[mlir][Vector] Canonicalize empty vector.mask into arith.select
#140976
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
dcaballe
requested review from
Groverkss,
banach-space,
kuhar and
nicolasvasilache
as code owners
Member
|
@llvm/pr-subscribers-mlir Author: Diego Caballero (dcaballe) ChangesThis MR adds a missing canonicalization for empty Full diff: https://github.com/llvm/llvm-project/pull/140976.diff 3 Files Affected:
diff --git a/mlir/include/mlir/Dialect/Vector/IR/VectorOps.td b/mlir/include/mlir/Dialect/Vector/IR/VectorOps.td
index 3f5564541554e..5e8421ed67d66 100644
--- a/mlir/include/mlir/Dialect/Vector/IR/VectorOps.td
+++ b/mlir/include/mlir/Dialect/Vector/IR/VectorOps.td
@@ -2559,6 +2559,7 @@ def Vector_MaskOp : Vector_Op<"mask", [
Location loc);
}];
+ let hasCanonicalizer = 1;
let hasFolder = 1;
let hasCustomAssemblyFormat = 1;
let hasVerifier = 1;
diff --git a/mlir/lib/Dialect/Vector/IR/VectorOps.cpp b/mlir/lib/Dialect/Vector/IR/VectorOps.cpp
index 41777347975da..389950af821a2 100644
--- a/mlir/lib/Dialect/Vector/IR/VectorOps.cpp
+++ b/mlir/lib/Dialect/Vector/IR/VectorOps.cpp
@@ -6661,6 +6661,7 @@ LogicalResult MaskOp::verify() {
///
/// %0 = user_op %a : vector<8xf32>
///
+/// Empty `vector.mask` with passthru operand are handled by the canonicalizer.
static LogicalResult foldEmptyMaskOp(MaskOp maskOp, MaskOp::FoldAdaptor adaptor,
SmallVectorImpl<OpFoldResult> &results) {
if (!maskOp.isEmpty() || maskOp.hasPassthru())
@@ -6696,6 +6697,47 @@ LogicalResult MaskOp::fold(FoldAdaptor adaptor,
return success();
}
+// Canonialize empty `vector.mask` operations that can't be handled in
+// `VectorMask::fold`.
+//
+// Example 1: Empty `vector.mask` with passthru operand.
+//
+// %0 = vector.mask %mask, %passthru { vector.yield %a : vector<8xf32> } :
+// vector<8xi1> -> vector<8xf32>
+//
+// becomes:
+//
+// %0 = arith.select %mask, %a, %passthru : vector<8xf32>
+//
+class CanonializeEmptyMaskOp : public OpRewritePattern<MaskOp> {
+ using OpRewritePattern::OpRewritePattern;
+
+ LogicalResult matchAndRewrite(MaskOp maskOp,
+ PatternRewriter &rewriter) const override {
+ if (!maskOp.isEmpty())
+ return failure();
+
+ if (!maskOp.hasPassthru())
+ return failure();
+
+ Block *block = maskOp.getMaskBlock();
+ auto terminator = cast<vector::YieldOp>(block->front());
+ assert(terminator.getNumOperands() == 1 &&
+ "expected one result when passthru is provided");
+
+ rewriter.replaceOpWithNewOp<arith::SelectOp>(
+ maskOp, maskOp.getResultTypes(), maskOp.getMask(),
+ terminator.getOperand(0), maskOp.getPassthru());
+
+ return success();
+ }
+};
+
+void MaskOp::getCanonicalizationPatterns(RewritePatternSet &results,
+ MLIRContext *context) {
+ results.add<CanonializeEmptyMaskOp>(context);
+}
+
// MaskingOpInterface definitions.
