {{.*}} {
-//     CHECK-PEEL-0:       scf.for %{{.*}} = %[[c0]] to %{{.*}} step %[[c512]] {{.*}} {
-//     CHECK-PEEL-0:         linalg.matmul ins({{.*}} : tensor<256x?xf32>, tensor<?x?xf32>) outs({{.*}} : tensor<256x?xf32>)
-//     CHECK-PEEL-0:       }
-//     CHECK-PEEL-0:     }
-//     CHECK-PEEL-0:   }
-//     CHECK-PEEL-0:   scf.for %{{.*}} {
-//     CHECK-PEEL-0:     scf.for %{{.*}} = %[[c0]] to %{{.*}} step %[[c128]] {{.*}} {
-//     CHECK-PEEL-0:       scf.for %{{.*}} = %[[c0]] to %{{.*}} step %[[c512]] {{.*}} {
-//     CHECK-PEEL-0:         linalg.matmul ins({{.*}} : tensor<?x?xf32>, tensor<?x?xf32>) outs({{.*}} : tensor<?x?xf32>)
-//     CHECK-PEEL-0:       }
-//     CHECK-PEEL-0:     }
-//     CHECK-PEEL-0:   }
-
-//     CHECK-PEEL-12: func @matmul_dynamic_tensor
-// CHECK-PEEL-12-DAG:   %[[c0:.*]] = arith.constant 0 : index
-// CHECK-PEEL-12-DAG:   %[[c128:.*]] = arith.constant 128 : index
-// CHECK-PEEL-12-DAG:   %[[c256:.*]] = arith.constant 256 : index
-// CHECK-PEEL-12-DAG:   %[[c512:.*]] = arith.constant 512 : index
-//     CHECK-PEEL-12:   scf.for %{{.*}} = %[[c0]] to %{{.*}} step %[[c256]] {{.*}} {
-//     CHECK-PEEL-12:     scf.for %{{.*}} = %[[c0]] to %{{.*}} step %[[c128]] {{.*}} {
-//     CHECK-PEEL-12:       scf.for %{{.*}} = %[[c0]] to %{{.*}} step %[[c512]] {{.*}} {
-//     CHECK-PEEL-12:         linalg.matmul ins({{.*}} : tensor<?x512xf32>, tensor<512x128xf32>) outs({{.*}} : tensor<?x128xf32>)
-//     CHECK-PEEL-12:       }
-//     CHECK-PEEL-12:       scf.for %{{.*}} {
-//     CHECK-PEEL-12:         linalg.matmul ins({{.*}} : tensor<?x?xf32>, tensor<?x128xf32>) outs({{.*}} : tensor<?x128xf32>)
-//     CHECK-PEEL-12:       }
-//     CHECK-PEEL-12:     }
-//     CHECK-PEEL-12:     scf.for %{{.*}} {
-//     CHECK-PEEL-12:       scf.for %{{.*}} = %[[c0]] to %{{.*}} step %[[c512]] {{.*}} {
-//     CHECK-PEEL-12:         linalg.matmul ins({{.*}} : tensor<?x?xf32>, tensor<?x?xf32>) outs({{.*}} : tensor<?x?xf32>)
-//     CHECK-PEEL-12:       }
-//     CHECK-PEEL-12:     }
-//     CHECK-PEEL-12:   }
-func.func @matmul_dynamic_tensor(%arg0: tensor<?x?xf32>, %arg1: tensor<?x?xf32>)
-    -> tensor<?x?xf32> {
-  %c0 = arith.constant 0 : index
-  %c1 = arith.constant 1 : index
-  %d0 = tensor.dim %arg0, %c0 : tensor<?x?xf32>
-  %d1 = tensor.dim %arg1, %c1 : tensor<?x?xf32>
-  %out = tensor.empty(%d0, %d1) : tensor<?x?xf32>
-  %r = linalg.matmul {__internal_linalg_transform__ = "tile"}
-      ins(%arg0, %arg1: tensor<?x?xf32>, tensor<?x?xf32>)
-      outs(%out: tensor<?x?xf32>) -> tensor<?x?xf32>
-  return %r : tensor<?x?xf32>
-}

mlir/test/Dialect/Linalg/tile-conv.mlir

-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,3" | FileCheck %s
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -canonicalize | FileCheck %s
 
 //  CHECK-DAG: #[[MAP0:.*]] = affine_map<(d0)[s0, s1] -> (-d0 + s0 + s1 - 1, s1 + 1)>
 //  CHECK-DAG: #[[MAP1:.*]] = affine_map<(d0)[s0, s1] -> (-d0 + s0 + s1 - 1, s1 + 2)>
 //  CHECK-DAG: #[[MAP2:.*]] = affine_map<(d0)[s0] -> (-d0 + s0, 2)>
 //  CHECK-DAG: #[[MAP3:.*]] = affine_map<(d0)[s0] -> (-d0 + s0, 3)>
 
 func.func @conv(%arg0 : memref<?x?xf32>, %arg1 : memref<?x?xf32>, %arg2 : memref<?x?xf32>) {
   linalg.conv_2d ins(%arg0, %arg1 : memref<?x?xf32>, memref<?x?xf32>) outs(%arg2 : memref<?x?xf32>)
   return
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.conv_2d"]} in %arg1
+    %1, %loop:2 = transform.structured.tile %0 [2, 3]
+}
+
 //       CHECK: func @conv
 //  CHECK-SAME:   %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 //  CHECK-SAME:   %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 //  CHECK-SAME:   %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 //   CHECK-DAG:   %[[C0:.*]] = arith.constant 0 : index
 //   CHECK-DAG:   %[[C1:.*]] = arith.constant 1 : index
 //   CHECK-DAG:   %[[C2:.*]] = arith.constant 2 : index
 //   CHECK-DAG:   %[[C3:.*]] = arith.constant 3 : index

mlir/test/Dialect/Linalg/tile-indexed.mlir

-// RUN: mlir-opt %s -linalg-tile="tile-sizes=10,25" -split-input-file | FileCheck %s -check-prefix=TILE-10n25
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=25,0" -split-input-file | FileCheck %s -check-prefix=TILE-25n0
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=0,25" -split-input-file | FileCheck %s -check-prefix=TILE-0n25
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -canonicalize -split-input-file | FileCheck %s -check-prefix=TILE-10n25
 
 func.func @indexed_vector(%arg0: memref<50xindex>) {
   linalg.generic {indexing_maps = [affine_map<(i) -> (i)>],
                   iterator_types = ["parallel"]}
      outs(%arg0 : memref<50xindex>) {
     ^bb0(%a: index):
       %i = linalg.index 0 : index
       linalg.yield %i : index
   }
   return
 }
+
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.generic"]} in %arg1
+    %1, %loop:2 = transform.structured.tile %0 [10, 25]
+}
+
 // TILE-10n25-DAG: [[$MAP:#[a-zA-Z0-9_]*]] = affine_map<(d0, d1) -> (d0 + d1)>
 // TILE-10n25-LABEL: func @indexed_vector
 // TILE-10n25: %[[C10:.*]] = arith.constant 10 : index
 // TILE-10n25: scf.for %[[J:.*]] = {{.*}} step %[[C10]]
 // TILE-10n25:   linalg.generic
 // TILE-10n25:     %[[I:.*]] = linalg.index 0 : index
 // TILE-10n25:     %[[NEW_I:.*]] = affine.apply [[$MAP]](%[[I]], %[[J]])
 // TILE-10n25:     linalg.yield %[[NEW_I]] : index
 
-// TILE-25n0-DAG: [[$MAP:#[a-zA-Z0-9_]*]] = affine_map<(d0, d1) -> (d0 + d1)>
-// TILE-25n0-LABEL: func @indexed_vector
-// TILE-25n0: %[[C25:.*]] = arith.constant 25 : index
-// TILE-25n0: scf.for %[[J:.*]] = {{.*}} step %[[C25]]
-// TILE-25n0:   linalg.generic
-// TILE-25n0:     %[[I:.*]] = linalg.index 0 : index
-// TILE-25n0:     %[[NEW_I:.*]] = affine.apply [[$MAP]](%[[I]], %[[J]])
-// TILE-25n0:     linalg.yield %[[NEW_I]] : index
-
-// TILE-0n25-LABEL: func @indexed_vector
-// TILE-0n25-NOT: scf.for %[[J:.*]] = {{.*}} step %
-// TILE-0n25: linalg.generic
-
 // -----
 
 func.func @indexed_matrix(%arg0: memref<50x50xindex>) {
   linalg.generic {indexing_maps = [affine_map<(i, j) -> (i, j)>],
                   iterator_types = ["parallel", "parallel"]}
     outs(%arg0 : memref<50x50xindex>) {
     ^bb0(%a: index):
       %i = linalg.index 0 : index
       %j = linalg.index 1 : index
       %sum = arith.addi %i, %j : index
       linalg.yield %sum : index
   }
   return
 }
+
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.generic"]} in %arg1
+    %1, %loop:2 = transform.structured.tile %0 [10, 25]
+}
+
 // TILE-10n25-DAG: [[$MAP:#[a-zA-Z0-9_]*]] = affine_map<(d0, d1) -> (d0 + d1)>
 // TILE-10n25-LABEL: func @indexed_matrix
 // TILE-10n25-DAG: %[[C25:.*]] = arith.constant 25 : index
 // TILE-10n25-DAG: %[[C10:.*]] = arith.constant 10 : index
 // TILE-10n25: scf.for %[[K:.*]] = {{.*}} step %[[C10]]
 // TILE-10n25:   scf.for %[[L:.*]] = {{.*}} step %[[C25]]
 // TILE-10n25:     linalg.generic
 // TILE-10n25:       %[[I:.*]] = linalg.index 0 : index
 // TILE-10n25:       %[[NEW_I:.*]] = affine.apply [[$MAP]](%[[I]], %[[K]])
 // TILE-10n25:       %[[J:.*]] = linalg.index 1 : index
 // TILE-10n25:       %[[NEW_J:.*]] = affine.apply [[$MAP]](%[[J]], %[[L]])
 // TILE-10n25:       %[[SUM:.*]] = arith.addi %[[NEW_I]], %[[NEW_J]] : index
 // TILE-10n25:       linalg.yield %[[SUM]] : index
-
-// TILE-25n0-DAG: [[$MAP:#[a-zA-Z0-9_]*]] = affine_map<(d0, d1) -> (d0 + d1)>
-// TILE-25n0-LABEL: func @indexed_matrix
-// TILE-25n0: %[[C25:.*]] = arith.constant 25 : index
-// TILE-25n0: scf.for %[[L:.*]] = {{.*}} step %[[C25]]
-// TILE-25n0:   linalg.generic
-// TILE-25n0:     %[[I:.*]] = linalg.index 0 : index
-// TILE-25n0:     %[[NEW_I:.*]] = affine.apply [[$MAP]](%[[I]], %[[L]])
-// TILE-25n0:     %[[J:.*]] = linalg.index 1 : index
-// TILE-25n0:     %[[SUM:.*]] = arith.addi %[[NEW_I]], %[[J]] : index
-// TILE-25n0:     linalg.yield %[[SUM]] : index
-
-// TILE-0n25-DAG: [[$MAP:#[a-zA-Z0-9_]*]] = affine_map<(d0, d1) -> (d0 + d1)>
-// TILE-0n25-LABEL: func @indexed_matrix
-// TILE-0n25: %[[C25:.*]] = arith.constant 25 : index
-// TILE-0n25: scf.for %[[L:.*]] = {{.*}} step %[[C25]]
-// TILE-0n25:   linalg.generic
-// TILE-0n25:     %[[I:.*]] = linalg.index 0 : index
-// TILE-0n25:     %[[J:.*]] = linalg.index 1 : index
-// TILE-0n25:     %[[NEW_J:.*]] = affine.apply [[$MAP]](%[[J]], %[[L]])
-// TILE-0n25:     %[[SUM:.*]] = arith.addi %[[I]], %[[NEW_J]] : index
-// TILE-0n25:     linalg.yield %[[SUM]] : index

mlir/test/Dialect/Linalg/tile-pad-tensor-op.mlir

-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,3" -cse -split-input-file | \
-// RUN: FileCheck %s -check-prefix=TILE2
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=0,3" -resolve-shaped-type-result-dims -cse -split-input-file | \
-// RUN: FileCheck %s -check-prefix=TILE1
-// This test only checks that tiling does not crash.
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2" -resolve-shaped-type-result-dims -cse -split-input-file
-
-//  TILE2-DAG:  #[[MAP0:.*]] = affine_map<()[s0] -> (s0 + 8)>
-//  TILE2-DAG:  #[[MAP1:.*]] = affine_map<()[s0] -> (s0 + 7)>
-//       TILE2: func @dynamic_pad_tensor(
-//  TILE2-SAME:     %[[IN:.*]]: tensor<?x?xf32>
-//   TILE2-DAG:   %[[C0:.*]] = arith.constant 0 : index
-//   TILE2-DAG:   %[[C1:.*]] = arith.constant 1 : index
-//   TILE2-DAG:   %[[C2:.*]] = arith.constant 2 : index
-//   TILE2-DAG:   %[[C3:.*]] = arith.constant 3 : index
-//       TILE2:   %[[DIM_IN0:.*]] = tensor.dim %[[IN]], %[[C0]]
-//       TILE2:   %[[DIM0:.*]] = affine.apply #[[MAP0]]()[%[[DIM_IN0]]]
-//       TILE2:   %[[DIM_IN1:.*]] = tensor.dim %[[IN]], %[[C1]]
-//       TILE2:   %[[DIM1:.*]] = affine.apply #[[MAP1]]()[%[[DIM_IN1]]]
-//       TILE2:   %[[RESULT:.*]] = scf.for {{.*}} = %[[C0]] to %[[DIM0]] step %[[C2]]
-//       TILE2:     scf.for {{.*}} = %[[C0]] to %[[DIM1]] step %[[C3]] iter_args(%[[INNER_OUT:.*]] =
-//       TILE2:       %[[SWAP_RESULT:.*]] = scf.if
-//       TILE2:         tensor.generate
-//       TILE2:       else
-//       TILE2:         %[[SLICE:.*]] = tensor.extract_slice %[[IN]][{{.*}}, {{.*}}] [{{.*}}, {{.*}}] [1, 1]
-//       TILE2:         %[[PAD:.*]] = tensor.pad %[[SLICE]]
-//       TILE2:       tensor.insert_slice %[[SWAP_RESULT]] into %[[INNER_OUT]][{{.*}}, {{.*}}] [{{.*}}, {{.*}}] [1, 1]
-//       TILE2:   return %[[RESULT]]
-
-//   TILE1-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 + 7)>
-//   TILE1-DAG: #[[MAP1:.*]] = affine_map<()[s0] -> (s0 + 8)>
-//       TILE1: func @dynamic_pad_tensor(
-//  TILE1-SAME:     %[[IN:.*]]: tensor<?x?xf32>
-//   TILE1-DAG:   %[[C0:.*]] = arith.constant 0 : index
-//   TILE1-DAG:   %[[C1:.*]] = arith.constant 1 : index
-//   TILE1-DAG:   %[[C3:.*]] = arith.constant 3 : index
-//       TILE1:   %[[DIM_IN1:.*]] = tensor.dim %[[IN]], %[[C1]]
-//       TILE1:   %[[DIM1:.*]] = affine.apply #[[MAP0]]()[%[[DIM_IN1]]]
-//       TILE1:   %[[DIM_IN0:.*]] = tensor.dim %[[IN]], %[[C0]]
-//       TILE1:   %[[DIM0:.*]] = affine.apply #[[MAP1]]()[%[[DIM_IN0]]]
-//       TILE1:   %[[RESULT:.*]] = scf.for {{.*}} = %[[C0]] to %[[DIM1]] step %[[C3]] iter_args(%[[INNER_OUT:.*]] =
-//       TILE1:     %[[SWAP_RESULT:.*]] = scf.if
-//       TILE1:       tensor.generate
-//       TILE1:     else
-//       TILE1:       %[[SLICE:.*]] = tensor.extract_slice %[[IN]][{{.*}}, {{.*}}] [{{.*}}, {{.*}}] [1, 1]
-//       TILE1:       %[[PAD:.*]] = tensor.pad %[[SLICE]] low[3, %{{.*}}] high[{{.*}}, {{.*}}]
-//       TILE1:     tensor.insert_slice %[[SWAP_RESULT]] into %[[INNER_OUT]][0, {{.*}}] [%[[DIM0]], {{.*}}] [1, 1]
-//       TILE1:   return %[[RESULT]]
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -canonicalize -cse -split-input-file
 
