[mlir][sparse] More allocate -> empty tensor migration #66720
Merged
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
This also allows tensor.empty in the "conversion" path of the sparse compiler, further paving the way to deprecate the bufferization.allocated_tensor() op.
aartbik
requested review from
matthias-springer,
PeimingLiu,
maerhart and
yinying-lisa-li
|
@llvm/pr-subscribers-mlir-sparse @llvm/pr-subscribers-mlir ChangesThis also allows tensor.empty in the "conversion" path of the sparse compiler, further paving the way to Patch is 37.16 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/66720.diff 9 Files Affected:
diff --git a/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorConversion.cpp b/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorConversion.cpp
index 871686a4ada0f70..d75601e369a0d25 100644
--- a/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorConversion.cpp
+++ b/mlir/lib/Dialect/SparseTensor/Transforms/SparseTensorConversion.cpp
@@ -830,6 +830,7 @@ class SparseTensorNewConverter : public OpConversionPattern<NewOp> {
};
/// Sparse conversion rule for the alloc operator.
+/// TODO(springerm): remove when bufferization.alloc_tensor is gone
class SparseTensorAllocConverter
: public OpConversionPattern<bufferization::AllocTensorOp> {
public:
@@ -864,6 +865,37 @@ class SparseTensorAllocConverter
}
};
+/// Sparse conversion rule for the empty tensor.
+class SparseTensorEmptyConverter : public OpConversionPattern<tensor::EmptyOp> {
+public:
+ using OpConversionPattern::OpConversionPattern;
+ LogicalResult
+ matchAndRewrite(tensor::EmptyOp op, OpAdaptor adaptor,
+ ConversionPatternRewriter &rewriter) const override {
+ Location loc = op.getLoc();
+ const auto stt = getSparseTensorType(op);
+ if (!stt.hasEncoding())
+ return failure();
+ // Gather all dimension sizes as SSA values.
+ const Dimension dimRank = stt.getDimRank();
+ SmallVector<Value> dimSizes;
+ dimSizes.reserve(dimRank);
+ auto shape = op.getType().getShape();
+ unsigned operandCtr = 0;
+ for (Dimension d = 0; d < dimRank; ++d) {
+ dimSizes.push_back(stt.isDynamicDim(d)
+ ? adaptor.getOperands()[operandCtr++]
+ : constantIndex(rewriter, loc, shape[d]));
+ }
+ // Generate the call to construct empty tensor. The sizes are
+ // explicitly defined by the arguments to the alloc operator.
+ rewriter.replaceOp(op, NewCallParams(rewriter, loc)
+ .genBuffers(stt, dimSizes)
+ .genNewCall(Action::kEmpty));
+ return success();
+ }
+};
+
/// Sparse conversion rule for the convert operator.
class SparseTensorConvertConverter : public OpConversionPattern<ConvertOp> {
public:
@@ -1503,19 +1535,19 @@ mlir::SparseTensorTypeToPtrConverter::SparseTensorTypeToPtrConverter() {
void mlir::populateSparseTensorConversionPatterns(
TypeConverter &typeConverter, RewritePatternSet &patterns,
const SparseTensorConversionOptions &options) {
- patterns.add<SparseReturnConverter, SparseTensorToDimSizeConverter,
- SparseCastConverter, SparseTensorNewConverter,
- SparseReshapeConverter<tensor::ExpandShapeOp>,
- SparseReshapeConverter<tensor::CollapseShapeOp>,