/// Returns the operation masked by this 'vector.mask'.
diff --git a/mlir/test/Dialect/Vector/canonicalize.mlir b/mlir/test/Dialect/Vector/canonicalize.mlir
index 974f4506a2ef0..a6543aafd1c77 100644
--- a/mlir/test/Dialect/Vector/canonicalize.mlir
+++ b/mlir/test/Dialect/Vector/canonicalize.mlir
@@ -719,7 +719,7 @@ func.func @fold_extract_transpose(
// CHECK-LABEL: fold_extract_broadcast_same_input_output_scalar
// CHECK-SAME: %[[A:.*]]: f32
// CHECK: return %[[A]] : f32
-func.func @fold_extract_broadcast_same_input_output_scalar(%a : f32,
+func.func @fold_extract_broadcast_same_input_output_scalar(%a : f32,
%idx0 : index, %idx1 : index, %idx2 : index) -> f32 {
%b = vector.broadcast %a : f32 to vector<1x2x4xf32>
%r = vector.extract %b[%idx0, %idx1, %idx2] : f32 from vector<1x2x4xf32>
@@ -731,7 +731,7 @@ func.func @fold_extract_broadcast_same_input_output_scalar(%a : f32,
// CHECK-LABEL: fold_extract_broadcast_same_input_output_vec
// CHECK-SAME: %[[A:.*]]: vector<4xf32>
// CHECK: return %[[A]] : vector<4xf32>
-func.func @fold_extract_broadcast_same_input_output_vec(%a : vector<4xf32>,
+func.func @fold_extract_broadcast_same_input_output_vec(%a : vector<4xf32>,
%idx0 : index, %idx1 : index) -> vector<4xf32> {
%b = vector.broadcast %a : vector<4xf32> to vector<1x2x4xf32>
%r = vector.extract %b[%idx0, %idx1] : vector<4xf32> from vector<1x2x4xf32>
@@ -744,7 +744,7 @@ func.func @fold_extract_broadcast_same_input_output_vec(%a : vector<4xf32>,
// CHECK-SAME: %[[A:.*]]: vector<f32>
// CHECK: %[[B:.+]] = vector.extract %[[A]][] : f32 from vector<f32>
// CHECK: return %[[B]] : f32
-func.func @fold_extract_broadcast_0dvec_input_scalar_output(%a : vector<f32>,
+func.func @fold_extract_broadcast_0dvec_input_scalar_output(%a : vector<f32>,
%idx0 : index, %idx1 : index, %idx2: index) -> f32 {
%b = vector.broadcast %a : vector<f32> to vector<1x2x4xf32>
%r = vector.extract %b[%idx0, %idx1, %idx2] : f32 from vector<1x2x4xf32>
@@ -780,7 +780,7 @@ func.func @fold_extract_splat(%a : f32, %idx0 : index, %idx1 : index, %idx2 : in
// CHECK-SAME: %[[IDX:.*]]: index, %[[IDX1:.*]]: index, %[[IDX2:.*]]: index
// CHECK: %[[R:.*]] = vector.extract %[[A]][%[[IDX1]], 0] : f32 from vector<2x1xf32>
// CHECK: return %[[R]] : f32
-func.func @fold_extract_broadcast_dim1_broadcasting(%a : vector<2x1xf32>,
+func.func @fold_extract_broadcast_dim1_broadcasting(%a : vector<2x1xf32>,
%idx : index, %idx1 : index, %idx2 : index) -> f32 {
%b = vector.broadcast %a : vector<2x1xf32> to vector<1x2x4xf32>
%r = vector.extract %b[%idx, %idx1, %idx2] : f32 from vector<1x2x4xf32>
@@ -795,7 +795,7 @@ func.func @fold_extract_broadcast_dim1_broadcasting(%a : vector<2x1xf32>,
// CHECK: %[[B:.+]] = vector.extract %[[A]][%[[IDX1]]] : vector<4xf32> from vector<2x4xf32>
// CHECK: return %[[B]] : vector<4xf32>
// rank(extract_output) < rank(broadcast_input)
-func.func @fold_extract_broadcast_to_lower_rank(%a : vector<2x4xf32>,
+func.func @fold_extract_broadcast_to_lower_rank(%a : vector<2x4xf32>,
%idx0 : index, %idx1 : index) -> vector<4xf32> {
%b = vector.broadcast %a : vector<2x4xf32> to vector<1x2x4xf32>
%r = vector.extract %b[%idx0, %idx1] : vector<4xf32> from vector<1x2x4xf32>
@@ -808,7 +808,7 @@ func.func @fold_extract_broadcast_to_lower_rank(%a : vector<2x4xf32>,