-func.func @dynamic_pad_tensor(%input_tensor: tensor<?x?xf32>,
+//  CHECK-DAG:  #[[MAP0:.*]] = affine_map<()[s0] -> (s0 + 8)>
+//  CHECK-DAG:  #[[MAP1:.*]] = affine_map<()[s0] -> (s0 + 7)>
+//       CHECK: func @dynamic_pad_tensor_3_4(
+//  CHECK-SAME:     %[[IN:.*]]: tensor<?x?xf32>
+//   CHECK-DAG:   %[[C0:.*]] = arith.constant 0 : index
+//   CHECK-DAG:   %[[C1:.*]] = arith.constant 1 : index
+//   CHECK-DAG:   %[[C2:.*]] = arith.constant 2 : index
+//   CHECK-DAG:   %[[C3:.*]] = arith.constant 3 : index
+//   CHECK-DAG:   %[[DIM_IN0:.*]] = tensor.dim %[[IN]], %[[C0]]
+//   CHECK-DAG:   %[[DIM_IN1:.*]] = tensor.dim %[[IN]], %[[C1]]
+//   CHECK-DAG:   %[[DIM0:.*]] = affine.apply #[[MAP0]]()[%[[DIM_IN0]]]
+//   CHECK-DAG:   %[[DIM1:.*]] = affine.apply #[[MAP1]]()[%[[DIM_IN1]]]
+//       CHECK:   %[[RESULT:.*]] = scf.for {{.*}} = %[[C0]] to %[[DIM0]] step %[[C2]]
+//       CHECK:     scf.for {{.*}} = %[[C0]] to %[[DIM1]] step %[[C3]] iter_args(%[[INNER_OUT:.*]] =
+//       CHECK:       %[[SWAP_RESULT:.*]] = scf.if
+//       CHECK:         tensor.generate
+//       CHECK:       else
+//       CHECK:         %[[SLICE:.*]] = tensor.extract_slice %[[IN]][{{.*}}, {{.*}}] [{{.*}}, {{.*}}] [1, 1]
+//       CHECK:         %[[PAD:.*]] = tensor.pad %[[SLICE]]
+//       CHECK:       tensor.insert_slice %[[SWAP_RESULT]] into %[[INNER_OUT]][{{.*}}, {{.*}}] [{{.*}}, {{.*}}] [1, 1]
+//       CHECK:   return %[[RESULT]]
+
+func.func @dynamic_pad_tensor_3_4(%input_tensor: tensor<?x?xf32>,
                          %pad_value: f32) -> tensor<?x?xf32> {
   %0 = tensor.pad %input_tensor low[3, 4] high[5, 3] {
     ^bb0(%arg1: index, %arg2: index):
       tensor.yield %pad_value : f32
     } : tensor<?x?xf32> to tensor<?x?xf32>
   return %0 : tensor<?x?xf32>
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["tensor.pad"]} in %arg1
+    %1, %loops:2 = transform.structured.tile_to_scf_for %0 [2, 3]
+}
+
+// -----
+
+//   CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 + 7)>
+//   CHECK-DAG: #[[MAP1:.*]] = affine_map<()[s0] -> (s0 + 8)>
+//       CHECK: func @dynamic_pad_tensor_0_3(
+//  CHECK-SAME:     %[[IN:.*]]: tensor<?x?xf32>
+//   CHECK-DAG:   %[[C0:.*]] = arith.constant 0 : index
+//   CHECK-DAG:   %[[C1:.*]] = arith.constant 1 : index
+//   CHECK-DAG:   %[[C3:.*]] = arith.constant 3 : index
+//   CHECK-DAG:   %[[DIM_IN1:.*]] = tensor.dim %[[IN]], %[[C1]]
+//   CHECK-DAG:   %[[DIM1:.*]] = affine.apply #[[MAP0]]()[%[[DIM_IN1]]]
+//   CHECK-DAG:   %[[DIM_IN0:.*]] = tensor.dim %[[IN]], %[[C0]]
+//   CHECK-DAG:   %[[DIM0:.*]] = affine.apply #[[MAP1]]()[%[[DIM_IN0]]]
+//       CHECK:   %[[RESULT:.*]] = scf.for {{.*}} = %[[C0]] to %[[DIM1]] step %[[C3]] iter_args(%[[INNER_OUT:.*]] =
+//       CHECK:     %[[SWAP_RESULT:.*]] = scf.if
+//       CHECK:       tensor.generate
+//       CHECK:     else
+//       CHECK:       %[[SLICE:.*]] = tensor.extract_slice %[[IN]][{{.*}}, {{.*}}] [{{.*}}, {{.*}}] [1, 1]
+//       CHECK:       %[[PAD:.*]] = tensor.pad %[[SLICE]] low[3, %{{.*}}] high[{{.*}}, {{.*}}]
+//       CHECK:     tensor.insert_slice %[[SWAP_RESULT]] into %[[INNER_OUT]][0, {{.*}}] [%[[DIM0]], {{.*}}] [1, 1]
+//       CHECK:   return %[[RESULT]]
+
+func.func @dynamic_pad_tensor_0_3(%input_tensor: tensor<?x?xf32>,
+                         %pad_value: f32) -> tensor<?x?xf32> {
+  %0 = tensor.pad %input_tensor low[3, 4] high[5, 3] {
+    ^bb0(%arg1: index, %arg2: index):
+      tensor.yield %pad_value : f32
+    } : tensor<?x?xf32> to tensor<?x?xf32>
+  return %0 : tensor<?x?xf32>
+}
+
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["tensor.pad"]} in %arg1
+    %1, %loop = transform.structured.tile_to_scf_for %0 [0, 3]
+}
+
+// -----
+
+// CHECK-LABEL: func @static_pad_tensor_3_4(
+//  CHECK-SAME:     %[[IN:.*]]: tensor<7x9xf32>
+//   CHECK-DAG:   %[[C0:.*]] = arith.constant 0 : index
+//   CHECK-DAG:   %[[C2:.*]] = arith.constant 2 : index
+//   CHECK-DAG:   %[[C3:.*]] = arith.constant 3 : index
+//   CHECK-DAG:   %[[C15:.*]] = arith.constant 15 : index
+//   CHECK-DAG:   %[[C16:.*]] = arith.constant 16 : index
+//       CHECK:   %[[RESULT:.*]] = scf.for {{.*}} = %[[C0]] to %[[C15]] step %[[C2]]
+//       CHECK:     scf.for {{.*}} = %[[C0]] to %[[C16]] step %[[C3]] iter_args(%[[INNER_OUT:.*]] =
+//       CHECK:       %[[SWAP_RESULT:.*]] = scf.if
+//       CHECK:         tensor.generate
+//       CHECK:       else
+//       CHECK:         %[[SLICE:.*]] = tensor.extract_slice %[[IN]][{{.*}}, {{.*}}] [{{.*}}, {{.*}}] [1, 1]
+//       CHECK:         %[[PAD:.*]] = tensor.pad %[[SLICE]]
+//       CHECK:       tensor.insert_slice %[[SWAP_RESULT]] into %[[INNER_OUT]][{{.*}}, {{.*}}] [{{.*}}, {{.*}}] [1, 1]
+//       CHECK:   return %[[RESULT]]
+
+func.func @static_pad_tensor_3_4(%input_tensor: tensor<7x9xf32>,
+                        %pad_value: f32) -> tensor<15x16xf32> {
+  %0 = tensor.pad %input_tensor low[3, 4] high[5, 3] {
+    ^bb0(%arg1: index, %arg2: index):
+      tensor.yield %pad_value : f32
+    } : tensor<7x9xf32> to tensor<15x16xf32>
+  return %0 : tensor<15x16xf32>
+}
+
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["tensor.pad"]} in %arg1
+    %1, %loops:2 = transform.structured.tile_to_scf_for %0 [2, 3]
+}
+
 // -----
 
-// TILE2-LABEL: func @static_pad_tensor(
-//  TILE2-SAME:     %[[IN:.*]]: tensor<7x9xf32>
-//   TILE2-DAG:   %[[C0:.*]] = arith.constant 0 : index
-//   TILE2-DAG:   %[[C2:.*]] = arith.constant 2 : index
-//   TILE2-DAG:   %[[C3:.*]] = arith.constant 3 : index
-//   TILE2-DAG:   %[[C15:.*]] = arith.constant 15 : index
-//   TILE2-DAG:   %[[C16:.*]] = arith.constant 16 : index
-//       TILE2:   %[[RESULT:.*]] = scf.for {{.*}} = %[[C0]] to %[[C15]] step %[[C2]]
-//       TILE2:     scf.for {{.*}} = %[[C0]] to %[[C16]] step %[[C3]] iter_args(%[[INNER_OUT:.*]] =
-//       TILE2:       %[[SWAP_RESULT:.*]] = scf.if
-//       TILE2:         tensor.generate
-//       TILE2:       else
-//       TILE2:         %[[SLICE:.*]] = tensor.extract_slice %[[IN]][{{.*}}, {{.*}}] [{{.*}}, {{.*}}] [1, 1]
-//       TILE2:         %[[PAD:.*]] = tensor.pad %[[SLICE]]
-//       TILE2:       tensor.insert_slice %[[SWAP_RESULT]] into %[[INNER_OUT]][{{.*}}, {{.*}}] [{{.*}}, {{.*}}] [1, 1]
-//       TILE2:   return %[[RESULT]]
-
-
-// TILE1-LABEL: func @static_pad_tensor(
-//  TILE1-SAME:     %[[IN:.*]]: tensor<7x9xf32>
-//   TILE1-DAG:   %[[C0:.*]] = arith.constant 0 : index
-//   TILE1-DAG:   %[[C3:.*]] = arith.constant 3 : index
-//   TILE1-DAG:   %[[C16:.*]] = arith.constant 16 : index
-//       TILE1:   %[[RESULT:.*]] = scf.for {{.*}} = %[[C0]] to %[[C16]] step %[[C3]] iter_args(%[[INNER_OUT:.*]] =
-//       TILE1:     %[[SWAP_RESULT:.*]] = scf.if
-//       TILE1:       tensor.generate
-//       TILE1:     else
-//       TILE1:       %[[SLICE:.*]] = tensor.extract_slice %[[IN]][0, {{.*}}] [7, {{.*}}] [1, 1]
-//       TILE1:       %[[PAD:.*]] = tensor.pad %[[SLICE]] low[3, %{{.*}}] high[5, {{.*}}]
-//       TILE1:     tensor.insert_slice %[[SWAP_RESULT]] into %[[INNER_OUT]][0, {{.*}}] [15, {{.*}}] [1, 1]
-//       TILE1:   return %[[RESULT]]
+// CHECK-LABEL: func @static_pad_tensor_0_3(
+//  CHECK-SAME:     %[[IN:.*]]: tensor<7x9xf32>
+//   CHECK-DAG:   %[[C0:.*]] = arith.constant 0 : index
+//   CHECK-DAG:   %[[C3:.*]] = arith.constant 3 : index
+//   CHECK-DAG:   %[[C16:.*]] = arith.constant 16 : index
+//       CHECK:   %[[RESULT:.*]] = scf.for {{.*}} = %[[C0]] to %[[C16]] step %[[C3]] iter_args(%[[INNER_OUT:.*]] =
+//       CHECK:     %[[SWAP_RESULT:.*]] = scf.if
+//       CHECK:       tensor.generate
+//       CHECK:     else
+//       CHECK:       %[[SLICE:.*]] = tensor.extract_slice %[[IN]][0, {{.*}}] [7, {{.*}}] [1, 1]
+//       CHECK:       %[[PAD:.*]] = tensor.pad %[[SLICE]] low[3, %{{.*}}] high[5, {{.*}}]
+//       CHECK:     %[[CAST_SWAP_RESULT:.*]] = tensor.cast %[[SWAP_RESULT]] : tensor<?x?xf32> to tensor<15x?xf32> 
+//       CHECK:     tensor.insert_slice %[[CAST_SWAP_RESULT]] into %[[INNER_OUT]][0, {{.*}}] [15, {{.*}}] [1, 1]
+//       CHECK:   return %[[RESULT]]
 
-func.func @static_pad_tensor(%input_tensor: tensor<7x9xf32>,
+func.func @static_pad_tensor_0_3(%input_tensor: tensor<7x9xf32>,
                         %pad_value: f32) -> tensor<15x16xf32> {
   %0 = tensor.pad %input_tensor low[3, 4] high[5, 3] {
     ^bb0(%arg1: index, %arg2: index):
       tensor.yield %pad_value : f32
     } : tensor<7x9xf32> to tensor<15x16xf32>
   return %0 : tensor<15x16xf32>
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["tensor.pad"]} in %arg1
+    %1, %loop = transform.structured.tile_to_scf_for %0 [0, 3]
+}
+
 // -----
 
-// TILE1-LABEL: func @static_pad_tile_evenly(
-//  TILE1-SAME:     %[[IN:.*]]: tensor<7x9xf32>, %[[OUT:.*]]: tensor<14x15xf32>
-//   TILE1-DAG:   %[[C0:.*]] = arith.constant 0 : index
-//   TILE1-DAG:   %[[C3:.*]] = arith.constant 3 : index
-//   TILE1-DAG:   %[[C15:.*]] = arith.constant 15 : index
-//       TILE1:   %[[RESULT:.*]] = scf.for %[[IV:.*]] = %[[C0]] to %[[C15]] step %[[C3]] iter_args(%[[INNER_OUT:.*]] =
-//       TILE1:     %[[R2:.*]] = scf.if
-//       TILE1:       %[[GEN:.*]] = tensor.generate
-//       TILE1:       scf.yield %[[GEN]] : tensor<14x3xf32>
-//       TILE1:     else
-//       TILE1:       %[[SLICE:.*]] = tensor.extract_slice %arg0[0, %{{.*}}] [7, %{{.*}}] [1, 1] : tensor<7x9xf32> to tensor<7x?xf32>
-//       TILE1:       %[[PAD:.*]] = tensor.pad %[[SLICE]] low[0, 0] high[7, %{{.*}}]
-//       TILE1:       scf.yield %[[PAD]] : tensor<14x3xf32>
-//       TILE1:     %[[R3:.*]] = tensor.insert_slice %[[R2]] into %[[INNER_OUT]][0, %[[IV]]] [14, 3] [1, 1] : tensor<14x3xf32> into tensor<14x15xf32>
-//       TILE1:     scf.yield %[[R3]] : tensor<14x15xf32>
-//       TILE1:   return %[[RESULT]] : tensor<14x15xf32>
-func.func @static_pad_tile_evenly(%input_tensor: tensor<7x9xf32>,
+// CHECK-LABEL: func @static_pad_tile_evenly_0_3(
+//  CHECK-SAME:     %[[IN:.*]]: tensor<7x9xf32>, %[[OUT:.*]]: tensor<14x15xf32>
+//   CHECK-DAG:   %[[C0:.*]] = arith.constant 0 : index
+//   CHECK-DAG:   %[[C3:.*]] = arith.constant 3 : index
+//   CHECK-DAG:   %[[C15:.*]] = arith.constant 15 : index
+//       CHECK:   %[[RESULT:.*]] = scf.for %[[IV:.*]] = %[[C0]] to %[[C15]] step %[[C3]] iter_args(%[[INNER_OUT:.*]] =
+//       CHECK:     %[[R2:.*]] = scf.if
+//       CHECK:       %[[GEN:.*]] = tensor.generate
+//       CHECK:       %[[cast_0:.*]] = tensor.cast %[[GEN]] : tensor<14x3xf32> to tensor<?x?xf32>
+//       CHECK:       scf.yield %[[cast_0]] : tensor<?x?xf32>
+//       CHECK:     else
+//       CHECK:       %[[SLICE:.*]] = tensor.extract_slice %arg0[0, %{{.*}}] [7, %{{.*}}] [1, 1] : tensor<7x9xf32> to tensor<7x?xf32>
+//       CHECK:       %[[PAD:.*]] = tensor.pad %[[SLICE]] low[0, 0] high[7, %{{.*}}]
+//       CHECK:       %[[cast_1:.*]] = tensor.cast %[[PAD]] : tensor<14x?xf32> to tensor<?x?xf32>
+//       CHECK:       scf.yield %[[cast_1]] : tensor<?x?xf32>
+//       CHECK:     %[[cast:.*]] = tensor.cast %[[R2]] : tensor<?x?xf32> to tensor<14x3xf32>
+//       CHECK:     %[[R3:.*]] = tensor.insert_slice %[[cast]] into %[[INNER_OUT]][0, %[[IV]]] [14, 3] [1, 1] : tensor<14x3xf32> into tensor<14x15xf32>
+//       CHECK:     scf.yield %[[R3]] : tensor<14x15xf32>
+//       CHECK:   return %[[RESULT]] : tensor<14x15xf32>
+
+func.func @static_pad_tile_evenly_0_3(%input_tensor: tensor<7x9xf32>,
                              %output_tensor: tensor<14x15xf32>,
                              %pad_value: f32) -> tensor<14x15xf32> {
   %0 = tensor.pad %input_tensor low[0, 0] high[7, 6] {
     ^bb0(%arg1: index, %arg2: index):
       tensor.yield %pad_value : f32
     } : tensor<7x9xf32> to tensor<14x15xf32>
   return %0 : tensor<14x15xf32>
 }
+
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["tensor.pad"]} in %arg1
+    %1, %loop = transform.structured.tile_to_scf_for %0 [0, 3]
+}