- SparseTensorConcatConverter, SparseTensorAllocConverter,
- SparseTensorDeallocConverter, SparseTensorToPositionsConverter,
- SparseTensorToCoordinatesConverter,
- SparseTensorToValuesConverter, SparseNumberOfEntriesConverter,
- SparseTensorLoadConverter, SparseTensorInsertConverter,
- SparseTensorExpandConverter, SparseTensorCompressConverter,
- SparseTensorOutConverter, SparseTensorPackConverter>(
- typeConverter, patterns.getContext());
-
+ patterns
+ .add<SparseReturnConverter, SparseTensorToDimSizeConverter,
+ SparseCastConverter, SparseTensorNewConverter,
+ SparseReshapeConverter<tensor::ExpandShapeOp>,
+ SparseReshapeConverter<tensor::CollapseShapeOp>,
+ SparseTensorConcatConverter, SparseTensorAllocConverter,
+ SparseTensorEmptyConverter, SparseTensorDeallocConverter,
+ SparseTensorToPositionsConverter, SparseTensorToCoordinatesConverter,
+ SparseTensorToValuesConverter, SparseNumberOfEntriesConverter,
+ SparseTensorLoadConverter, SparseTensorInsertConverter,
+ SparseTensorExpandConverter, SparseTensorCompressConverter,
+ SparseTensorOutConverter, SparseTensorPackConverter>(
+ typeConverter, patterns.getContext());
patterns.add<SparseTensorConvertConverter>(typeConverter,
patterns.getContext(), options);
}
diff --git a/mlir/test/Dialect/SparseTensor/constant_index_map.mlir b/mlir/test/Dialect/SparseTensor/constant_index_map.mlir
index bfb4503edbc4e40..9eb535385790146 100644
--- a/mlir/test/Dialect/SparseTensor/constant_index_map.mlir
+++ b/mlir/test/Dialect/SparseTensor/constant_index_map.mlir
@@ -13,7 +13,7 @@
// CHECK-DAG: %[[VAL_2:.*]] = arith.constant 77 : index
// CHECK-DAG: %[[VAL_3:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[VAL_4:.*]] = arith.constant 1 : index
-// CHECK-DAG: %[[VAL_5:.*]] = bufferization.alloc_tensor() : tensor<77xi1, #{{.*}}>
+// CHECK-DAG: %[[VAL_5:.*]] = tensor.empty() : tensor<77xi1, #{{.*}}>
// CHECK-DAG: %[[VAL_6:.*]] = bufferization.to_memref %[[VAL_0]] : memref<1x77xi1>
// CHECK-DAG: %[[VAL_7:.*]] = bufferization.to_memref %[[VAL_1]] : memref<1x77xi1>
// CHECK: %[[VAL_8:.*]] = scf.for %[[VAL_9:.*]] = %[[VAL_3]] to %[[VAL_2]] step %[[VAL_4]] iter_args(%[[VAL_10:.*]] = %[[VAL_5]]) -> (tensor<77xi1, #{{.*}}>) {
@@ -27,7 +27,7 @@
// CHECK: return %[[VAL_15]] : tensor<77xi1, #{{.*}}>
// CHECK: }
func.func @main(%arg0: tensor<1x77xi1>, %arg1: tensor<1x77xi1>) -> tensor<77xi1, #SpVec> {
- %0 = bufferization.alloc_tensor() : tensor<77xi1, #SpVec>
+ %0 = tensor.empty() : tensor<77xi1, #SpVec>
%1 = linalg.generic {
indexing_maps = [#map1, #map1, #map2],
iterator_types = ["parallel"]}
diff --git a/mlir/test/Dialect/SparseTensor/sparse_affine.mlir b/mlir/test/Dialect/SparseTensor/sparse_affine.mlir
index b3f6ae9f12ee4d6..fc97685b8378bf5 100644
--- a/mlir/test/Dialect/SparseTensor/sparse_affine.mlir
+++ b/mlir/test/Dialect/SparseTensor/sparse_affine.mlir
@@ -1,4 +1,3 @@
-// NOTE: Assertions have been autogenerated by utils/generate-test-checks.py
// RUN: mlir-opt %s -sparsification | FileCheck %s
#SpVec = #sparse_tensor.encoding<{ map = (d0) -> (d0 : compressed) }>
@@ -17,9 +16,9 @@
}
// CHECK-LABEL: func @mul_inv_dense1d(