// CHECK: %[[B:.*]] = vector.broadcast %{{.*}} : f32 to vector<4xf32>
// CHECK: return %[[B]] : vector<4xf32>
// rank(extract_output) > rank(broadcast_input)
-func.func @fold_extract_broadcast_to_higher_rank(%a : f32, %idx0 : index, %idx1 : index)
+func.func @fold_extract_broadcast_to_higher_rank(%a : f32, %idx0 : index, %idx1 : index)
-> vector<4xf32> {
%b = vector.broadcast %a : f32 to vector<1x2x4xf32>
%r = vector.extract %b[%idx0, %idx1] : vector<4xf32> from vector<1x2x4xf32>
@@ -822,7 +822,7 @@ func.func @fold_extract_broadcast_to_higher_rank(%a : f32, %idx0 : index, %idx1
// CHECK: %[[R:.*]] = vector.broadcast %[[A]] : vector<1xf32> to vector<8xf32>
// CHECK: return %[[R]] : vector<8xf32>
// rank(extract_output) == rank(broadcast_input)
-func.func @fold_extract_broadcast_to_equal_rank(%a : vector<1xf32>, %idx0 : index)
+func.func @fold_extract_broadcast_to_equal_rank(%a : vector<1xf32>, %idx0 : index)
-> vector<8xf32> {
%b = vector.broadcast %a : vector<1xf32> to vector<1x8xf32>
%r = vector.extract %b[%idx0] : vector<8xf32> from vector<1x8xf32>
@@ -1169,7 +1169,7 @@ func.func @broadcast_poison() -> vector<4x6xi8> {
return %broadcast : vector<4x6xi8>
}
-// -----
+// -----
// CHECK-LABEL: broadcast_splat_constant
// CHECK: %[[CONST:.*]] = arith.constant dense<1> : vector<4x6xi8>
@@ -2756,6 +2756,19 @@ func.func @empty_vector_mask_with_return(%a : vector<8xf32>, %mask : vector<8xi1
// -----
+// CHECK-LABEL: func @empty_vector_mask_with_passthru
+// CHECK-SAME: %[[IN:.*]]: vector<8xf32>, %[[MASK:.*]]: vector<8xi1>, %[[PASSTHRU:.*]]: vector<8xf32>
+func.func @empty_vector_mask_with_passthru(%a : vector<8xf32>, %mask : vector<8xi1>,
+ %passthru : vector<8xf32>) -> vector<8xf32> {
+// CHECK-NOT: vector.mask
+// CHECK: %[[SEL:.*]] = arith.select %[[MASK]], %[[IN]], %[[PASSTHRU]] : vector<8xi1>, vector<8xf32>
+// CHECK: return %[[SEL]] : vector<8xf32>
+ %0 = vector.mask %mask, %passthru { vector.yield %a : vector<8xf32> } : vector<8xi1> -> vector<8xf32>
+ return %0 : vector<8xf32>
+}
+
+// -----
+
// CHECK-LABEL: func @all_true_vector_mask
// CHECK-SAME: %[[IN:.*]]: tensor<3x4xf32>
func.func @all_true_vector_mask(%ta : tensor<3x4xf32>) -> vector<3x4xf32> {
|
Member
|
@llvm/pr-subscribers-mlir-vector Author: Diego Caballero (dcaballe) ChangesThis MR adds a missing canonicalization for empty Full diff: https://github.com/llvm/llvm-project/pull/140976.diff 3 Files Affected:
diff --git a/mlir/include/mlir/Dialect/Vector/IR/VectorOps.td b/mlir/include/mlir/Dialect/Vector/IR/VectorOps.td
index 3f5564541554e..5e8421ed67d66 100644
--- a/mlir/include/mlir/Dialect/Vector/IR/VectorOps.td
+++ b/mlir/include/mlir/Dialect/Vector/IR/VectorOps.td
@@ -2559,6 +2559,7 @@ def Vector_MaskOp : Vector_Op<"mask", [
Location loc);
}];
+ let hasCanonicalizer = 1;
let hasFolder = 1;
let hasCustomAssemblyFormat = 1;
let hasVerifier = 1;
diff --git a/mlir/lib/Dialect/Vector/IR/VectorOps.cpp b/mlir/lib/Dialect/Vector/IR/VectorOps.cpp
index 41777347975da..389950af821a2 100644
--- a/mlir/lib/Dialect/Vector/IR/VectorOps.cpp
+++ b/mlir/lib/Dialect/Vector/IR/VectorOps.cpp
@@ -6661,6 +6661,7 @@ LogicalResult MaskOp::verify() {
///
/// %0 = user_op %a : vector<8xf32>
///
+/// Empty `vector.mask` with passthru operand are handled by the canonicalizer.