mlir/test/Dialect/Linalg/tile-parallel-reduce.mlir

-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,4,8 loop-type=parallel" -split-input-file | FileCheck %s
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2 loop-type=parallel" -split-input-file | FileCheck %s -check-prefix=TILE1
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,4 loop-type=parallel" -split-input-file | FileCheck %s -check-prefix=TILE2
-
-func.func @gemm(%arg0 : memref<?x?xf32>,
-           %arg1 : memref<?x?xf32>,
-           %arg2 : memref<?x?xf32>)
-{
-  linalg.matmul ins(%arg0, %arg1: memref<?x?xf32>, memref<?x?xf32>)
-               outs(%arg2: memref<?x?xf32>)
-  return
-}
-// CHECK-LABEL: func @gemm
-//   CHECK-DAG:   %[[C2:.*]] = arith.constant 2 : index
-//   CHECK-DAG:   %[[C4:.*]] = arith.constant 4 : index
-//   CHECK-DAG:   %[[C8:.*]] = arith.constant 8 : index
-//       CHECK:   scf.parallel (%[[ARG3:.*]], %[[ARG4:.*]]) =
-//  CHECK-SAME:     step (%[[C2]], %[[C4]])
-//       CHECK:     scf.for %[[ARG5:.*]] =
-//  CHECK-SAME:       step %[[C8]]
-//       CHECK:       %[[SV1:.*]] = memref.subview %{{.*}}[%[[ARG3]], %[[ARG5]]]
-//       CHECK:       %[[SV2:.*]] = memref.subview %{{.*}}[%[[ARG5]], %[[ARG4]]]
-//       CHECK:       %[[SV3:.*]] = memref.subview %{{.*}}[%[[ARG3]], %[[ARG4]]]
-//       CHECK:       linalg.matmul ins(%[[SV1]], %[[SV2]]{{.*}} outs(%[[SV3]]
-
-// TILE1-LABEL: func @gemm
-//   TILE1-DAG:   %[[C2:.*]] = arith.constant 2 : index
-//       TILE1:   scf.parallel (%[[ARG3:.*]]) =
-//  TILE1-SAME:     step (%[[C2]])
-//       TILE1:     %[[SV1:.*]] = memref.subview %{{.*}}[%[[ARG3]], 0]
-//       TILE1:     %[[SV3:.*]] = memref.subview %{{.*}}[%[[ARG3]], 0]
-//   TILE1-NOT:     memref.subview
-//       TILE1:     linalg.matmul ins(%[[SV1]], %{{.*}} outs(%[[SV3]]
-
-// TILE2-LABEL: func @gemm
-//   TILE2-DAG:   %[[C2:.*]] = arith.constant 2 : index
-//   TILE2-DAG:   %[[C4:.*]] = arith.constant 4 : index
-//       TILE2:   scf.parallel (%[[ARG3:.*]], %[[ARG4:.*]]) =
-//  TILE2-SAME:     step (%[[C2]], %[[C4]])
-//       TILE2:       %[[SV1:.*]] = memref.subview %{{.*}}[%[[ARG3]], 0]
-//       TILE2:       %[[SV2:.*]] = memref.subview %{{.*}}[0, %[[ARG4]]]
-//       TILE2:       %[[SV3:.*]] = memref.subview %{{.*}}[%[[ARG3]], %[[ARG4]]]
-//       TILE2:       linalg.matmul ins(%[[SV1]], %[[SV2]]{{.*}} outs(%[[SV3]]
-
-// -----
-
-#map0 = affine_map<(d0, d1, d2) -> (d0, d1, d2)>
-#map1 = affine_map<(d0, d1, d2) -> (d0, d2)>
-#map2 = affine_map<(d0, d1, d2) -> (d1)>
-#accesses = [#map0, #map1, #map2]
-#trait = {
-  args_in = 2 : i64,
-  args_out = 1 : i64,
-  iterator_types = ["reduction", "parallel", "reduction"],
-  indexing_maps = #accesses
-}
-
-func.func @reduction(%arg0 : memref<?x?x?xf32>,
-                %arg1 : memref<?x?xf32>,
-                %arg2 : memref<?xf32>)
-{
-  linalg.generic #trait
-    ins(%arg0, %arg1 : memref<?x?x?xf32>, memref<?x?xf32>)
-   outs(%arg2 : memref<?xf32>) {
-  ^bb0(%arg3 : f32, %arg4 : f32, %arg5 : f32):
-    %0 = arith.addf %arg3, %arg4 : f32
-    %1 = arith.addf %0, %arg5 : f32
-    linalg.yield %1 : f32
-  }
-  return
-}
-
-// CHECK-LABEL: func @reduction
-//   CHECK-DAG:   %[[C2:.*]] = arith.constant 2 : index
-//   CHECK-DAG:   %[[C4:.*]] = arith.constant 4 : index
-//   CHECK-DAG:   %[[C8:.*]] = arith.constant 8 : index
-//       CHECK:   scf.for %[[ARG3:.*]] =
-//  CHECK-SAME:     step %[[C2]]
-//       CHECK:     scf.parallel (%[[ARG4:.*]]) =
-//  CHECK-SAME:       step (%[[C4]])
-//       CHECK:       scf.for %[[ARG5:.*]] =
-//  CHECK-SAME:         step %[[C8]]
-//       CHECK:         %[[SV1:.*]] = memref.subview %{{.*}}[%[[ARG3]], %[[ARG4]], %[[ARG5]]]
-//       CHECK:         %[[SV2:.*]] = memref.subview %{{.*}}[%[[ARG3]], %[[ARG5]]]
-//       CHECK:         %[[SV3:.*]] = memref.subview %{{.*}}[%[[ARG4]]]
-//       CHECK:         linalg.generic
-//  CHECK-SAME:           ins(%[[SV1]], %[[SV2]]
-//  CHECK-SAME:          outs(%[[SV3]]
-
-// TILE1-LABEL: func @reduction
-//   TILE1-DAG:   %[[C2:.*]] = arith.constant 2 : index
-//       TILE1:   scf.for %[[ARG3:.*]] =
-//  TILE1-SAME:     step %[[C2]]
-//       TILE1:         %[[SV1:.*]] = memref.subview %{{.*}}[%[[ARG3]], 0, 0]
-//       TILE1:         %[[SV2:.*]] = memref.subview %{{.*}}[%[[ARG3]], 0]
-//   TILE1-NOT:         memref.subview
-//       TILE1:         linalg.generic
-//  TILE1-SAME:           ins(%[[SV1]], %[[SV2]]
-//  TILE1-SAME:          outs(%{{.*}}
-
-// TILE2-LABEL: func @reduction
-//   TILE2-DAG:   %[[C2:.*]] = arith.constant 2 : index
-//   TILE2-DAG:   %[[C4:.*]] = arith.constant 4 : index
-//       TILE2:   scf.for %[[ARG3:.*]] =
-//  TILE2-SAME:     step %[[C2]]
-//       TILE2:     scf.parallel (%[[ARG4:.*]]) =
-//  TILE2-SAME:       step (%[[C4]])
-//       TILE2:         %[[SV1:.*]] = memref.subview %{{.*}}[%[[ARG3]], %[[ARG4]], 0]
-//       TILE2:         %[[SV2:.*]] = memref.subview %{{.*}}[%[[ARG3]], 0]
-//       TILE2:         %[[SV3:.*]] = memref.subview %{{.*}}[%[[ARG4]]]
-//       TILE2:         linalg.generic
-//  TILE2-SAME:           ins(%[[SV1]], %[[SV2]]
-//  TILE2-SAME:          outs(%[[SV3]]

mlir/test/Dialect/Linalg/tile-parallel.mlir

-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2 loop-type=parallel" | FileCheck %s -check-prefix=TILE-2
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=0,2 loop-type=parallel" | FileCheck %s -check-prefix=TILE-02
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=0,0,2 loop-type=parallel" | FileCheck %s -check-prefix=TILE-002
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,3,4 loop-type=parallel" | FileCheck %s -check-prefix=TILE-234
-
-#id_2d = affine_map<(i, j) -> (i, j)>
-#pointwise_2d_trait = {
-  args_in = 2,
-  args_out = 1,
-  indexing_maps = [#id_2d, #id_2d, #id_2d],
-  iterator_types = ["parallel", "parallel"]
-}
-
-func.func @sum(%lhs: memref<?x?xf32, strided<[?, 1], offset: ?>>,
-          %rhs: memref<?x?xf32, strided<[?, 1], offset: ?>>,
-          %sum: memref<?x?xf32, strided<[?, 1], offset: ?>>) {
-  linalg.generic #pointwise_2d_trait
-     ins(%lhs, %rhs: memref<?x?xf32, strided<[?, 1], offset: ?>>,
-                     memref<?x?xf32, strided<[?, 1], offset: ?>>)
-    outs(%sum : memref<?x?xf32, strided<[?, 1], offset: ?>>) {
-  ^bb0(%lhs_in: f32, %rhs_in: f32, %sum_out: f32):
-    %result = arith.addf %lhs_in, %rhs_in : f32
-    linalg.yield %result : f32
-  }
-  return
-}
-// TILE-2-LABEL: func @sum(
-// TILE-2-SAME:    [[LHS:%.*]]: memref{{.*}}, [[RHS:%.*]]: memref{{.*}}, [[SUM:%.*]]: memref{{.*}}) {
-// TILE-2-DAG: [[C0:%.*]] = arith.constant 0 : index
-// TILE-2-DAG: [[C2:%.*]] = arith.constant 2 : index
-// TILE-2: [[LHS_ROWS:%.*]] = memref.dim [[LHS]], %c0
-// TILE-2: scf.parallel ([[I:%.*]]) = ([[C0]]) to ([[LHS_ROWS]]) step ([[C2]]) {
-// TILE-2-NO: scf.parallel
-// TILE-2:   [[LHS_SUBVIEW:%.*]] = memref.subview [[LHS]]
-// TILE-2:   [[RHS_SUBVIEW:%.*]] = memref.subview [[RHS]]
-// TILE-2:   [[SUM_SUBVIEW:%.*]] = memref.subview [[SUM]]
-// TILE-2:   linalg.generic {{.*}} ins([[LHS_SUBVIEW]], [[RHS_SUBVIEW]]{{.*}} outs([[SUM_SUBVIEW]]
-
-// TILE-02-LABEL: func @sum(
-// TILE-02-SAME:    [[LHS:%.*]]: memref{{.*}}, [[RHS:%.*]]: memref{{.*}}, [[SUM:%.*]]: memref{{.*}}) {
-// TILE-02-DAG: [[C0:%.*]] = arith.constant 0 : index
-// TILE-02-DAG: [[C2:%.*]] = arith.constant 2 : index
-// TILE-02: [[LHS_COLS:%.*]] = memref.dim [[LHS]], %c1
-// TILE-02: scf.parallel ([[I:%.*]]) = ([[C0]]) to ([[LHS_COLS]]) step ([[C2]]) {
-// TILE-02-NO: scf.parallel
-// TILE-02:   [[LHS_SUBVIEW:%.*]] = memref.subview [[LHS]]
-// TILE-02:   [[RHS_SUBVIEW:%.*]] = memref.subview [[RHS]]
-// TILE-02:   [[SUM_SUBVIEW:%.*]] = memref.subview [[SUM]]
-// TILE-02:    linalg.generic {{.*}} ins([[LHS_SUBVIEW]], [[RHS_SUBVIEW]]{{.*}} outs([[SUM_SUBVIEW]]
-
-// TILE-002-LABEL: func @sum(
-// TILE-002-SAME:    [[LHS:%.*]]: memref{{.*}}, [[RHS:%.*]]: memref{{.*}}, [[SUM:%.*]]: memref{{.*}}) {
-// TILE-002-NO: scf.parallel
-// TILE-002:   linalg.generic {{.*}} ins([[LHS]], [[RHS]]{{.*}} outs([[SUM]]
-
-// TILE-234-LABEL: func @sum(
-// TILE-234-SAME:    [[LHS:%.*]]: memref{{.*}}, [[RHS:%.*]]: memref{{.*}}, [[SUM:%.*]]: memref{{.*}}) {
-// TILE-234-DAG: [[C0:%.*]] = arith.constant 0 : index
-// TILE-234-DAG: [[C2:%.*]] = arith.constant 2 : index
-// TILE-234-DAG: [[C3:%.*]] = arith.constant 3 : index
-// TILE-234: [[LHS_ROWS:%.*]] = memref.dim [[LHS]], %c0
-// TILE-234: [[LHS_COLS:%.*]] = memref.dim [[LHS]], %c1
-// TILE-234: scf.parallel ([[I:%.*]], [[J:%.*]]) = ([[C0]], [[C0]]) to ([[LHS_ROWS]], [[LHS_COLS]]) step ([[C2]], [[C3]]) {
-// TILE-234-NO: scf.parallel
-// TILE-234:   [[LHS_SUBVIEW:%.*]] = memref.subview [[LHS]]
-// TILE-234:   [[RHS_SUBVIEW:%.*]] = memref.subview [[RHS]]
-// TILE-234:   [[SUM_SUBVIEW:%.*]] = memref.subview [[SUM]]
-// TILE-234:   linalg.generic {{.*}} ins([[LHS_SUBVIEW]], [[RHS_SUBVIEW]]{{.*}} outs([[SUM_SUBVIEW]]

