Fold linalg.fill -> linalg.copy

[MaheshRavishankar]

No description provided.

[llvmbot]

@llvm/pr-subscribers-mlir

@llvm/pr-subscribers-mlir-linalg

Author: None (MaheshRavishankar)

Changes

---

Full diff: https://github.com/llvm/llvm-project/pull/72920.diff

2 Files Affected:

• (modified) mlir/lib/Dialect/Linalg/IR/LinalgOps.cpp (+26-4)
• (modified) mlir/test/Dialect/Linalg/canonicalize.mlir (+13)

diff --git a/mlir/lib/Dialect/Linalg/IR/LinalgOps.cpp b/mlir/lib/Dialect/Linalg/IR/LinalgOps.cpp
index d12ba8c4c59b33f..58af9995548e939 100644
--- a/mlir/lib/Dialect/Linalg/IR/LinalgOps.cpp
+++ b/mlir/lib/Dialect/Linalg/IR/LinalgOps.cpp
@@ -803,14 +803,36 @@ struct FoldFillWithPack : public OpRewritePattern<tensor::PackOp> {
   }
 };
 
+/// Fold fill with copy.
+struct FoldFillWithCopy : OpRewritePattern<linalg::CopyOp> {
+  using OpRewritePattern<linalg::CopyOp>::OpRewritePattern;
+
+  LogicalResult matchAndRewrite(linalg::CopyOp copyOp,
+                                PatternRewriter &rewriter) const override {
+    if (auto fillOp = copyOp.getInputs().front().getDefiningOp<FillOp>()) {
+      rewriter.replaceOpWithNewOp<FillOp>(copyOp, copyOp.getResultTypes(),
+                                          fillOp.getInputs(),
+                                          copyOp.getOutputs());
+      return success();
+    }
+    if (auto fillOp = copyOp.getOutputs().front().getDefiningOp<FillOp>()) {
+      rewriter.replaceOpWithNewOp<linalg::CopyOp>(copyOp, copyOp.getInputs(),
+                                                  fillOp.getOutputs());
+      return success();
+    }
+    return failure();
+  }
+};
+
 } // namespace
 
 void FillOp::getCanonicalizationPatterns(RewritePatternSet &results,
                                          MLIRContext *context) {
-  results.add<FoldFillWithTensorExtract, FoldFillWithPack, FoldFillWithPad,
-              FoldFillWithTensorReshape<tensor::CollapseShapeOp>,
-              FoldFillWithTensorReshape<tensor::ExpandShapeOp>,
-              FoldInsertPadIntoFill>(context);
+  results
+      .add<FoldFillWithCopy, FoldFillWithTensorExtract, FoldFillWithPack,
+           FoldFillWithPad, FoldFillWithTensorReshape<tensor::CollapseShapeOp>,
+           FoldFillWithTensorReshape<tensor::ExpandShapeOp>,
+           FoldInsertPadIntoFill>(context);
 }
 
 //===----------------------------------------------------------------------===//
diff --git a/mlir/test/Dialect/Linalg/canonicalize.mlir b/mlir/test/Dialect/Linalg/canonicalize.mlir
index 7793e435582746c..c054829a915d7ba 100644
--- a/mlir/test/Dialect/Linalg/canonicalize.mlir
+++ b/mlir/test/Dialect/Linalg/canonicalize.mlir
@@ -972,3 +972,16 @@ func.func @canonicalize_dim_of_dest_style_op(%arg0 : tensor<?x?xf32>) -> tensor<
   %3 = linalg.copy ins(%1 : tensor<?x?xf32>) outs(%2 : tensor<?x?xf32>) -> tensor<?x?xf32>
   return %3: tensor<?x?xf32>
 }
+
+// -----
+
+// CHECK-LABEL: func @canonicalize_fill_to_copy_dest(
+//  CHECK-SAME:     %[[ARG0:[a-zA-Z0-9]+]]: tensor<?x?xf32>
+//  CHECK-SAME:     %[[ARG1:[a-zA-Z0-9]+]]: tensor<?x?xf32>)
+//       CHECK:   linalg.copy ins(%[[ARG1]] : tensor<?x?xf32>) outs(%[[ARG0]] : tensor<?x?xf32>)
+func.func @canonicalize_fill_to_copy_dest(%arg0 : tensor<?x?xf32>, %arg1 : tensor<?x?xf32>) -> tensor<?x?xf32> {
+  %c0 = arith.constant 0.0 : f32
+  %fill = linalg.fill ins(%c0 : f32) outs(%arg0 : tensor<?x?xf32>) -> tensor<?x?xf32>
+  %copy = linalg.copy ins(%arg1 : tensor<?x?xf32>) outs(%fill : tensor<?x?xf32>) -> tensor<?x?xf32>
+  return %copy : tensor<?x?xf32>
+}

[banach-space]

Wouldn't DCE get rid of linalg.copy here?

Suggested change

	// CHECK: %[[ZERO:.+]] = arith.constant 0.0
	// CHECK: %[[ZERO:.+]] = arith.constant 0.0
	// CHECK-NOT: linalg.copy

Similar comment below.

[MaheshRavishankar]

DCE does remove it. It isn't removed by the pattern though explicitly. So I'd rather not test for that here.