static LogicalResult foldEmptyMaskOp(MaskOp maskOp, MaskOp::FoldAdaptor adaptor,
SmallVectorImpl<OpFoldResult> &results) {
if (!maskOp.isEmpty() || maskOp.hasPassthru())
@@ -6696,6 +6697,47 @@ LogicalResult MaskOp::fold(FoldAdaptor adaptor,
return success();
}
+// Canonialize empty `vector.mask` operations that can't be handled in
+// `VectorMask::fold`.
+//
+// Example 1: Empty `vector.mask` with passthru operand.
+//
+// %0 = vector.mask %mask, %passthru { vector.yield %a : vector<8xf32> } :
+// vector<8xi1> -> vector<8xf32>
+//
+// becomes:
+//
+// %0 = arith.select %mask, %a, %passthru : vector<8xf32>
+//
+class CanonializeEmptyMaskOp : public OpRewritePattern<MaskOp> {
+ using OpRewritePattern::OpRewritePattern;
+
+ LogicalResult matchAndRewrite(MaskOp maskOp,
+ PatternRewriter &rewriter) const override {
+ if (!maskOp.isEmpty())
+ return failure();
+
+ if (!maskOp.hasPassthru())
+ return failure();
+
+ Block *block = maskOp.getMaskBlock();
+ auto terminator = cast<vector::YieldOp>(block->front());
+ assert(terminator.getNumOperands() == 1 &&
+ "expected one result when passthru is provided");
+
+ rewriter.replaceOpWithNewOp<arith::SelectOp>(
+ maskOp, maskOp.getResultTypes(), maskOp.getMask(),
+ terminator.getOperand(0), maskOp.getPassthru());
+
+ return success();
+ }
+};
+
+void MaskOp::getCanonicalizationPatterns(RewritePatternSet &results,
+ MLIRContext *context) {
+ results.add<CanonializeEmptyMaskOp>(context);
+}
+
// MaskingOpInterface definitions.
/// Returns the operation masked by this 'vector.mask'.