mlir/test/Dialect/Linalg/tile-scalarize-dynamic-dims.mlir

-// RUN: mlir-opt %s -test-linalg-transform-patterns="test-tile-scalarize-dynamic-dims" -scf-for-loop-canonicalization -canonicalize -split-input-file | \
-// RUN:     FileCheck %s
-
-// CHECK-LABEL: func @matmul_partly_dynamic_tensor(
-//  CHECK-SAME:     %[[ARG0:.*]]: tensor<?x?xf32>, %[[ARG1:.*]]: tensor<?x2000xf32>
-//   CHECK-DAG:   %[[C0:.*]] = arith.constant 0 : index
-//   CHECK-DAG:   %[[C1:.*]] = arith.constant 1 : index
-//       CHECK:   tensor.dim %[[ARG0]], %[[C0]] : tensor<?x?xf32>
-//       CHECK:   %[[UB1:.*]] = tensor.dim %[[ARG0]], %[[C0]] : tensor<?x?xf32>
-//       CHECK:   %[[UB2:.*]] = tensor.dim %[[ARG0]], %[[C1]] : tensor<?x?xf32>
-//       CHECK:   scf.for %[[IV0:.*]] = %[[C0]] to %[[UB1]] step %[[C1]]
-//       CHECK:     scf.for %[[IV1:.*]] = %[[C0]] to %[[UB2]] step %[[C1]]
-//       CHECK:       %[[S1:.*]] = tensor.extract_slice %[[ARG0]][%[[IV0]], %[[IV1]]] [1, 1] [1, 1] : tensor<?x?xf32> to tensor<1x1xf32>
-//       CHECK:       %[[S2:.*]] = tensor.extract_slice %[[ARG1]][%[[IV1]], 0] [1, 2000] [1, 1] : tensor<?x2000xf32> to tensor<1x2000xf32>
-//       CHECK:       %[[S3:.*]] = tensor.extract_slice %{{.*}}[%[[IV0]], 0] [1, 2000] [1, 1] : tensor<?x2000xf32> to tensor<1x2000xf32>
-//       CHECK:       linalg.matmul ins(%[[S1]], %[[S2]] : tensor<1x1xf32>, tensor<1x2000xf32>) outs(%[[S3]] : tensor<1x2000xf32>) -> tensor<1x2000xf32>
-func.func @matmul_partly_dynamic_tensor(%arg0: tensor<?x?xf32>, %arg1: tensor<?x2000xf32>)
-    -> tensor<?x2000xf32> {
-  %c0 = arith.constant 0 : index
-  %c1 = arith.constant 1 : index
-  %d0 = tensor.dim %arg0, %c0 : tensor<?x?xf32>
-  %out = tensor.empty(%d0) : tensor<?x2000xf32>
-  %r = linalg.matmul {__internal_linalg_transform__ = "tile"}
-      ins(%arg0, %arg1: tensor<?x?xf32>, tensor<?x2000xf32>)
-      outs(%out: tensor<?x2000xf32>) -> tensor<?x2000xf32>
-  return %r : tensor<?x2000xf32>
-}
-
-// -----
-
-// The input IR of this test case is a tiled and peeled linalg.matmul op.
-
-// CHECK-LABEL: func @tiled_and_peeled_matmul(
-//       CHECK:   linalg.matmul ins({{.*}} : tensor<32x259xf32>, tensor<259x258xf32>) outs({{.*}} : tensor<32x258xf32>) -> tensor<32x258xf32>
-//       CHECK:   linalg.matmul ins({{.*}} : tensor<1x259xf32>, tensor<259x258xf32>) outs({{.*}} : tensor<1x258xf32>) -> tensor<1x258xf32>
-#map0 = affine_map<(d0) -> (64, -d0 + 257)>
-#map1 = affine_map<()[s0] -> ((s0 floordiv 32) * 32)>
-#map2 = affine_map<(d0)[s0] -> (d0 - (s0 floordiv 32) * 32)>
-
-func.func @tiled_and_peeled_matmul(%arg0: tensor<257x259xf32>, %arg1: tensor<259x258xf32>, %arg2: tensor<257x258xf32>) -> tensor<257x258xf32> {
-  %c257 = arith.constant 257 : index
-  %c64 = arith.constant 64 : index
-  %cst = arith.constant 0.000000e+00 : f32
-  %c0 = arith.constant 0 : index
-  %c32 = arith.constant 32 : index
-  %0 = linalg.fill ins(%cst : f32) outs(%arg2 : tensor<257x258xf32>) -> tensor<257x258xf32>
-  %1 = scf.for %arg3 = %c0 to %c257 step %c64 iter_args(%arg4 = %0) -> (tensor<257x258xf32>) {
-    %2 = affine.min #map0(%arg3)
-    %3 = tensor.extract_slice %arg0[%arg3, 0] [%2, 259] [1, 1] : tensor<257x259xf32> to tensor<?x259xf32>
-    %4 = tensor.extract_slice %arg4[%arg3, 0] [%2, 258] [1, 1] : tensor<257x258xf32> to tensor<?x258xf32>
-    %5 = affine.apply #map1()[%2]
-    %6 = scf.for %arg5 = %c0 to %5 step %c32 iter_args(%arg6 = %4) -> (tensor<?x258xf32>) {
-      %10 = tensor.extract_slice %3[%arg5, 0] [32, 259] [1, 1] : tensor<?x259xf32> to tensor<32x259xf32>
-      %11 = tensor.extract_slice %arg6[%arg5, 0] [32, 258] [1, 1] : tensor<?x258xf32> to tensor<32x258xf32>
-      %12 = linalg.matmul {__internal_linalg_transform__ = "tile"} ins(%10, %arg1 : tensor<32x259xf32>, tensor<259x258xf32>) outs(%11 : tensor<32x258xf32>) -> tensor<32x258xf32>
-      %13 = tensor.insert_slice %12 into %arg6[%arg5, 0] [32, 258] [1, 1] : tensor<32x258xf32> into tensor<?x258xf32>
-      scf.yield %13 : tensor<?x258xf32>
-    }
-    %7 = arith.cmpi slt, %5, %2 : index
-    %8 = scf.if %7 -> (tensor<?x258xf32>) {
-      %10 = affine.apply #map2(%2)[%2]
-      %11 = tensor.extract_slice %3[%5, 0] [%10, 259] [1, 1] : tensor<?x259xf32> to tensor<?x259xf32>
-      %12 = tensor.extract_slice %6[%5, 0] [%10, 258] [1, 1] : tensor<?x258xf32> to tensor<?x258xf32>
-      %13 = linalg.matmul {__internal_linalg_transform__ = "tile"} ins(%11, %arg1 : tensor<?x259xf32>, tensor<259x258xf32>) outs(%12 : tensor<?x258xf32>) -> tensor<?x258xf32>
-      %14 = tensor.insert_slice %13 into %6[%5, 0] [%10, 258] [1, 1] : tensor<?x258xf32> into tensor<?x258xf32>
-      scf.yield %14 : tensor<?x258xf32>
-    } else {
-      scf.yield %6 : tensor<?x258xf32>
-    }
-    %9 = tensor.insert_slice %8 into %arg4[%arg3, 0] [%2, 258] [1, 1] : tensor<?x258xf32> into tensor<257x258xf32>
-    scf.yield %9 : tensor<257x258xf32>
-  }
-  return %1 : tensor<257x258xf32>
-}

mlir/test/Dialect/Linalg/tile-tensors.mlir

-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,3,4" -split-input-file | FileCheck %s
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -split-input-file | FileCheck %s
 
 // CHECK-LABEL: func @matmul_tensors(
 // CHECK-SAME:    %[[TA:[0-9a-z]+]]: tensor<?x?xf32>
 // CHECK-SAME:    %[[TB:[0-9a-z]+]]: tensor<?x?xf32>
 // CHECK-SAME:    %[[TC:[0-9a-z]+]]: tensor<?x?xf32>) -> tensor<?x?xf32> {
 func.func @matmul_tensors(
   %arg0: tensor<?x?xf32>, %arg1: tensor<?x?xf32>, %arg2: tensor<?x?xf32>)
     -> tensor<?x?xf32> {
//      CHECK:   scf.yield %[[TD1]] : tensor<?x?xf32>
   %0 = linalg.matmul  ins(%arg0, %arg1: tensor<?x?xf32>, tensor<?x?xf32>)
                      outs(%arg2: tensor<?x?xf32>)
     -> tensor<?x?xf32>
 
 //      CHECK: return %[[TD0]] : tensor<?x?xf32>
   return %0 : tensor<?x?xf32>
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.matmul"]} in %arg1
+    %1, %loops:3 = transform.structured.tile %0 [2, 3, 4]
+}
+
 // -----
 
 func.func @generic_op_tensors(
   %arg0 : tensor<?x?x?xf32>, %arg1 : tensor<?x?x?xf32>) -> tensor<?x?x?xf32> {
   %c0 = arith.constant 0 : index
   %c1 = arith.constant 1 : index
   %c2 = arith.constant 2 : index
   %0 = tensor.dim %arg0, %c0 : tensor<?x?x?xf32>
%4 = linalg.generic
     outs(%3 : tensor<?x?x?xf32>) {
     ^bb0(%arg2 : f32, %arg3: f32, %arg4: f32):
       %5 = arith.addf %arg2, %arg3 : f32
       linalg.yield %5 : f32
     } -> tensor<?x?x?xf32>
   return %4 : tensor<?x?x?xf32>
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.generic"]} in %arg1
+    %1, %loops:3 = transform.structured.tile %0 [2, 3, 4]
+}
+
 // CHECK-LABEL: func @generic_op_tensors
 //  CHECK-SAME:   %[[ARG0:[a-zA-Z0-9_]+]]: tensor<?x?x?xf32>
 //  CHECK-SAME:   %[[ARG1:[a-zA-Z0-9_]+]]: tensor<?x?x?xf32>
 //       CHECK:   %[[INIT:.+]] = tensor.empty
 //       CHECK:   %[[TD0:.+]] = scf.for %{{.+}} to %{{.+}} step %{{.+}} iter_args(%[[TC0:.+]] = %[[INIT]]) -> (tensor<?x?x?xf32>) {
 //       CHECK:     %[[TD1:.+]] = scf.for %{{.+}} to %{{.+}} step %{{.+}} iter_args(%[[TC1:.+]] = %[[TC0]]) -> (tensor<?x?x?xf32>) {
 //       CHECK:       %[[TD2:.+]] = scf.for %{{.+}} to %{{.+}} step %{{.+}} iter_args(%[[TC2:.+]] = %[[TC1]]) -> (tensor<?x?x?xf32>) {
 //       CHECK:       %[[STARG0:.+]] = tensor.extract_slice %[[ARG0]][{{.+}}] : tensor<?x?x?xf32> to tensor<?x?x?xf32>
%2 = linalg.generic
     outs(%arg1 : tensor<?x42xf32>) {
     ^bb0(%arg3 : f32, %arg4: f32, %arg5: f32):
       %5 = arith.addf %arg3, %arg5 : f32
       linalg.yield %5 : f32
     } -> tensor<?x42xf32>
   return %2 : tensor<?x42xf32>
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.generic"]} in %arg1
+    %1, %loops:3 = transform.structured.tile %0 [2, 3, 4]
+}

mlir/test/Dialect/Linalg/tile-zero.mlir

-// RUN: mlir-opt -test-linalg-transform-patterns=test-tile-pattern %s | FileCheck %s
-
-func.func @matmul_zero_tile(
-  %arg0: tensor<?x?xf32>, %arg1 : tensor<?x?xf32>, %arg2 : tensor<?x?xf32>) -> tensor<?x?xf32> {
-  %0 = linalg.matmul {__internal_linalg_transform__ = "tile"}
-      ins(%arg0, %arg1 : tensor<?x?xf32>, tensor<?x?xf32>)
-      outs(%arg2 : tensor<?x?xf32>) -> tensor<?x?xf32>
-  return %0 : tensor<?x?xf32>
-}
-// CHECK-LABEL: matmul_zero_tile
-//       CHECK:   linalg.matmul
-//   CHECK-NOT:   __internal_linalg_transform__