-// CHECK-SAME: %[[VAL_0:.*]]: tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>,
-// CHECK-SAME: %[[VAL_1:.*]]: tensor<4xf32>,
-// CHECK-SAME: %[[VAL_2:.*]]: tensor<32xf32>) -> tensor<32xf32> {
+// CHECK-SAME: %[[VAL_0:.*]]: tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>,
+// CHECK-SAME: %[[VAL_1:.*]]: tensor<4xf32>,
+// CHECK-SAME: %[[VAL_2:.*]]: tensor<32xf32>) -> tensor<32xf32> {
// CHECK-DAG: %[[VAL_3:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[VAL_4:.*]] = arith.constant 3 : index
// CHECK-DAG: %[[VAL_5:.*]] = arith.constant 1 : index
@@ -57,13 +56,13 @@ func.func @mul_inv_dense1d(%arga: tensor<32xf32, #SpVec>,
}
// CHECK-LABEL: func.func @mul_inv_sparse1d(
-// CHECK-SAME: %[[VAL_0:.*]]: tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>,
-// CHECK-SAME: %[[VAL_1:.*]]: tensor<4xf32, #sparse_tensor.encoding<{{{.*}}}>>)
+// CHECK-SAME: %[[VAL_0:.*]]: tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>,
+// CHECK-SAME: %[[VAL_1:.*]]: tensor<4xf32, #sparse_tensor.encoding<{{{.*}}}>>)
// CHECK: %[[VAL_2:.*]] = arith.constant 0 : index
// CHECK: %[[VAL_3:.*]] = arith.constant 1 : index
// CHECK: %[[VAL_4:.*]] = arith.constant 3 : index
// CHECK: %[[VAL_5:.*]] = arith.constant 0.000000e+00 : f32
-// CHECK: %[[VAL_6:.*]] = bufferization.alloc_tensor() : tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>
+// CHECK: %[[VAL_6:.*]] = tensor.empty() : tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>
// CHECK: %[[VAL_7:.*]] = sparse_tensor.positions %[[VAL_0]] {level = 0 : index} : tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xindex>
// CHECK: %[[VAL_8:.*]] = sparse_tensor.values %[[VAL_0]] : tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xf32>
// CHECK: %[[VAL_9:.*]] = sparse_tensor.positions %[[VAL_1]] {level = 0 : index} : tensor<4xf32, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xindex>
@@ -95,7 +94,7 @@ func.func @mul_inv_dense1d(%arga: tensor<32xf32, #SpVec>,
// CHECK: return %[[VAL_32]] : tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>
func.func @mul_inv_sparse1d(%arga: tensor<32xf32, #SpVec>,
%argb: tensor<4xf32, #SpVec>) -> tensor<32xf32, #SpVec> {
- %argx = bufferization.alloc_tensor() : tensor<32xf32, #SpVec>
+ %argx = tensor.empty() : tensor<32xf32, #SpVec>
%0 = linalg.generic #trait1
ins(%arga, %argb: tensor<32xf32, #SpVec>, tensor<4xf32, #SpVec>)
outs(%argx: tensor<32xf32, #SpVec>) {
@@ -109,13 +108,13 @@ func.func @mul_inv_sparse1d(%arga: tensor<32xf32, #SpVec>,
// CHECK-LABEL: func.func @mul_inv_enc_dense1d(
-// CHECK-SAME: %[[VAL_0:.*]]: tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>,
-// CHECK-SAME: %[[VAL_1:.*]]: tensor<4xf32, #sparse_tensor.encoding<{{{.*}}}>>) -> tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>> {
+// CHECK-SAME: %[[VAL_0:.*]]: tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>,
+// CHECK-SAME: %[[VAL_1:.*]]: tensor<4xf32, #sparse_tensor.encoding<{{{.*}}}>>) -> tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>> {
// CHECK: %[[VAL_2:.*]] = arith.constant 32 : index
// CHECK: %[[VAL_3:.*]] = arith.constant 3 : index