diff --git a/mlir/test/Dialect/Vector/canonicalize.mlir b/mlir/test/Dialect/Vector/canonicalize.mlir
index 974f4506a2ef0..a6543aafd1c77 100644
--- a/mlir/test/Dialect/Vector/canonicalize.mlir
+++ b/mlir/test/Dialect/Vector/canonicalize.mlir
@@ -719,7 +719,7 @@ func.func @fold_extract_transpose(
// CHECK-LABEL: fold_extract_broadcast_same_input_output_scalar
// CHECK-SAME: %[[A:.*]]: f32
// CHECK: return %[[A]] : f32
-func.func @fold_extract_broadcast_same_input_output_scalar(%a : f32,
+func.func @fold_extract_broadcast_same_input_output_scalar(%a : f32,
%idx0 : index, %idx1 : index, %idx2 : index) -> f32 {
%b = vector.broadcast %a : f32 to vector<1x2x4xf32>
%r = vector.extract %b[%idx0, %idx1, %idx2] : f32 from vector<1x2x4xf32>
@@ -731,7 +731,7 @@ func.func @fold_extract_broadcast_same_input_output_scalar(%a : f32,
// CHECK-LABEL: fold_extract_broadcast_same_input_output_vec
// CHECK-SAME: %[[A:.*]]: vector<4xf32>
// CHECK: return %[[A]] : vector<4xf32>
-func.func @fold_extract_broadcast_same_input_output_vec(%a : vector<4xf32>,
+func.func @fold_extract_broadcast_same_input_output_vec(%a : vector<4xf32>,
%idx0 : index, %idx1 : index) -> vector<4xf32> {
%b = vector.broadcast %a : vector<4xf32> to vector<1x2x4xf32>
%r = vector.extract %b[%idx0, %idx1] : vector<4xf32> from vector<1x2x4xf32>
@@ -744,7 +744,7 @@ func.func @fold_extract_broadcast_same_input_output_vec(%a : vector<4xf32>,
// CHECK-SAME: %[[A:.*]]: vector<f32>
// CHECK: %[[B:.+]] = vector.extract %[[A]][] : f32 from vector<f32>
// CHECK: return %[[B]] : f32
-func.func @fold_extract_broadcast_0dvec_input_scalar_output(%a : vector<f32>,
+func.func @fold_extract_broadcast_0dvec_input_scalar_output(%a : vector<f32>,
%idx0 : index, %idx1 : index, %idx2: index) -> f32 {
%b = vector.broadcast %a : vector<f32> to vector<1x2x4xf32>
%r = vector.extract %b[%idx0, %idx1, %idx2] : f32 from vector<1x2x4xf32>
@@ -780,7 +780,7 @@ func.func @fold_extract_splat(%a : f32, %idx0 : index, %idx1 : index, %idx2 : in
// CHECK-SAME: %[[IDX:.*]]: index, %[[IDX1:.*]]: index, %[[IDX2:.*]]: index
// CHECK: %[[R:.*]] = vector.extract %[[A]][%[[IDX1]], 0] : f32 from vector<2x1xf32>
// CHECK: return %[[R]] : f32
-func.func @fold_extract_broadcast_dim1_broadcasting(%a : vector<2x1xf32>,
+func.func @fold_extract_broadcast_dim1_broadcasting(%a : vector<2x1xf32>,
%idx : index, %idx1 : index, %idx2 : index) -> f32 {
%b = vector.broadcast %a : vector<2x1xf32> to vector<1x2x4xf32>
%r = vector.extract %b[%idx, %idx1, %idx2] : f32 from vector<1x2x4xf32>
@@ -795,7 +795,7 @@ func.func @fold_extract_broadcast_dim1_broadcasting(%a : vector<2x1xf32>,
// CHECK: %[[B:.+]] = vector.extract %[[A]][%[[IDX1]]] : vector<4xf32> from vector<2x4xf32>
// CHECK: return %[[B]] : vector<4xf32>
// rank(extract_output) < rank(broadcast_input)
-func.func @fold_extract_broadcast_to_lower_rank(%a : vector<2x4xf32>,
+func.func @fold_extract_broadcast_to_lower_rank(%a : vector<2x4xf32>,
%idx0 : index, %idx1 : index) -> vector<4xf32> {
%b = vector.broadcast %a : vector<2x4xf32> to vector<1x2x4xf32>
%r = vector.extract %b[%idx0, %idx1] : vector<4xf32> from vector<1x2x4xf32>
@@ -808,7 +808,7 @@ func.func @fold_extract_broadcast_to_lower_rank(%a : vector<2x4xf32>,