mlir/test/Dialect/Linalg/tile.mlir

-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2" -mlir-disable-threading=true | FileCheck %s -check-prefix=TILE-2
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=0,2" -mlir-disable-threading=true | FileCheck %s -check-prefix=TILE-02
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=0,0,2" -mlir-disable-threading=true | FileCheck %s -check-prefix=TILE-002
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,3,4" -mlir-disable-threading=true | FileCheck %s -check-prefix=TILE-234
-
-//   TILE-2-DAG: #[[$bound_map:.*]] = affine_map<(d0)[s0] -> (-d0 + s0, 2)>
-//  TILE-02-DAG: #[[$bound_map:.*]] = affine_map<(d0)[s0] -> (-d0 + s0, 2)>
-// TILE-002-DAG: #[[$bound_map:.*]] = affine_map<(d0)[s0] -> (-d0 + s0, 2)>
-// TILE-234-DAG: #[[$bound_map_2:.*]] = affine_map<(d0)[s0] -> (-d0 + s0, 2)>
-// TILE-234-DAG: #[[$bound_map_3:.*]] = affine_map<(d0)[s0] -> (-d0 + s0, 3)>
-// TILE-234-DAG: #[[$bound_map_4:.*]] = affine_map<(d0)[s0] -> (-d0 + s0, 4)>
-
-func.func @matmul(%arg0: memref<?x?xf32, strided<[?, 1], offset: ?>>,
-             %arg1: memref<?x?xf32, strided<[?, 1], offset: ?>>,
-             %arg2: memref<?x?xf32, strided<[?, 1], offset: ?>>) {
-  linalg.matmul
-    ins(%arg0, %arg1: memref<?x?xf32, strided<[?, 1], offset: ?>>,
-                      memref<?x?xf32, strided<[?, 1], offset: ?>>)
-   outs(%arg2: memref<?x?xf32, strided<[?, 1], offset: ?>>)
-  return
-}
-// TILE-2-LABEL: func @matmul(
-//       TILE-2-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-2-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-2: %[[M:.*]] = memref.dim %{{.*}}, %c0 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-2: scf.for %[[I:.*]] = %{{.*}}{{.*}} to %[[M]] step %{{.*}} {
-//       TILE-2:   %[[szM:.*]] = affine.min #[[$bound_map]](%[[I]])[%[[M]]]
-//       TILE-2:   %[[K:.*]] = memref.dim %{{.*}}, %c1 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-2:   %[[szK:.*]] = affine.min #[[$bound_map]](%[[I]])[%[[M]]]
-//       TILE-2:   %[[N:.*]] = memref.dim %{{.*}}, %c1 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-2:   %[[sAi:.*]] = memref.subview %{{.*}}[%[[I]], 0] [%[[szM]], %[[K]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-2:   %[[sCi:.*]] = memref.subview %{{.*}}[%[[I]], 0] [%[[szK]], %[[N]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-2:   linalg.matmul ins(%[[sAi]]{{.*}} outs(%[[sCi]]
-
-// TILE-02-LABEL: func @matmul(
-//       TILE-02-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-02-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-02: %[[N:.*]] = memref.dim %arg1, %c1 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-02: scf.for %[[J:.*]] = %{{.*}} to %[[N]] step %{{.*}} {
-//       TILE-02:   %[[K:.*]] = memref.dim %{{.*}}, %c0 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-02:   %[[szN:.*]] = affine.min #[[$bound_map]](%[[J]])[%[[N]]]
-//       TILE-02:   %[[M:.*]] = memref.dim %{{.*}}, %c0 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-02:   %[[szK:.*]] = affine.min #[[$bound_map]](%[[J]])[%[[N]]]
-//       TILE-02:   %[[sBj:.*]] = memref.subview %{{.*}}[0, %[[J]]] [%[[K]], %[[szN]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-02:   %[[sCj:.*]] = memref.subview %{{.*}}[0, %[[J]]] [%[[M]], %[[szK]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-02:   linalg.matmul ins(%{{.*}}, %[[sBj]]{{.*}} outs(%[[sCj]]
-
-// TILE-002-LABEL: func @matmul(
-//       TILE-002-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-002-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-002: %[[ubK:.*]] = memref.dim %{{.*}}, %c1 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-002: scf.for %[[K:.*]] = %{{.*}}{{.*}} to %[[ubK]] step %{{.*}} {
-//       TILE-002:   %[[M:.*]] = memref.dim %{{.*}}, %c0 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-002:   %[[szK:.*]] = affine.min #[[$bound_map]](%[[K]])[%[[ubK]]]
-//       TILE-002:   %[[szK_1:.*]] = affine.min #[[$bound_map]](%[[K]])[%[[ubK]]]
-//       TILE-002:   %[[N:.*]] = memref.dim %{{.*}}, %c1 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-002:   %[[sAj:.*]] = memref.subview %{{.*}}[0, %[[K]]] [%[[M]], %[[szK]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-002:   %[[sBj:.*]] = memref.subview %{{.*}}[%[[K]], 0] [%[[szK_1]], %[[N]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-002:   linalg.matmul ins(%[[sAj]], %[[sBj]]{{.*}} outs(%{{.*}}
-
-// TILE-234-LABEL: func @matmul(
-//       TILE-234-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-234-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-234-DAG: %[[C3:.*]] = arith.constant 3 : index
-//       TILE-234-DAG: %[[C4:.*]] = arith.constant 4 : index
-//       TILE-234: %[[ubM:.*]] = memref.dim %{{.*}}, %c0 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-234: %[[ubK:.*]] = memref.dim %{{.*}}, %c1 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-234: %[[ubN:.*]] = memref.dim %{{.*}}, %c1 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-234:  scf.for %[[I:.*]] = %{{.*}}{{.*}} to %[[ubM]] step %{{.*}} {
-//       TILE-234:    scf.for %[[J:.*]] = %{{.*}}{{.*}} to %[[ubN]] step %{{.*}} {
-//       TILE-234:      scf.for %[[K:.*]] = %{{.*}}{{.*}} to %[[ubK]] step %{{.*}} {
-//       TILE-234:        %[[szM:.*]] = affine.min #[[$bound_map_2]](%[[I]])[%[[ubM]]]
-//       TILE-234:        %[[szK:.*]] = affine.min #[[$bound_map_4]](%[[K]])[%[[ubK]]]
-//       TILE-234:        %[[szK_1:.*]] = affine.min #[[$bound_map_4]](%[[K]])[%[[ubK]]]
-//       TILE-234:        %[[szN:.*]] = affine.min #[[$bound_map_3]](%[[J]])[%[[ubN]]]
-//       TILE-234:        %[[szM_1:.*]] = affine.min #[[$bound_map_2]](%[[I]])[%[[ubM]]]
-//       TILE-234:        %[[szN_1:.*]] = affine.min #[[$bound_map_3]](%[[J]])[%[[ubN]]]
-//       TILE-234:        %[[sAik:.*]] = memref.subview %{{.*}}[%[[I]], %[[K]]] [%[[szM]], %[[szK]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-234:        %[[sBkj:.*]] = memref.subview %{{.*}}[%[[K]], %[[J]]] [%[[szK_1]], %[[szN]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-234:        %[[sCij:.*]] = memref.subview %{{.*}}[%[[I]], %[[J]]] [%[[szM_1]], %[[szN_1]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//
-//       TILE-234:        linalg.matmul ins(%[[sAik]], %[[sBkj]]{{.*}} outs(%[[sCij]]
-
-// When the buffer shapes are known at compile time, it is possible to avoid
-// the "min" in subview size computation. This test uses buffer sizes divisible
-// by respective tile sizes (M=10 divisble by 2, N=12 divisible by 2 and 3,
-// K=16 divisble by 2 and 4).
-func.func @matmul_static(%arg0: memref<10x16xf32, strided<[?, 1], offset: ?>>,
-                    %arg1: memref<16x12xf32, strided<[?, 1], offset: ?>>,
-                    %arg2: memref<10x12xf32, strided<[?, 1], offset: ?>>) {
-  linalg.matmul
-    ins(%arg0, %arg1: memref<10x16xf32, strided<[?, 1], offset: ?>>,
-                      memref<16x12xf32, strided<[?, 1], offset: ?>>)
-   outs(%arg2: memref<10x12xf32, strided<[?, 1], offset: ?>>)
-  return
-}
-// TILE-2-LABEL: func @matmul_static(
-//  TILE-2-SAME: %[[ARG0:[0-9a-zA-Z]*]]: memref
-//  TILE-2-SAME: %[[ARG1:[0-9a-zA-Z]*]]: memref
-//  TILE-2-SAME: %[[ARG2:[0-9a-zA-Z]*]]: memref
-//       TILE-2-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-2-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-2-DAG: %[[M:.*]] = arith.constant 10 : index
-//       TILE-2: scf.for %[[I:.*]] = %{{.*}} to %[[M]] step %{{.*}} {
-//       TILE-2:   %[[sAi:.*]] = memref.subview %{{.*}}[%[[I]], 0] [2, 16] [1, 1] : memref<10x16xf32, strided<[?, 1], offset: ?>> to memref<2x16xf32, strided<[?, 1], offset: ?>>
-//       TILE-2:   %[[sCi:.*]] = memref.subview %{{.*}}[%[[I]], 0] [2, 12] [1, 1] : memref<10x12xf32, strided<[?, 1], offset: ?>> to memref<2x12xf32, strided<[?, 1], offset: ?>>
-//       TILE-2:   linalg.matmul ins(%[[sAi]], %{{.*}}{{.*}} outs(%[[sCi]]
-
-// TILE-02-LABEL: func @matmul_static(
-//       TILE-02-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-02-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-02-DAG: %[[N:.*]] = arith.constant 12 : index
-//       TILE-02: scf.for %[[J:.*]] = %{{.*}} to %[[N]] step %{{.*}} {
-//       TILE-02:   %[[sBj:.*]] = memref.subview %{{.*}}[0, %[[J]]] [16, 2] [1, 1] : memref<16x12xf32, strided<[?, 1], offset: ?>> to memref<16x2xf32, strided<[?, 1], offset: ?>>
-//       TILE-02:   %[[sCj:.*]] = memref.subview %{{.*}}[0, %[[J]]] [10, 2] [1, 1] : memref<10x12xf32, strided<[?, 1], offset: ?>> to memref<10x2xf32, strided<[?, 1], offset: ?>>
-//       TILE-02:   linalg.matmul ins(%{{.*}}, %[[sBj]]{{.*}} outs(%[[sCj]]
-
-// TILE-002-LABEL: func @matmul_static(
-//       TILE-002-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-002-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-002-DAG: %[[C16:.*]] = arith.constant 16 : index
-//       TILE-002: scf.for %[[K:.*]] = %{{.*}}{{.*}} to %[[C16]] step %{{.*}} {
-//       TILE-002:   %[[sAj:.*]] = memref.subview %{{.*}}[0, %[[K]]] [10, 2] [1, 1] : memref<10x16xf32, strided<[?, 1], offset: ?>> to memref<10x2xf32, strided<[?, 1], offset: ?>>
-//       TILE-002:   %[[sBj:.*]] = memref.subview %{{.*}}[%[[K]], 0] [2, 12] [1, 1] : memref<16x12xf32, strided<[?, 1], offset: ?>> to memref<2x12xf32, strided<[?, 1], offset: ?>>
-//       TILE-002:   linalg.matmul ins(%[[sAj]], %[[sBj]]{{.*}} outs(%{{.*}}
-
-// TILE-234-LABEL: func @matmul_static(
-//       TILE-234-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-234-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-234-DAG: %[[C3:.*]] = arith.constant 3 : index
-//       TILE-234-DAG: %[[C4:.*]] = arith.constant 4 : index
-//       TILE-234-DAG: %[[C10:.*]] = arith.constant 10 : index
-//       TILE-234-DAG: %[[C16:.*]] = arith.constant 16 : index
-//       TILE-234-DAG: %[[C12:.*]] = arith.constant 12 : index
-//       TILE-234:  scf.for %[[I:.*]] = %{{.*}}{{.*}} to %[[C10]] step %{{.*}} {
-//       TILE-234:    scf.for %[[J:.*]] = %{{.*}}{{.*}} to %[[C12]] step %{{.*}} {
-//       TILE-234:      scf.for %[[K:.*]] = %{{.*}}{{.*}} to %[[C16]] step %{{.*}} {
-//       TILE-234:        %[[sAik:.*]] = memref.subview %{{.*}}[%[[I]], %[[K]]] [2, 4] [1, 1] : memref<10x16xf32, strided<[?, 1], offset: ?>> to memref<2x4xf32, strided<[?, 1], offset: ?>>
-//       TILE-234:        %[[sBkj:.*]] = memref.subview %{{.*}}[%[[K]], %[[J]]] [4, 3] [1, 1] : memref<16x12xf32, strided<[?, 1], offset: ?>> to memref<4x3xf32, strided<[?, 1], offset: ?>>
-//       TILE-234:        %[[sCij:.*]] = memref.subview %{{.*}}[%[[I]], %[[J]]] [2, 3] [1, 1] : memref<10x12xf32, strided<[?, 1], offset: ?>> to memref<2x3xf32, strided<[?, 1], offset: ?>>
-//
-//       TILE-234:        linalg.matmul ins(%[[sAik]], %[[sBkj]]{{.*}} outs(%[[sCij]]
-
-func.func @matvec(%arg0: memref<?x?xf32, strided<[?, 1], offset: ?>>, %arg1: memref<?xf32, strided<[1], offset: ?>>, %arg2: memref<?xf32, strided<[1], offset: ?>>) {
-  linalg.matvec
-    ins(%arg0, %arg1: memref<?x?xf32, strided<[?, 1], offset: ?>>,
-                      memref<?xf32, strided<[1], offset: ?>>)
-   outs(%arg2: memref<?xf32, strided<[1], offset: ?>>)
-  return
-}
-// TILE-2-LABEL: func @matvec(
-//  TILE-2-SAME: %[[ARG0:[0-9a-zA-Z]*]]: memref
-//  TILE-2-SAME: %[[ARG1:[0-9a-zA-Z]*]]: memref
-//  TILE-2-SAME: %[[ARG2:[0-9a-zA-Z]*]]: memref
-//       TILE-2-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-2-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-2: %[[M:.*]] = memref.dim %{{.*}}, %c0 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-2: scf.for %[[I:.*]] = %{{.*}}{{.*}} to %[[M]] step %{{.*}} {
-//       TILE-2:   %[[szM:.*]] = affine.min #[[$bound_map]](%[[I]])[%[[M]]]
-//       TILE-2:   %[[N:.*]] = memref.dim %{{.*}}, %c1 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-2:   %[[szN:.*]] = affine.min #[[$bound_map]](%[[I]])[%[[M]]]
-//       TILE-2:   %[[sAi:.*]] = memref.subview %{{.*}}[%[[I]], 0] [%[[szM]], %[[N]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-2:   %[[sCi:.*]] = memref.subview %{{.*}}[%[[I]]] [%[[szN]]] [1] : memref<?xf32, strided<[1], offset: ?>> to memref<?xf32, strided<[1], offset: ?>>
-//       TILE-2:   linalg.matvec ins(%[[sAi]], %{{.*}} outs(%[[sCi]]
-
-// TILE-02-LABEL: func @matvec(
-// TILE-02-SAME: %[[ARG0:[0-9a-zA-Z]*]]: memref
-// TILE-02-SAME: %[[ARG1:[0-9a-zA-Z]*]]: memref
-// TILE-02-SAME: %[[ARG2:[0-9a-zA-Z]*]]: memref
-//       TILE-02-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-02-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-02: %[[K:.*]] = memref.dim %{{.*}}, %c1 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-02: scf.for %[[J:.*]] = %{{.*}}{{.*}} to %[[K]] step %{{.*}} {
-//       TILE-02:   %[[M:.*]] = memref.dim %{{.*}}, %c0 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-02:   %[[szN:.*]] = affine.min #[[$bound_map]](%[[J]])[%[[K]]]
-//       TILE-02:   %[[szN_1:.*]] = affine.min #[[$bound_map]](%[[J]])[%[[K]]]
-//       TILE-02:   %[[sAj:.*]] = memref.subview %{{.*}}[0, %[[J]]] [%[[M]], %[[szN]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-02:   %[[sBj:.*]] = memref.subview %{{.*}}[%[[J]]] [%[[szN_1]]] [1] : memref<?xf32, strided<[1], offset: ?>> to memref<?xf32, strided<[1], offset: ?>>
-//       TILE-02:   linalg.matvec ins(%[[sAj]], %[[sBj]]{{.*}} outs(%{{.*}}
-
-// TILE-002-LABEL: func @matvec(
-// TILE-002-SAME: %[[ARG0:[0-9a-zA-Z]*]]: memref
-// TILE-002-SAME: %[[ARG1:[0-9a-zA-Z]*]]: memref
-// TILE-002-SAME: %[[ARG2:[0-9a-zA-Z]*]]: memref
-//   TILE-002-NOT: scf.for
-
-// TILE-234-LABEL: func @matvec(
-// TILE-234-SAME: %[[ARG0:[0-9a-zA-Z]*]]: memref
-// TILE-234-SAME: %[[ARG1:[0-9a-zA-Z]*]]: memref
-// TILE-234-SAME: %[[ARG2:[0-9a-zA-Z]*]]: memref
-//       TILE-234-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-234-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-234-DAG: %[[C3:.*]] = arith.constant 3 : index
-//       TILE-234: %[[M:.*]] = memref.dim %{{.*}}, %c0 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-234: %[[K:.*]] = memref.dim %{{.*}}, %c1 : memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-234:  scf.for %[[I:.*]] = %{{.*}}{{.*}} to %[[M]] step %{{.*}} {
-//       TILE-234:    scf.for %[[J:.*]] = %{{.*}}{{.*}} to %[[K]] step %{{.*}} {
-//       TILE-234:      %[[szM:.*]] = affine.min #[[$bound_map_2]](%[[I]])[%[[M]]]
-//       TILE-234:      %[[szN:.*]] = affine.min #[[$bound_map_3]](%[[J]])[%[[K]]]
-//       TILE-234:      %[[szN_1:.*]] = affine.min #[[$bound_map_3]](%[[J]])[%[[K]]]
-//       TILE-234:      %[[szM_1:.*]] = affine.min #[[$bound_map_2]](%[[I]])[%[[M]]]
-//       TILE-234:      %[[sAij:.*]] = memref.subview %{{.*}}[%[[I]], %[[J]]] [%[[szM]], %[[szN]]] [1, 1] : memref<?x?xf32, strided<[?, 1], offset: ?>> to memref<?x?xf32, strided<[?, 1], offset: ?>>
-//       TILE-234:      %[[sBj:.*]] = memref.subview %{{.*}}[%[[J]]] [%[[szN_1]]] [1] : memref<?xf32, strided<[1], offset: ?>> to memref<?xf32, strided<[1], offset: ?>>
-//       TILE-234:      %[[sCi:.*]] = memref.subview %{{.*}}[%[[I]]] [%[[szM_1]]] [1] : memref<?xf32, strided<[1], offset: ?>> to memref<?xf32, strided<[1], offset: ?>>
-//
-//       TILE-234:      linalg.matvec ins(%[[sAij]], %[[sBj]]{{.*}} outs(%[[sCi]]
-
-func.func @dot(%arg0: memref<?xf32, strided<[1], offset: ?>>, %arg1: memref<?xf32, strided<[1], offset: ?>>, %arg2: memref<f32>) {
-  linalg.dot
-    ins(%arg0, %arg1: memref<?xf32, strided<[1], offset: ?>>, memref<?xf32, strided<[1], offset: ?>>)
-   outs(%arg2: memref<f32>)
-  return
-}
-// TILE-2-LABEL: func @dot(
-//       TILE-2-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-2-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-2: %[[M:.*]] = memref.dim %{{.*}}, %c0 : memref<?xf32, strided<[1], offset: ?>>
-//       TILE-2: scf.for %[[I:.*]] = %{{.*}}{{.*}} to %[[M]] step %{{.*}} {
-//       TILE-2:   %[[szM:.*]] = affine.min #[[$bound_map]](%[[I]])[%[[M]]]
-//       TILE-2:   %[[szM_1:.*]] = affine.min #[[$bound_map]](%[[I]])[%[[M]]]
-//       TILE-2:   %[[sAi:.*]] = memref.subview %{{.*}}[%[[I]]] [%[[szM]]] [1] : memref<?xf32, strided<[1], offset: ?>> to memref<?xf32, strided<[1], offset: ?>>
-//       TILE-2:   %[[sBi:.*]] = memref.subview %{{.*}}[%[[I]]] [%[[szM_1]]] [1] : memref<?xf32, strided<[1], offset: ?>> to memref<?xf32, strided<[1], offset: ?>>
-//       TILE-2:   linalg.dot ins(%[[sAi]], %[[sBi]]{{.*}} outs(
-
-// TILE-02-LABEL: func @dot(
-//   TILE-02-NOT: scf.for
-
-// TILE-002-LABEL: func @dot(
-//   TILE-002-NOT: scf.for
-
-// TILE-234-LABEL: func @dot(
-//       TILE-234-DAG: %[[C0:.*]] = arith.constant 0 : index
-//       TILE-234-DAG: %[[C2:.*]] = arith.constant 2 : index
-//       TILE-234:  %[[ubK:.*]] = memref.dim %{{.*}}, %c0 : memref<?xf32, strided<[1], offset: ?>>
-//       TILE-234:  scf.for %[[I:.*]] = %{{.*}} to %[[ubK]] step %{{.*}} {
-//       TILE-234:    %[[szM:.*]] = affine.min #[[$bound_map_2]](%[[I]])[%[[ubK]]]
-//       TILE-234:    %[[szM_1:.*]] = affine.min #[[$bound_map_2]](%[[I]])[%[[ubK]]]
-//       TILE-234:    %[[sAi:.*]] = memref.subview %{{.*}}[%[[I]]] [%[[szM]]] [1] : memref<?xf32, strided<[1], offset: ?>> to memref<?xf32, strided<[1], offset: ?>>
-//       TILE-234:    %[[sBi:.*]] = memref.subview %{{.*}}[%[[I]]] [%[[szM_1]]] [1] : memref<?xf32, strided<[1], offset: ?>> to memref<?xf32, strided<[1], offset: ?>>
-//       TILE-234:    linalg.dot ins(%[[sAi]], %[[sBi]]{{.*}} outs(
-
-func.func @fill_static(%arg0: memref<127x99xf32>, %arg1: f32) {
-  linalg.fill ins(%arg1 : f32) outs(%arg0 : memref<127x99xf32>)
-  return
-}
-// TILE-2-LABEL: func @fill_static
-//       TILE-2:   for
-//   TILE-2-NOT:   for
-//       TILE-2:       memref.subview{{.*}} : memref<127x99xf32>
-//       TILE-2:       linalg.fill{{.*}} : memref<?x99xf32, strided<[99, 1], offset: ?>>
-
-// TILE-02-LABEL: func @fill_static
-//       TILE-02:   for
-//   TILE-02-NOT:   for
-//       TILE-02:       memref.subview{{.*}} : memref<127x99xf32>
-//       TILE-02:       linalg.fill{{.*}} : memref<127x?xf32, strided<[99, 1], offset: ?>>
-
-// TILE-002-LABEL: func @fill_static
-//   TILE-002-NOT:   for
-//       TILE-002:     linalg.fill{{.*}} : memref<127x99xf32>
-
-// TILE-234-LABEL: func @fill_static
-//       TILE-234:   for
-//       TILE-234:     for
-//   TILE-234-NOT:   for
-//       TILE-234:       memref.subview{{.*}} : memref<127x99xf32>
-//       TILE-234:       linalg.fill{{.*}} : memref<?x3xf32, strided<[99, 1], offset: ?>>
-
-
-func.func @fill(%arg0: memref<?x?xf32, strided<[?, 1], offset: ?>>, %arg1: f32) {
-  linalg.fill ins(%arg1 : f32) outs(%arg0 : memref<?x?xf32, strided<[?, 1], offset: ?>>)
-  return
-}
-// TILE-2-LABEL: func @fill
-//       TILE-2:   for
-//   TILE-2-NOT:   for
-//       TILE-2:   fill{{.*}} f32
-
-// TILE-02-LABEL: func @fill
-//       TILE-02:   for
-//   TILE-02-NOT:   for
-//       TILE-02:     fill{{.*}} f32
-
-// TILE-002-LABEL: func @fill
-//   TILE-002-NOT:   for
-//       TILE-002:     fill{{.*}} f32
-
-// TILE-234-LABEL: func @fill
-//       TILE-234:   for
-//       TILE-234:     for
-//   TILE-234-NOT:   for
-//       TILE-234:       fill{{.*}} f32
-
-#id_2d = affine_map<(i, j) -> (i, j)>
-#pointwise_2d_trait = {
-  args_in = 2,
-  args_out = 1,
-  indexing_maps = [#id_2d, #id_2d, #id_2d],
-  iterator_types = ["parallel", "parallel"]
-}
-
-func.func @pointwise(%arg0: memref<?x?xf32, strided<[?, 1], offset: ?>>, %arg1: memref<?x?xf32, strided<[?, 1], offset: ?>>,
-                %arg2: memref<?x?xf32, strided<[?, 1], offset: ?>>) {
-  linalg.generic #pointwise_2d_trait
-    ins(%arg0, %arg1 : memref<?x?xf32, strided<[?, 1], offset: ?>>, memref<?x?xf32, strided<[?, 1], offset: ?>>)
-    outs(%arg2 : memref<?x?xf32, strided<[?, 1], offset: ?>>) {
-  ^bb0(%arg4: f32, %arg5: f32, %arg6: f32):
-    %4 = arith.addf %arg4, %arg5 : f32
-    linalg.yield %4 : f32
-  }
-  return
-}
-// TILE-2-LABEL: func @pointwise
-//       TILE-2:   for
-//   TILE-2-NOT:   for
-//       TILE-2:   linalg.generic
-
-// TILE-02-LABEL: func @pointwise
-//       TILE-02:   for
-//   TILE-02-NOT:   for
-//       TILE-02:     linalg.generic
-
-// TILE-002-LABEL: func @pointwise
-//   TILE-002-NOT:   for
-//       TILE-002:     linalg.generic
-
-// TILE-234-LABEL: func @pointwise
-//       TILE-234:   for
-//       TILE-234:     for
-//   TILE-234-NOT:   for
-//       TILE-234:       linalg.generic

mlir/test/Dialect/Linalg/transform-patterns.mlir

-// RUN: mlir-opt %s -test-linalg-transform-patterns=test-patterns -split-input-file -test-transform-dialect-interpreter | FileCheck %s
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -test-linalg-transform-patterns=test-patterns -split-input-file | FileCheck %s
 