// CHECK: %[[VAL_4:.*]] = arith.constant 0 : index
// CHECK: %[[VAL_5:.*]] = arith.constant 1 : index
-// CHECK: %[[VAL_6:.*]] = bufferization.alloc_tensor() : tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>
+// CHECK: %[[VAL_6:.*]] = tensor.empty() : tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>>
// CHECK: %[[VAL_7:.*]] = sparse_tensor.values %[[VAL_0]] : tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xf32>
// CHECK: %[[VAL_8:.*]] = sparse_tensor.values %[[VAL_1]] : tensor<4xf32, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xf32>
// CHECK: %[[VAL_9:.*]] = sparse_tensor.values %[[VAL_6]] : tensor<32xf32, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xf32>
@@ -132,7 +131,7 @@ func.func @mul_inv_sparse1d(%arga: tensor<32xf32, #SpVec>,
// CHECK: }
func.func @mul_inv_enc_dense1d(%arga: tensor<32xf32, #EncDenseVec>,
%argb: tensor<4xf32, #EncDenseVec>) -> tensor<32xf32, #EncDenseVec> {
- %argx = bufferization.alloc_tensor() : tensor<32xf32, #EncDenseVec>
+ %argx = tensor.empty() : tensor<32xf32, #EncDenseVec>
%0 = linalg.generic #trait1
ins(%arga, %argb: tensor<32xf32, #EncDenseVec>, tensor<4xf32, #EncDenseVec>)
outs(%argx: tensor<32xf32, #EncDenseVec>) {
@@ -155,9 +154,9 @@ func.func @mul_inv_enc_dense1d(%arga: tensor<32xf32, #EncDenseVec>,
}
// CHECK-LABEL: func @and_affine_dense1d(
-// CHECK-SAME: %[[VAL_0:.*]]: tensor<32xi32, #sparse_tensor.encoding<{{{.*}}}>>,
-// CHECK-SAME: %[[VAL_1:.*]]: tensor<34xi32>,
-// CHECK-SAME: %[[VAL_2:.*]]: tensor<32xi32>) -> tensor<32xi32> {
+// CHECK-SAME: %[[VAL_0:.*]]: tensor<32xi32, #sparse_tensor.encoding<{{{.*}}}>>,
+// CHECK-SAME: %[[VAL_1:.*]]: tensor<34xi32>,
+// CHECK-SAME: %[[VAL_2:.*]]: tensor<32xi32>) -> tensor<32xi32> {
// CHECK-DAG: %[[ZERO:.*]] = arith.constant 0 : i32
// CHECK-DAG: %[[VAL_3:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[VAL_4:.*]] = arith.constant 1 : index
@@ -195,12 +194,12 @@ func.func @and_affine_dense1d(%arga: tensor<32xi32, #SpVec>,
}
// CHECK-LABEL: func.func @and_affine_sparse1d(
-// CHECK-SAME: %[[VAL_0:.*]]: tensor<32xi32, #sparse_tensor.encoding<{{{.*}}}>>,
-// CHECK-SAME: %[[VAL_1:.*]]: tensor<34xi32, #sparse_tensor.encoding<{{{.*}}}>>)
+// CHECK-SAME: %[[VAL_0:.*]]: tensor<32xi32, #sparse_tensor.encoding<{{{.*}}}>>,
+// CHECK-SAME: %[[VAL_1:.*]]: tensor<34xi32, #sparse_tensor.encoding<{{{.*}}}>>)
// CHECK: %[[VAL_2:.*]] = arith.constant 0 : index
// CHECK: %[[VAL_3:.*]] = arith.constant 1 : index
// CHECK: %[[VAL_4:.*]] = arith.constant 2 : index
-// CHECK: %[[VAL_5:.*]] = bufferization.alloc_tensor() : tensor<32xi32, #sparse_tensor.encoding<{{{.*}}}>>
+// CHECK: %[[VAL_5:.*]] = tensor.empty() : tensor<32xi32, #sparse_tensor.encoding<{{{.*}}}>>
// CHECK: %[[VAL_6:.*]] = sparse_tensor.positions %[[VAL_0]] {level = 0 : index} : tensor<32xi32, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xindex>
// CHECK: %[[VAL_7:.*]] = sparse_tensor.coordinates %[[VAL_0]] {level = 0 : index} : tensor<32xi32, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xindex>