// CHECK: %[[B:.*]] = vector.broadcast %{{.*}} : f32 to vector<4xf32>
// CHECK: return %[[B]] : vector<4xf32>
// rank(extract_output) > rank(broadcast_input)
-func.func @fold_extract_broadcast_to_higher_rank(%a : f32, %idx0 : index, %idx1 : index)
+func.func @fold_extract_broadcast_to_higher_rank(%a : f32, %idx0 : index, %idx1 : index)
-> vector<4xf32> {
%b = vector.broadcast %a : f32 to vector<1x2x4xf32>
%r = vector.extract %b[%idx0, %idx1] : vector<4xf32> from vector<1x2x4xf32>
@@ -822,7 +822,7 @@ func.func @fold_extract_broadcast_to_higher_rank(%a : f32, %idx0 : index, %idx1
// CHECK: %[[R:.*]] = vector.broadcast %[[A]] : vector<1xf32> to vector<8xf32>
// CHECK: return %[[R]] : vector<8xf32>
// rank(extract_output) == rank(broadcast_input)
-func.func @fold_extract_broadcast_to_equal_rank(%a : vector<1xf32>, %idx0 : index)
+func.func @fold_extract_broadcast_to_equal_rank(%a : vector<1xf32>, %idx0 : index)
-> vector<8xf32> {
%b = vector.broadcast %a : vector<1xf32> to vector<1x8xf32>
%r = vector.extract %b[%idx0] : vector<8xf32> from vector<1x8xf32>
@@ -1169,7 +1169,7 @@ func.func @broadcast_poison() -> vector<4x6xi8> {
return %broadcast : vector<4x6xi8>
}
-// -----
+// -----
// CHECK-LABEL: broadcast_splat_constant
// CHECK: %[[CONST:.*]] = arith.constant dense<1> : vector<4x6xi8>
@@ -2756,6 +2756,19 @@ func.func @empty_vector_mask_with_return(%a : vector<8xf32>, %mask : vector<8xi1
// -----
+// CHECK-LABEL: func @empty_vector_mask_with_passthru
+// CHECK-SAME: %[[IN:.*]]: vector<8xf32>, %[[MASK:.*]]: vector<8xi1>, %[[PASSTHRU:.*]]: vector<8xf32>
+func.func @empty_vector_mask_with_passthru(%a : vector<8xf32>, %mask : vector<8xi1>,
+ %passthru : vector<8xf32>) -> vector<8xf32> {
+// CHECK-NOT: vector.mask
+// CHECK: %[[SEL:.*]] = arith.select %[[MASK]], %[[IN]], %[[PASSTHRU]] : vector<8xi1>, vector<8xf32>
+// CHECK: return %[[SEL]] : vector<8xf32>
+ %0 = vector.mask %mask, %passthru { vector.yield %a : vector<8xf32> } : vector<8xi1> -> vector<8xf32>
+ return %0 : vector<8xf32>
+}
+
+// -----
+
// CHECK-LABEL: func @all_true_vector_mask
// CHECK-SAME: %[[IN:.*]]: tensor<3x4xf32>
func.func @all_true_vector_mask(%ta : tensor<3x4xf32>) -> vector<3x4xf32> {
|
banach-space
approved these changes
May 22, 2025
Contributor
banach-space
left a comment
banach-space
left a comment
There was a problem hiding this comment.
LGTM % kind request to expand the documentation :)
Btw, did you mean arith.select in the PR title?
Comment on lines
6700
to
6701
Contributor
There was a problem hiding this comment.
[nit] Could you document why this can't be handled by the folder?
Contributor
There was a problem hiding this comment.
Could you document _why)?
dcaballe
changed the title
vector.mask into vector.selectvector.mask into arith.select
This MR adds a missing canonicalization for empty `vector.mask` ops with
a passthru value.
```
%0 = vector.mask %mask, %passthru { vector.yield %a : vector<8xf32> } :
vector<8xi1> -> vector<8xf32>
becomes:
%0 = arith.select %mask, %a, %passthru : vector<8xf32>
```
dcaballe
force-pushed
the
dcaballero/fold-empty-vector-mask-passthru
branch
from
9a21ad5 to
fcf2152
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This MR adds a missing canonicalization for empty
vector.maskops with a passthru value.