-// Map corresponding to a 2D memory access where the stride along the last dim is known to be 1.
-// CHECK-DAG: #[[$kn:.*]] = affine_map<(d0, d1, d2) -> (d2, d1)>
-// CHECK-DAG: #[[$nm:.*]] = affine_map<(d0, d1, d2) -> (d1, d0)>
-// CHECK-DAG: #[[$km:.*]] = affine_map<(d0, d1, d2) -> (d2, d0)>
+// -----
 
 func.func @dot(%x: memref<?xf32, strided<[1], offset: ?>>,
           %y: memref<?xf32, strided<[1], offset: ?>>,
           %v: memref<f32>) {
-  linalg.dot { __internal_linalg_transform__ = "MEM" }
-    ins(%x, %y: memref<?xf32, strided<[1], offset: ?>>,
+  linalg.dot ins(%x, %y: memref<?xf32, strided<[1], offset: ?>>,
                          memref<?xf32, strided<[1], offset: ?>>)
             outs(%v: memref<f32>)
-
   return
 }
+
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.dot"]} in %arg1
+    %1, %loop = transform.structured.tile %0 [8000]
+}
+
 // CHECK-LABEL: func @dot
 // CHECK-DAG:     %[[c0:.*]] = arith.constant 0 : index
 // CHECK-DAG:     %[[c1:.*]] = arith.constant 1 : index
 // CHECK-DAG:     %[[c8000:.*]] = arith.constant 8000 : index
 // CHECK:         scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c8000]] {
 // CHECK:             scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c1]] {
 // CHECK:               load
 // CHECK:               load
 // CHECK:               load
 // CHECK:               arith.mulf
 // CHECK:               arith.addf
 // CHECK:               store
 
+// -----
+
 func.func @matvec(%A: memref<?x?xf32, strided<[?, 1], offset: ?>>,
              %x: memref<?xf32, strided<[1], offset: ?>>,
              %y: memref<?xf32, strided<[1], offset: ?>>) {
   linalg.matvec
     ins(%A, %x: memref<?x?xf32, strided<[?, 1], offset: ?>>,
                 memref<?xf32, strided<[1], offset: ?>>)
     outs(%y: memref<?xf32, strided<[1], offset: ?>>)
   return
 }
+
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.matvec"]} in %arg1
+    %1, %loops:2 = transform.structured.tile %0 [5, 6]
+}
+
 // CHECK-LABEL: func @matvec
 // CHECK-DAG:     %[[c0:.*]] = arith.constant 0 : index
 // CHECK-DAG:     %[[c5:.*]] = arith.constant 5 : index
 // CHECK-DAG:     %[[c6:.*]] = arith.constant 6 : index
-// CHECK:         scf.parallel {{.*}} step (%[[c5]])
+// CHECK:         scf.for {{.*}} step %[[c5]]
 // CHECK:           scf.for {{.*}} step %[[c6]]
 // CHECK:             linalg.matvec
 // CHECK:               ins({{.*}}: memref<?x?xf32, strided<[?, 1], offset: ?>>, memref<?xf32, strided<[1], offset: ?>>)
 // CHECK:              outs({{.*}}: memref<?xf32, strided<[1], offset: ?>>)
 
+// -----
+
 func.func @matmul(%A: memref<?x?xf32, strided<[?, 1], offset: ?>>,
              %B: memref<?x?xf32, strided<[?, 1], offset: ?>>,
              %C: memref<?x?xf32, strided<[?, 1], offset: ?>>) {
-  linalg.matmul { __internal_linalg_transform__ = "MEM" }
-    ins(%A, %B: memref<?x?xf32, strided<[?, 1], offset: ?>>,
+  linalg.matmul ins(%A, %B: memref<?x?xf32, strided<[?, 1], offset: ?>>,
                             memref<?x?xf32, strided<[?, 1], offset: ?>>)
                outs(%C: memref<?x?xf32, strided<[?, 1], offset: ?>>)
   return
 }
+
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.matmul"]} in %arg1
+    %1, %loops:3 = transform.structured.tile %0 [2000, 3000, 4000]
+    %2, %loops_2:3 = transform.structured.tile %1 [200, 300, 400]
+    %3, %loops_3:3 = transform.structured.tile %2 [20, 30, 40]
+    %4, %loops_4:3 = transform.structured.tile %3 [2, 3, 4]
+}
+
 // CHECK-LABEL: func @matmul
 // CHECK-DAG:     %[[c0:.*]] = arith.constant 0 : index
 // CHECK-DAG:     %[[c2:.*]] = arith.constant 2 : index
 // CHECK-DAG:     %[[c3:.*]] = arith.constant 3 : index
 // CHECK-DAG:     %[[c4:.*]] = arith.constant 4 : index
 // CHECK-DAG:     %[[c20:.*]] = arith.constant 20 : index
 // CHECK-DAG:     %[[c30:.*]] = arith.constant 30 : index
 // CHECK-DAG:     %[[c40:.*]] = arith.constant 40 : index

 // CHECK:                         scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c40]] {
 // CHECK:                           scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c2]] {
 // CHECK:                             scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c3]] {
 // CHECK:                               scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c4]] {
 // CHECK:                                 linalg.matmul
 // CHECK:                                   ins({{.*}}: memref<?x?xf32, strided<[?, 1], offset: ?>>, memref<?x?xf32, strided<[?, 1], offset: ?>>)
 // CHECK:                                  outs({{.*}}: memref<?x?xf32, strided<[?, 1], offset: ?>>)
 
+// -----
+
+// Map corresponding to a 2D memory access where the stride along the last dim is known to be 1.
+// CHECK-DAG: #[[$kn:.*]] = affine_map<(d0, d1, d2) -> (d2, d1)>
+// CHECK-DAG: #[[$nm:.*]] = affine_map<(d0, d1, d2) -> (d1, d0)>
+// CHECK-DAG: #[[$km:.*]] = affine_map<(d0, d1, d2) -> (d2, d0)>
+
 #matmul_accesses = [
   affine_map<(m, n, k) -> (m, k)>,
   affine_map<(m, n, k) -> (k, n)>,
   affine_map<(m, n, k) -> (m, n)>
 ]
 #generic_matmul_trait = {
   args_in = 2,
   args_out = 1,
func.func @permute_generic(%A: memref<?x?xf32, strided<[?, 1], offset: ?>>,
    outs(%C : memref<?x?xf32, strided<[?, 1], offset: ?>>) {
     ^bb(%a: f32, %b: f32, %c: f32):
       %d = arith.mulf %a, %b: f32
       %e = arith.addf %c, %d: f32
       linalg.yield %e: f32
   }
   return
 }
+
 transform.with_pdl_patterns {
 ^bb0(%arg0: !pdl.operation):
   transform.sequence %arg0 failures(propagate) {
   ^bb1(%arg1: !pdl.operation):
     %0 = transform.structured.match ops{["linalg.generic"]} in %arg1
     transform.structured.interchange %0 { iterator_interchange = [1, 2, 0]}
   }
 }
+
 // CHECK-LABEL:  func @permute_generic
 // CHECK:        linalg.generic {
 // CHECK-SAME:   indexing_maps = [#[[$kn]], #[[$nm]], #[[$km]]],
 // CHECK-SAME:   iterator_types = ["parallel", "reduction", "parallel"],
 // CHECK-SAME:   library_call = "linalg_matmul"}
 // CHECK:          memref<?x?xf32, strided<[?, 1], offset: ?>>,
 // CHECK-SAME:     memref<?x?xf32, strided<[?, 1], offset: ?>>
 // CHECK-SAME:     memref<?x?xf32, strided<[?, 1], offset: ?>>
 
+// -----
+
 func.func @matvec_perm(%A: memref<?x?xf32, strided<[?, 1], offset: ?>>,
              %x: memref<?xf32, strided<[1], offset: ?>>,
              %y: memref<?xf32, strided<[1], offset: ?>>) {
-  linalg.matvec {__internal_linalg_transform__ = "__with_perm__"}
-    ins(%A, %x: memref<?x?xf32, strided<[?, 1], offset: ?>>,
+  linalg.matvec ins(%A, %x: memref<?x?xf32, strided<[?, 1], offset: ?>>,
                             memref<?xf32, strided<[1], offset: ?>>)
                outs(%y: memref<?xf32, strided<[1], offset: ?>>)
   return
 }
+
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.matvec"]} in %arg1
+    %1, %loops:2 = transform.structured.tile %0 [5, 6] {interchange = [1, 0]}
+}
+
 // CHECK-LABEL: func @matvec_perm
 // CHECK-DAG:     %[[c0:.*]] = arith.constant 0 : index
 // CHECK-DAG:     %[[c5:.*]] = arith.constant 5 : index
 // CHECK-DAG:     %[[c6:.*]] = arith.constant 6 : index
 // CHECK:         scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c6]]
 // CHECK:           scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c5]]
 // CHECK:             linalg.matvec
 // CHECK:               ins({{.*}}: memref<?x?xf32, strided<[?, 1], offset: ?>>, memref<?xf32, strided<[1], offset: ?>>)
 // CHECK:              outs({{.*}}: memref<?xf32, strided<[1], offset: ?>>)
 
+// -----
+
 func.func @matmul_perm(%A: memref<?x?xf32, strided<[?, 1], offset: ?>>,
              %B: memref<?x?xf32, strided<[?, 1], offset: ?>>,
              %C: memref<?x?xf32, strided<[?, 1], offset: ?>>) {
-  linalg.matmul {__internal_linalg_transform__ = "__with_perm__"}
-    ins(%A, %B: memref<?x?xf32, strided<[?, 1], offset: ?>>,
+  linalg.matmul ins(%A, %B: memref<?x?xf32, strided<[?, 1], offset: ?>>,
                             memref<?x?xf32, strided<[?, 1], offset: ?>>)
                outs(%C : memref<?x?xf32, strided<[?, 1], offset: ?>>)
   return
 }
+
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.matmul"]} in %arg1
+    %1, %loops:3 = transform.structured.tile %0 [2000, 3000, 4000] {interchange=[1, 2, 0]}
+    %2, %loops_2:3 = transform.structured.tile %1 [200, 300, 400] {interchange=[1, 0, 2]}
+    %3, %loops_3:3 = transform.structured.tile %2 [20, 30, 40]
+}
+
 // CHECK-LABEL: func @matmul_perm
 // CHECK-DAG:     %[[c0:.*]] = arith.constant 0 : index
 // CHECK-DAG:     %[[c20:.*]] = arith.constant 20 : index
 // CHECK-DAG:     %[[c30:.*]] = arith.constant 30 : index
 // CHECK-DAG:     %[[c40:.*]] = arith.constant 40 : index
 // CHECK-DAG:     %[[c200:.*]] = arith.constant 200 : index
 // CHECK-DAG:     %[[c300:.*]] = arith.constant 300 : index
 // CHECK-DAG:     %[[c400:.*]] = arith.constant 400 : index
 // CHECK-DAG:     %[[c2000:.*]] = arith.constant 2000 : index
 // CHECK-DAG:     %[[c3000:.*]] = arith.constant 3000 : index
 // CHECK-DAG:     %[[c4000:.*]] = arith.constant 4000 : index
 // CHECK:         scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c3000]] {
 // CHECK:           scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c4000]] {
 // CHECK:             scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c2000]] {
 // CHECK:               scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c300]] {
 // CHECK:                 scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c200]] {
 // CHECK:                   scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c400]] {
 // CHECK:                     scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c20]] {
 // CHECK:                       scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c30]] {
 // CHECK:                         scf.for {{.*}} = %[[c0]] to {{.*}} step %[[c40]] {
 // CHECK:                                 linalg.matmul
 // CHECK:                                  ins({{.*}}: memref<?x?xf32, strided<[?, 1], offset: ?>>, memref<?x?xf32, strided<[?, 1], offset: ?>>)
 // CHECK:                                   outs({{.*}}: memref<?x?xf32, strided<[?, 1], offset: ?>>)
-
-func.func @tile_permute_parallel_loop(%arg0: memref<?x?xf32>,
-                                 %arg1: memref<?x?xf32>,
-                                 %arg2: memref<?x?xf32>) {
-  linalg.matmul {__internal_linalg_transform__ = "par__with_perm__"}
-    ins(%arg0, %arg1: memref<?x?xf32>, memref<?x?xf32>)
-   outs(%arg2: memref<?x?xf32>)
-  return
-}
-// CHECK-LABEL: func @tile_permute_parallel_loop
-//  CHECK-SAME:   %[[ARG0:[a-zA-Z0-9_]+]]: memref<?x?xf32>
-//  CHECK-SAME:   %[[ARG1:[a-zA-Z0-9_]+]]: memref<?x?xf32>
-//  CHECK-SAME:   %[[ARG2:[a-zA-Z0-9_]+]]: memref<?x?xf32>
-//   CHECK-DAG:   %[[C16:.*]] = arith.constant 16 : index
-//   CHECK-DAG:   %[[C8:.*]] = arith.constant 8 : index
-//   CHECK-DAG:   %[[C4:.*]] = arith.constant 4 : index
-//   CHECK-DAG:   %[[C0:.*]] = arith.constant 0 : index
-//   CHECK-DAG:   %[[D0:.*]] = memref.dim %[[ARG0]], %c0
-//   CHECK-DAG:   %[[D1:.*]] = memref.dim %[[ARG0]], %c1
-//   CHECK-DAG:   %[[D2:.*]] = memref.dim %[[ARG1]], %c1
-//       CHECK:   scf.parallel (%{{.*}}) = (%[[C0]]) to (%[[D2]]) step (%[[C8]])
-//       CHECK:     scf.for %{{.*}} = %[[C0]] to %[[D1]] step %[[C4]]
-//       CHECK:       scf.parallel (%{{.*}}) = (%[[C0]]) to (%[[D0]]) step (%[[C16]])

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

-// RUN: mlir-opt %s -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: mlir-opt %s -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=4" -convert-linalg-to-loops -convert-scf-to-cf \
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf \
 // RUN:   -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
 func.func private @printMemrefF32(memref<*xf32>)
 
 // Creates and returns a 1-D buffer of size %s1 filled with the value %f
 func.func @alloc_1d_filled_f32(%s1 : index, %f : f32) -> memref<?xf32> {
   %buf = memref.alloc(%s1) : memref<?xf32>
   linalg.fill ins(%f : f32) outs(%buf : memref<?xf32>)
   return %buf : memref<?xf32>
 }
 
 func.func @conv_1d(%arg0: memref<?xf32>, %arg1: memref<?xf32>, %arg2: memref<?xf32>) {
   linalg.conv_1d ins (%arg0, %arg1: memref<?xf32>, memref<?xf32>)
                 outs (%arg2: memref<?xf32>)
   return
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.conv_1d"]} in %arg1
+    %1, %loop = transform.structured.tile %0 [4]
+}
+
 func.func @main() {
   %c3 = arith.constant 3 : index
   %c6 = arith.constant 6 : index
   %c8 = arith.constant 8 : index
   %f10 = arith.constant 10.00000e+00 : f32
   %val = arith.constant 2.00000e+00 : f32
   %zero = arith.constant 0.00000e+00 : f32
 

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

-// RUN: mlir-opt %s -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: mlir-opt %s -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,4" -convert-linalg-to-loops -convert-scf-to-cf \
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf \
 // RUN:   -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
 func.func private @printMemrefF32(memref<*xf32>)
 
 // Creates and returns 3-D buffer of size (%s1, %s2, %s3) filled with the value %f
 func.func @alloc_3d_filled_f32(%s1 : index, %s2 : index, %s3 : index, %f : f32) -> memref<?x?x?xf32> {
   %buf = memref.alloc(%s1, %s2, %s3) : memref<?x?x?xf32>
   linalg.fill ins(%f : f32) outs(%buf : memref<?x?x?xf32>)
   return %buf : memref<?x?x?xf32>
 }
 
 func.func @conv_1d_nwc_wcf(%arg0: memref<?x?x?xf32>, %arg1: memref<?x?x?xf32>, %arg2: memref<?x?x?xf32>) {
   linalg.conv_1d_nwc_wcf {dilations = dense<1> : tensor<1xi64>,
                          strides = dense<1> : tensor<1xi64>}
      ins (%arg0, %arg1: memref<?x?x?xf32>, memref<?x?x?xf32>)
     outs (%arg2: memref<?x?x?xf32>)
   return
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.conv_1d_nwc_wcf"]} in %arg1
+    %1, %loops:2 = transform.structured.tile %0 [2, 4]
+}
+
 func.func @main() {
   %c0 = arith.constant 0 : index
   %c1 = arith.constant 1 : index
   %c3 = arith.constant 3 : index
   %c6 = arith.constant 6 : index
   %c8 = arith.constant 8 : index
   %f10 = arith.constant 10.00000e+00 : f32
   %val = arith.constant 2.00000e+00 : f32

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

-// RUN: mlir-opt %s -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: mlir-opt %s -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,2" -convert-linalg-to-loops -convert-scf-to-cf \
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf \
 // RUN:   -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
 func.func private @printMemrefF32(memref<*xf32>)
 
 // Creates and returns a 2-D buffer of size (%s1, %s2) filled with the value %f
 func.func @alloc_2d_filled_f32(%s1 : index, %s2 : index, %f : f32) -> memref<?x?xf32> {
   %buf = memref.alloc(%s1, %s2) : memref<?x?xf32>
   linalg.fill ins(%f : f32) outs(%buf : memref<?x?xf32>)
   return %buf : memref<?x?xf32>
 }
 
 func.func @conv_2d(%arg0: memref<?x?xf32>, %arg1: memref<?x?xf32>, %arg2: memref<?x?xf32>) {
   linalg.conv_2d ins (%arg0, %arg1: memref<?x?xf32>, memref<?x?xf32>)
                 outs (%arg2: memref<?x?xf32>)
   return
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.conv_2d"]} in %arg1
+    %1, %loops:2 = transform.structured.tile %0 [2, 2]
+}
+
 func.func @main() {
   %c0 = arith.constant 0 : index
   %c1 = arith.constant 1 : index
   %c3 = arith.constant 3 : index
   %c6 = arith.constant 6 : index
   %c8 = arith.constant 8 : index
   %f10 = arith.constant 10.00000e+00 : f32
   %val = arith.constant 2.00000e+00 : f32

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

-// RUN: mlir-opt %s -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: mlir-opt %s -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,3,3,2" -convert-linalg-to-loops -convert-scf-to-cf \
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf \
 // RUN:   -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
 func.func private @printMemrefF32(memref<*xf32>)
 
 // Creates and returns 4-D buffer of size (%s1, %s2, %s3, %s4) filled with the value %f
 func.func @alloc_4d_filled_f32(%s1 : index, %s2 : index, %s3 : index, %s4 : index, %f : f32) -> memref<?x?x?x?xf32> {
   %buf = memref.alloc(%s1, %s2, %s3, %s4) : memref<?x?x?x?xf32>
   linalg.fill ins(%f : f32) outs(%buf : memref<?x?x?x?xf32>)
   return %buf : memref<?x?x?x?xf32>
 }
 
 func.func @conv_2d_nhwc_hwcf(%arg0: memref<?x?x?x?xf32>, %arg1: memref<?x?x?x?xf32>, %arg2: memref<?x?x?x?xf32>) {
   linalg.conv_2d_nhwc_hwcf {dilations = dense<1> : tensor<2xi64>,
                           strides = dense<1> : tensor<2xi64>}
      ins (%arg0, %arg1: memref<?x?x?x?xf32>, memref<?x?x?x?xf32>)
     outs (%arg2: memref<?x?x?x?xf32>)
   return
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.conv_2d_nhwc_hwcf"]} in %arg1
+    %1, %loops:4 = transform.structured.tile %0 [2, 3, 3, 2]
+}
+
 func.func @main() {
   %c0 = arith.constant 0 : index
   %c1 = arith.constant 1 : index
   %c3 = arith.constant 3 : index
   %c6 = arith.constant 6 : index
   %c8 = arith.constant 8 : index
   %f10 = arith.constant 10.00000e+00 : f32
   %val = arith.constant 2.00000e+00 : f32

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

-// RUN: mlir-opt %s -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: mlir-opt %s -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=2,2,2" -convert-linalg-to-loops -convert-scf-to-cf \
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf \
 // RUN:   -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
 func.func private @printMemrefF32(memref<*xf32>)
 
 // Creates and returns 3-D buffer of size (%s1, %s2, %s3) filled with the value %f
 func.func @alloc_3d_filled_f32(%s1 : index, %s2 : index, %s3 : index, %f : f32) -> memref<?x?x?xf32> {
   %buf = memref.alloc(%s1, %s2, %s3) : memref<?x?x?xf32>
   linalg.fill ins(%f : f32) outs(%buf : memref<?x?x?xf32>)
   return %buf : memref<?x?x?xf32>
 }
 
 func.func @conv_3d(%arg0: memref<?x?x?xf32>, %arg1: memref<?x?x?xf32>, %arg2: memref<?x?x?xf32>) {
   linalg.conv_3d ins (%arg0, %arg1: memref<?x?x?xf32>, memref<?x?x?xf32>)
                 outs (%arg2: memref<?x?x?xf32>)
   return
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.conv_3d"]} in %arg1
+    %1, %loops:3 = transform.structured.tile %0 [2, 2, 2]
+}
+
 func.func @main() {
   %c0 = arith.constant 0 : index
   %c1 = arith.constant 1 : index
   %c3 = arith.constant 3 : index
   %c6 = arith.constant 6 : index
   %c8 = arith.constant 8 : index
   %f10 = arith.constant 10.00000e+00 : f32
   %val = arith.constant 2.00000e+00 : f32

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

-// RUN: mlir-opt %s -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
+// RUN: mlir-opt %s -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s -linalg-tile="tile-sizes=0,5,5,5" -convert-linalg-to-loops -convert-scf-to-cf \
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -test-transform-dialect-erase-schedule -convert-linalg-to-loops -convert-scf-to-cf \
 // RUN:   -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
 func.func private @printMemrefF32(memref<*xf32>)
 
 // Creates and returns 5-D buffer of size (%s1, %s2, %s3, %s4, %s5) filled with the value %f
 func.func @alloc_5d_filled_f32(%s1 : index, %s2 : index, %s3 : index, %s4 : index, %s5 : index, %f : f32) -> memref<?x?x?x?x?xf32> {
   %buf = memref.alloc(%s1, %s2, %s3, %s4, %s5) : memref<?x?x?x?x?xf32>
   linalg.fill ins(%f : f32) outs(%buf : memref<?x?x?x?x?xf32>)
   return %buf : memref<?x?x?x?x?xf32>
 }
 
 func.func @conv_3d_ndhwc_dhwcf(%arg0: memref<?x?x?x?x?xf32>, %arg1: memref<?x?x?x?x?xf32>, %arg2: memref<?x?x?x?x?xf32>) {
   linalg.conv_3d_ndhwc_dhwcf {dilations = dense<1> : tensor<3xi64>,
                            strides = dense<1> : tensor<3xi64>}
      ins (%arg0, %arg1: memref<?x?x?x?x?xf32>, memref<?x?x?x?x?xf32>)
     outs (%arg2: memref<?x?x?x?x?xf32>)
   return
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.conv_3d_ndhwc_dhwcf"]} in %arg1
+    %1, %loops:3 = transform.structured.tile %0 [0, 5, 5, 5]
+}
 
 func.func @main() {
   %c0 = arith.constant 0 : index
   %c1 = arith.constant 1 : index
   %c3 = arith.constant 3 : index
   %c6 = arith.constant 6 : index
   %c8 = arith.constant 8 : index
   %f10 = arith.constant 10.00000e+00 : f32

mlir/test/Integration/Dialect/Linalg/CPU/test-tensor-matmul.mlir

 // UNSUPPORTED: asan
-// RUN: mlir-opt %s -linalg-bufferize -arith-bufferize \
+// RUN: mlir-opt %s -test-transform-dialect-erase-schedule -linalg-bufferize -arith-bufferize \
 // RUN: -tensor-bufferize -func-bufferize -finalizing-bufferize -buffer-deallocation -convert-linalg-to-loops -convert-scf-to-cf \
 // RUN: -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext,%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
-// RUN: mlir-opt %s  -linalg-tile="tile-sizes=1,2,3" -linalg-bufferize \
+// RUN: mlir-opt %s -test-transform-dialect-interpreter -test-transform-dialect-erase-schedule -linalg-bufferize \
 // RUN: -scf-bufferize -arith-bufferize -tensor-bufferize \
 // RUN: -func-bufferize \
 // RUN: -finalizing-bufferize -convert-linalg-to-loops -convert-scf-to-cf -convert-scf-to-cf \
 // RUN: -convert-linalg-to-llvm -lower-affine -convert-scf-to-cf --convert-memref-to-llvm -convert-func-to-llvm -reconcile-unrealized-casts | \
 // RUN: mlir-cpu-runner -e main -entry-point-result=void \
 // RUN:   -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext,%mlir_lib_dir/libmlir_runner_utils%shlibext \
 // RUN: | FileCheck %s
 
func.func @main() {
   //      CHECK: Unranked Memref base@ = {{0x[-9a-f]*}}
   // CHECK-SAME: rank = 2 offset = 0 sizes = [2, 4] strides = [4, 1] data =
   // CHECK-NEXT: [1038,   1044,   1050,   1056]
   // CHECK-NEXT: [1083,   1098,   1113,   1128]
 
   return
 }
 
+transform.sequence failures(propagate) {
+  ^bb0(%arg1: !pdl.operation):
+    %0 = transform.structured.match ops{["linalg.matmul"]} in %arg1
+    %1, %loops:3 = transform.structured.tile %0 [1, 2, 3]
+}
+
 func.func private @printMemrefF32(%ptr : tensor<*xf32>)

mlir/test/lib/Dialect/Linalg/TestLinalgTransforms.cpp

StringRef getDescription() const final {
     return "Test Linalg transformation patterns by applying them greedily.";
   }
 
   void runOnOperation() override;
 
   Option<bool> testPatterns{*this, "test-patterns",
                             llvm::cl::desc("Test a mixed set of patterns"),
                             llvm::cl::init(false)};
-  Option<bool> testTileAndDistributionOptions{
-      *this, "test-tile-and-distribute-options",
-      llvm::cl::desc("Test tile and distribute options"),
-      llvm::cl::init(false)};
   Option<bool> testVectorTransferForwardingPatterns{
       *this, "test-vector-transfer-forwarding-patterns",
       llvm::cl::desc(
           "Test a fused pass that forwards memref.copy to vector.transfer"),
       llvm::cl::init(false)};
   Option<bool> testGenericToVectorPattern{
       *this, "test-linalg-to-vector-patterns",
       llvm::cl::desc("Test a set of patterns that rewrite a linalg contraction "
                      "in vector.contract form"),
       llvm::cl::init(false)};
-  Option<bool> testTilePattern{*this, "test-tile-pattern",
-                               llvm::cl::desc("Test tile pattern"),
-                               llvm::cl::init(false)};
-  Option<bool> testTileScalarizeDynamicDims{
-      *this, "test-tile-scalarize-dynamic-dims",
-      llvm::cl::desc("Test tiling of dynamic dims by 1"),
-      llvm::cl::init(false)};
   Option<bool> testTransformPadTensor{
       *this, "test-transform-pad-tensor",
       llvm::cl::desc("Test transform pad tensor by copying with generic ops"),
       llvm::cl::init(false)};
   Option<bool> testGeneralizePadTensor{
       *this, "test-generalize-pad-tensor",
       llvm::cl::desc("Test transform pad tensor by copying with generic ops"),
       llvm::cl::init(false)};

 };
 } // namespace
 
 static void applyPatterns(func::FuncOp funcOp) {
   MLIRContext *ctx = funcOp.getContext();
   RewritePatternSet patterns(ctx);
 
   //===--------------------------------------------------------------------===//
-  // Linalg tiling patterns.
-  //===--------------------------------------------------------------------===//
-  patterns.add<LinalgTilingPattern>(
-      MatmulOp::getOperationName(), ctx,
-      LinalgTilingOptions().setTileSizes({2000, 3000, 4000}),
-      LinalgTransformationFilter(StringAttr::get(ctx, "MEM"),
-                                 StringAttr::get(ctx, "L3")));
-  patterns.add<LinalgTilingPattern>(
-      MatmulOp::getOperationName(), ctx,
-      LinalgTilingOptions().setTileSizes({200, 300, 400}),
-      LinalgTransformationFilter(StringAttr::get(ctx, "L3"),
-                                 StringAttr::get(ctx, "L2")));
-  patterns.add<LinalgTilingPattern>(
-      MatmulOp::getOperationName(), ctx,
-      LinalgTilingOptions().setTileSizes({20, 30, 40}),
-      LinalgTransformationFilter(StringAttr::get(ctx, "L2"),
-                                 StringAttr::get(ctx, "L1")));
-  patterns.add<LinalgTilingPattern>(
-      MatmulOp::getOperationName(), ctx,
-      LinalgTilingOptions().setTileSizes({2, 3, 4}),
-      LinalgTransformationFilter(StringAttr::get(ctx, "L1"),
-                                 StringAttr::get(ctx, "REG")));
-
-  patterns.add<LinalgTilingPattern>(
-      MatvecOp::getOperationName(), ctx,
-      LinalgTilingOptions().setTileSizes({5, 6}).setLoopType(
-          LinalgTilingLoopType::ParallelLoops),
-      LinalgTransformationFilter(ArrayRef<StringAttr>{},
-                                 StringAttr::get(ctx, "L1")));
-
-  patterns.add<LinalgTilingPattern>(
-      DotOp::getOperationName(), ctx, LinalgTilingOptions().setTileSizes(8000),
-      LinalgTransformationFilter(
-          ArrayRef<StringAttr>{StringAttr::get(ctx, "MEM"),
-                               StringAttr::get(ctx, "L3"),
-                               StringAttr::get(ctx, "L2")},
-          StringAttr::get(ctx, "REG")));
-
-  //===--------------------------------------------------------------------===//
-  // Linalg tiling and permutation patterns.
-  //===--------------------------------------------------------------------===//
-  patterns.add<LinalgTilingPattern>(
-      MatmulOp::getOperationName(), ctx,
-      LinalgTilingOptions()
-          .setTileSizes({2000, 3000, 4000})
-          .setInterchange({1, 2, 0}),
-      LinalgTransformationFilter(StringAttr::get(ctx, "__with_perm__"),
-                                 StringAttr::get(ctx, "L2__with_perm__")));
-  patterns.add<LinalgTilingPattern>(
-      MatmulOp::getOperationName(), ctx,
-      LinalgTilingOptions()
-          .setTileSizes({200, 300, 400})
-          .setInterchange({1, 0, 2}),
-      LinalgTransformationFilter(StringAttr::get(ctx, "L2__with_perm__"),
-                                 StringAttr::get(ctx, "L1__with_perm__")));
-  patterns.add<LinalgTilingPattern>(
-      MatmulOp::getOperationName(), ctx,
-      LinalgTilingOptions().setTileSizes({20, 30, 40}),
-      LinalgTransformationFilter(StringAttr::get(ctx, "L1__with_perm__"),
-                                 StringAttr::get(ctx, "REG__with_perm__")));
-
-  patterns.add<LinalgTilingPattern>(
-      MatvecOp::getOperationName(), ctx,
-      LinalgTilingOptions().setTileSizes({5, 6}).setInterchange({1, 0}),
-      LinalgTransformationFilter(StringAttr::get(ctx, "__with_perm__"),
-                                 StringAttr::get(ctx, "L1__with_perm__")));
-
-  patterns.add<LinalgTilingPattern>(
-      MatmulOp::getOperationName(), ctx,
-      LinalgTilingOptions()
-          .setTileSizes({16, 8, 4})
-          .setInterchange({1, 2, 0})
-          .setLoopType(LinalgTilingLoopType::ParallelLoops),
-      LinalgTransformationFilter(
-          StringAttr::get(ctx, "par__with_perm__"),
-          StringAttr::get(ctx, "after_par__with_perm__")));
-
-  //===--------------------------------------------------------------------===//
   // Linalg to loops patterns.
   //===--------------------------------------------------------------------===//
   patterns.add<LinalgLoweringPattern<DotOp>>(
       ctx,
-      /*loweringType=*/LinalgLoweringType::Loops,
-      LinalgTransformationFilter(StringAttr::get(ctx, "REG")));
+      /*loweringType=*/LinalgLoweringType::Loops);
 
   //===--------------------------------------------------------------------===//
   // Linalg distribution patterns.
   //===--------------------------------------------------------------------===//
   LinalgLoopDistributionOptions distributionOptions;
 
   //===--------------------------------------------------------------------===//
   // Linalg to vector contraction patterns.
   //===--------------------------------------------------------------------===//
   patterns.add<CopyVectorizationPattern>(ctx);
 
   (void)applyPatternsAndFoldGreedily(funcOp, std::move(patterns));
 
   // Drop the marker.
   funcOp.walk([](LinalgOp op) {
     op->removeAttr(LinalgTransforms::kLinalgTransformMarker);
   });
 }
 
-template <typename IdOp, typename NProcsOp>
-static SmallVector<ProcInfo, 2>
-getGpuProcIds(OpBuilder &b, Location loc, ArrayRef<Range> parallelLoopRanges,
-              ArrayRef<linalg::DistributionMethod> distributionMethod) {
-  size_t count = std::min<size_t>(3, parallelLoopRanges.size());
-  SmallVector<ProcInfo, 2> procInfo(count);
-  Type indexType = b.getIndexType();
-  for (unsigned i = 0; i < count; ++i) {
-    gpu::Dimension dim = *gpu::symbolizeDimension(i);
-    procInfo[count - 1 - i] = {b.create<IdOp>(loc, indexType, dim),
-                               b.create<NProcsOp>(loc, indexType, dim),
-                               distributionMethod[count - 1 - i]};
-  }
-  return procInfo;
-}
-
-static void fillTileAndDistributePatterns(MLIRContext *context,
-                                          RewritePatternSet &patterns) {
-  {
-    LinalgLoopDistributionOptions cyclicNprocsEqNiters;
-    SmallVector<linalg::DistributionMethod> distributionMethod = {
-        DistributionMethod::CyclicNumProcsEqNumIters,
-        DistributionMethod::CyclicNumProcsEqNumIters};
-    cyclicNprocsEqNiters.procInfo =
-        [distributionMethod](OpBuilder &b, Location loc,
-                             ArrayRef<Range> parallelLoopRanges) {
-          return getGpuProcIds<gpu::BlockIdOp, gpu::GridDimOp>(
-              b, loc, parallelLoopRanges, distributionMethod);
-        };
-    patterns.add<LinalgTilingPattern>(
-        MatmulOp::getOperationName(), context,
-        LinalgTilingOptions()
-            .setTileSizes({8, 8, 4})
-            .setLoopType(LinalgTilingLoopType::ParallelLoops)
-            .setDistributionOptions(cyclicNprocsEqNiters),
-        LinalgTransformationFilter(
-            StringAttr::get(context, "distribute1"),
-            StringAttr::get(context, "after_distribute1")));
-  }
-
-  {
-    LinalgLoopDistributionOptions cyclicNprocsGeNiters;
-    SmallVector<linalg::DistributionMethod> distributionMethod = {
-        DistributionMethod::CyclicNumProcsGeNumIters,
-        DistributionMethod::CyclicNumProcsGeNumIters};
-    cyclicNprocsGeNiters.procInfo =
-        [distributionMethod](OpBuilder &b, Location loc,
-                             ArrayRef<Range> parallelLoopRanges) {
-          return getGpuProcIds<gpu::BlockIdOp, gpu::GridDimOp>(
-              b, loc, parallelLoopRanges, distributionMethod);
-        };
-    patterns.add<LinalgTilingPattern>(
-        MatmulOp::getOperationName(), context,
-        LinalgTilingOptions()
-            .setTileSizes({8, 8, 4})
-            .setLoopType(LinalgTilingLoopType::ParallelLoops)
-            .setDistributionOptions(cyclicNprocsGeNiters),
-        LinalgTransformationFilter(
-            StringAttr::get(context, "distribute2"),
-            StringAttr::get(context, "after_distribute2")));
-  }
-
-  {
-    LinalgLoopDistributionOptions cyclicNprocsDefault;
-    SmallVector<linalg::DistributionMethod> distributionMethod = {
-        DistributionMethod::Cyclic, DistributionMethod::Cyclic};
-    cyclicNprocsDefault.procInfo =
-        [distributionMethod](OpBuilder &b, Location loc,
-                             ArrayRef<Range> parallelLoopRanges) {
-          return getGpuProcIds<gpu::BlockIdOp, gpu::GridDimOp>(
-              b, loc, parallelLoopRanges, distributionMethod);
-        };
-    patterns.add<LinalgTilingPattern>(
-        MatmulOp::getOperationName(), context,
-        LinalgTilingOptions()
-            .setTileSizes({8, 8, 4})
-            .setLoopType(LinalgTilingLoopType::ParallelLoops)
-            .setDistributionOptions(cyclicNprocsDefault),
-        LinalgTransformationFilter(
-            StringAttr::get(context, "distribute3"),
-            StringAttr::get(context, "after_distribute3")));
-  }
-
-  {
-    LinalgLoopDistributionOptions cyclicNprocsMixed1;
-    SmallVector<linalg::DistributionMethod> distributionMethod = {
-        DistributionMethod::CyclicNumProcsEqNumIters,
-        DistributionMethod::CyclicNumProcsGeNumIters};
-    cyclicNprocsMixed1.procInfo =
-        [distributionMethod](OpBuilder &b, Location loc,
-                             ArrayRef<Range> parallelLoopRanges) {
-          return getGpuProcIds<gpu::BlockIdOp, gpu::GridDimOp>(
-              b, loc, parallelLoopRanges, distributionMethod);
-        };
-    patterns.add<LinalgTilingPattern>(
-        MatmulOp::getOperationName(), context,
-        LinalgTilingOptions()
-            .setTileSizes({8, 8, 4})
-            .setLoopType(LinalgTilingLoopType::ParallelLoops)
-            .setDistributionOptions(cyclicNprocsMixed1),
-        LinalgTransformationFilter(
-            StringAttr::get(context, "distribute4"),
-            StringAttr::get(context, "after_distribute4")));
-  }
-
-  {
-    LinalgLoopDistributionOptions cyclicNprocsMixed2;
-    SmallVector<linalg::DistributionMethod> distributionMethod = {
-        DistributionMethod::CyclicNumProcsGeNumIters,
-        DistributionMethod::Cyclic};
-    cyclicNprocsMixed2.procInfo =
-        [distributionMethod](OpBuilder &b, Location loc,
-                             ArrayRef<Range> parallelLoopRanges) {
-          return getGpuProcIds<gpu::BlockIdOp, gpu::GridDimOp>(
-              b, loc, parallelLoopRanges, distributionMethod);
-        };
-    patterns.add<LinalgTilingPattern>(
-        MatmulOp::getOperationName(), context,
-        LinalgTilingOptions()
-            .setTileSizes({8, 8, 4})
-            .setLoopType(LinalgTilingLoopType::ParallelLoops)
-            .setDistributionOptions(cyclicNprocsMixed2),
-        LinalgTransformationFilter(
-            StringAttr::get(context, "distribute5"),
-            StringAttr::get(context, "after_distribute5")));
-  }
-
-  {
-    LinalgLoopDistributionOptions cyclicNprocsMixed3;
-    SmallVector<linalg::DistributionMethod> distributionMethod = {
-        DistributionMethod::Cyclic,
-        DistributionMethod::CyclicNumProcsEqNumIters};
-    cyclicNprocsMixed3.procInfo =
-        [distributionMethod](OpBuilder &b, Location loc,
-                             ArrayRef<Range> parallelLoopRanges) {
-          return getGpuProcIds<gpu::BlockIdOp, gpu::GridDimOp>(
-              b, loc, parallelLoopRanges, distributionMethod);
-        };
-
-    patterns.add<LinalgTilingPattern>(
-        MatmulOp::getOperationName(), context,
-        LinalgTilingOptions()
-            .setTileSizes({8, 8, 4})
-            .setLoopType(LinalgTilingLoopType::ParallelLoops)
-            .setDistributionOptions(cyclicNprocsMixed3),
-        LinalgTransformationFilter(
-            StringAttr::get(context, "distribute6"),
-            StringAttr::get(context, "after_distribute6")));
-  }
-
-  {
-    LinalgLoopDistributionOptions cyclicNprocsEqNiters;
-    SmallVector<linalg::DistributionMethod> distributionMethod = {
-        DistributionMethod::Cyclic, DistributionMethod::Cyclic};
-    cyclicNprocsEqNiters.procInfo =
-        [distributionMethod](OpBuilder &b, Location loc,
-                             ArrayRef<Range> parallelLoopRanges) {
-          return getGpuProcIds<gpu::BlockIdOp, gpu::GridDimOp>(
-              b, loc, parallelLoopRanges, distributionMethod);
-        };
-    patterns.add<LinalgTilingPattern>(
-        MatmulOp::getOperationName(), context,
-        LinalgTilingOptions()
-            .setTileSizes({8, 8, 4})
-            .setLoopType(LinalgTilingLoopType::Loops)
-            .setDistributionOptions(cyclicNprocsEqNiters),
-        LinalgTransformationFilter(
-            StringAttr::get(context, "tensors_distribute1"),
-            StringAttr::get(context, "tensors_after_distribute1")));
-  }
-}
-
 static void applyVectorTransferForwardingPatterns(func::FuncOp funcOp) {
   RewritePatternSet forwardPattern(funcOp.getContext());
   forwardPattern.add<LinalgCopyVTRForwardingPattern>(funcOp.getContext());
   forwardPattern.add<LinalgCopyVTWForwardingPattern>(funcOp.getContext());
   (void)applyPatternsAndFoldGreedily(funcOp, std::move(forwardPattern));
 }
 
 static void applyLinalgToVectorPatterns(func::FuncOp funcOp) {

 }
 
 static void applyExtractSliceOfPadTensorSwapPattern(func::FuncOp funcOp) {
   RewritePatternSet patterns(funcOp.getContext());
   patterns.add<ExtractSliceOfPadTensorSwapPattern>(funcOp.getContext());
   (void)applyPatternsAndFoldGreedily(funcOp, std::move(patterns));
 }
 
-static void applyTilePattern(func::FuncOp funcOp, const std::string &loopType,
-                             ArrayRef<int64_t> tileSizes,
-                             ArrayRef<int64_t> peeledLoops,
-                             bool scalarizeDynamicDims) {
-  MLIRContext *context = funcOp.getContext();
-  RewritePatternSet tilingPattern(context);
-  LinalgTilingLoopType type =
-      llvm::StringSwitch<LinalgTilingLoopType>(loopType)
-          .Case("for", LinalgTilingLoopType::Loops)
-          .Case("affine", LinalgTilingLoopType::AffineLoops)
-          .Case("parallel", LinalgTilingLoopType::ParallelLoops);
-  auto linalgTilingOptions = linalg::LinalgTilingOptions()
-                                 .setPeeledLoops(peeledLoops)
-                                 .setLoopType(type);
-  if (scalarizeDynamicDims) {
-    linalgTilingOptions.scalarizeDynamicDims();
-    assert(tileSizes.empty() &&
-           "tileSizes and scalarizeDynamicDims is mutually exclusive");
-  } else {
-    linalgTilingOptions.setTileSizes(tileSizes);
-  }
-  linalg::LinalgTransformationFilter f(StringAttr::get(context, "tile"));
-  TilingPatterns<linalg::MatmulOp, linalg::GenericOp>::insert(
-      tilingPattern, linalgTilingOptions, f);
-  (void)applyPatternsAndFoldGreedily(funcOp, std::move(tilingPattern));
-}
-
 static void applySplitReduction(func::FuncOp funcOp) {
   RewritePatternSet patterns(funcOp.getContext());
   linalg::populateSplitReductionPattern(
       patterns,
       [](LinalgOp op) {
         unsigned insertDimIndex = op.getNumLoops() - 1;
         return SplitReductionOptions{4, insertDimIndex, false};
       },

 void TestLinalgTransforms::runOnOperation() {
   auto lambda = [&](void *) {
     getOperation().walk([](LinalgOp op) {
       op->removeAttr(LinalgTransforms::kLinalgTransformMarker);
     });
   };
   std::unique_ptr<void, decltype(lambda)> cleanupGuard{(void *)1, lambda};
 
-  if (testTileAndDistributionOptions) {
-    RewritePatternSet patterns(&getContext());
-    fillTileAndDistributePatterns(&getContext(), patterns);
-    (void)applyPatternsAndFoldGreedily(getOperation(), std::move(patterns));
-    return;
-  }
   if (testPatterns)
     return applyPatterns(getOperation());
   if (testVectorTransferForwardingPatterns)
     return applyVectorTransferForwardingPatterns(getOperation());
   if (testGenericToVectorPattern)
     return applyLinalgToVectorPatterns(getOperation());
   if (testTransformPadTensor)
     return applyPadTensorToGenericPatterns(getOperation());
   if (testGeneralizePadTensor)
     return applyGeneralizePadTensorPatterns(getOperation());
   if (testSwapSubTensorPadTensor)
     return applyExtractSliceOfPadTensorSwapPattern(getOperation());
-  if (testTilePattern)
-    return applyTilePattern(getOperation(), loopType, tileSizes, peeledLoops,
-                            /*scalarizeDynamicDims=*/false);
-  if (testTileScalarizeDynamicDims)
-    return applyTilePattern(getOperation(), loopType, tileSizes,
-                            /*peeledLoops=*/{}, /*scalarizeDynamicDims=*/true);
   if (testSplitReduction)
     return applySplitReduction(getOperation());
   if (testSplitReductionInnerParallel)
     return applySplitReductionInnerParallel(getOperation());
   if (testBubbleUpExtractSliceOpPattern)
     return applyBubbleUpExtractSliceOpPattern(getOperation());
   if (testSwapExtractSliceWithFill)
     return applySwapExtractSliceWithFillPattern(getOperation());

mlir/test/lib/Dialect/Transform/TestTransformDialectInterpreter.cpp

void runOnOperation() override {
     }
   }
 
   Option<bool> enableExpensiveChecks{
       *this, "enable-expensive-checks", llvm::cl::init(false),
       llvm::cl::desc("perform expensive checks to better report errors in the "
                      "transform IR")};
 };
+
+struct TestTransformDialectEraseSchedulePass
+    : public PassWrapper<TestTransformDialectEraseSchedulePass,
+                         OperationPass<ModuleOp>> {
+  MLIR_DEFINE_EXPLICIT_INTERNAL_INLINE_TYPE_ID(
+      TestTransformDialectEraseSchedulePass)
+
+  StringRef getArgument() const final {
+    return "test-transform-dialect-erase-schedule";
+  }
+
+  StringRef getDescription() const final {
+    return "erase transform dialect schedule from the IR";
+  }
+
+  void runOnOperation() override {
+    getOperation()->walk<WalkOrder::PreOrder>([&](Operation *nestedOp) {
+      if (isa<transform::TransformOpInterface>(nestedOp)) {
+        nestedOp->erase();
+        return WalkResult::skip();
+      }
+      return WalkResult::advance();
+    });
+  }
+};
 } // namespace
 
 namespace mlir {
 namespace test {
+/// Registers the test pass for erasing transform dialect ops.
+void registerTestTransformDialectEraseSchedulePass() {
+  PassRegistration<TestTransformDialectEraseSchedulePass> reg;
+}
 /// Registers the test pass for applying transform dialect ops.
 void registerTestTransformDialectInterpreterPass() {
   PassRegistration<TestTransformDialectInterpreterPass> reg;
 }
 } // namespace test
 } // namespace mlir

mlir/tools/mlir-opt/mlir-opt.cpp

 void registerTestPreparationPassWithAllowedMemrefResults();
 void registerTestRecursiveTypesPass();
 void registerTestSCFUtilsPass();
 void registerTestShapeMappingPass();
 void registerTestSliceAnalysisPass();
 void registerTestTensorTransforms();
 void registerTestTilingInterface();
 void registerTestTopologicalSortAnalysisPass();
+void registerTestTransformDialectEraseSchedulePass();
 void registerTestTransformDialectInterpreterPass();
 void registerTestVectorLowerings();
 void registerTestNvgpuLowerings();
 } // namespace test
 } // namespace mlir
 
 namespace test {
 void registerTestDialect(DialectRegistry &);
#endif
   mlir::test::registerTestPDLLPasses();
   mlir::test::registerTestRecursiveTypesPass();
   mlir::test::registerTestSCFUtilsPass();
   mlir::test::registerTestShapeMappingPass();
   mlir::test::registerTestSliceAnalysisPass();
   mlir::test::registerTestTensorTransforms();
   mlir::test::registerTestTilingInterface();
   mlir::test::registerTestTopologicalSortAnalysisPass();
+  mlir::test::registerTestTransformDialectEraseSchedulePass();
   mlir::test::registerTestTransformDialectInterpreterPass();
   mlir::test::registerTestVectorLowerings();
   mlir::test::registerTestNvgpuLowerings();
 }
 #endif
 
 int main(int argc, char **argv) {
   registerAllPasses();