// CHECK: %[[VAL_8:.*]] = sparse_tensor.values %[[VAL_0]] : tensor<32xi32, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xi32>
@@ -234,7 +233,7 @@ func.func @and_affine_dense1d(%arga: tensor<32xi32, #SpVec>,
// CHECK: return %[[VAL_33]] : tensor<32xi32, #sparse_tensor.encoding<{{{.*}}}>>
func.func @and_affine_sparse1d(%arga: tensor<32xi32, #SpVec>,
%argb: tensor<34xi32, #SpVec>) -> tensor<32xi32, #SpVec> {
- %argx = bufferization.alloc_tensor() : tensor<32xi32, #SpVec>
+ %argx = tensor.empty() : tensor<32xi32, #SpVec>
%0 = linalg.generic #trait2
ins(%arga, %argb: tensor<32xi32, #SpVec>, tensor<34xi32, #SpVec>)
outs(%argx: tensor<32xi32, #SpVec>) {
@@ -256,9 +255,9 @@ func.func @and_affine_sparse1d(%arga: tensor<32xi32, #SpVec>,
}
// CHECK-LABEL: func @mul_affine_dense2d(
-// CHECK-SAME: %[[VAL_0:.*]]: tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>>,
-// CHECK-SAME: %[[VAL_1:.*]]: tensor<34x19xf64>,
-// CHECK-SAME: %[[VAL_2:.*]]: tensor<32x16xf64>) -> tensor<32x16xf64> {
+// CHECK-SAME: %[[VAL_0:.*]]: tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>>,
+// CHECK-SAME: %[[VAL_1:.*]]: tensor<34x19xf64>,
+// CHECK-SAME: %[[VAL_2:.*]]: tensor<32x16xf64>) -> tensor<32x16xf64> {
// CHECK-DAG: %[[VAL_3:.*]] = arith.constant 1 : index
// CHECK-DAG: %[[VAL_4:.*]] = arith.constant 32 : index
// CHECK-DAG: %[[VAL_5:.*]] = arith.constant 0 : index
@@ -304,8 +303,8 @@ func.func @mul_affine_dense2d(%arga: tensor<32x16xf64, #CSR>,
// CHECK-LABEL: func.func @mul_affine_sparse2d(
-// CHECK-SAME: %[[VAL_0:.*]]: tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>>,
-// CHECK-SAME: %[[VAL_1:.*]]: tensor<34x19xf64, #sparse_tensor.encoding<{{{.*}}}>>) -> tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>> {
+// CHECK-SAME: %[[VAL_0:.*]]: tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>>,
+// CHECK-SAME: %[[VAL_1:.*]]: tensor<34x19xf64, #sparse_tensor.encoding<{{{.*}}}>>) -> tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>> {
// CHECK-DAG: %[[VAL_2:.*]] = arith.constant 32 : index
// CHECK-DAG: %[[VAL_3:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[VAL_4:.*]] = arith.constant 1 : index
@@ -314,7 +313,7 @@ func.func @mul_affine_dense2d(%arga: tensor<32x16xf64, #CSR>,
// CHECK-DAG: %[[VAL_7:.*]] = arith.constant 3 : index
// CHECK-DAG: %[[VAL_TRUE:.*]] = arith.constant true
// CHECK-DAG: %[[VAL_FALSE:.*]] = arith.constant false
-// CHECK: %[[VAL_8:.*]] = bufferization.alloc_tensor() : tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>>
+// CHECK: %[[VAL_8:.*]] = tensor.empty() : tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>>
// CHECK: %[[VAL_9:.*]] = sparse_tensor.positions %[[VAL_0]] {level = 1 : index} : tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xindex>
// CHECK: %[[VAL_10:.*]] = sparse_tensor.coordinates %[[VAL_0]] {level = 1 : index} : tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xindex>
// CHECK: %[[VAL_11:.*]] = sparse_tensor.values %[[VAL_0]] : tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>> to memref<?xf64>
@@ -360,7 +359,7 @@ func.func @mul_affine_dense2d(%arga: tensor<32x16xf64, #CSR>,
// CHECK: return %[[VAL_45]] : tensor<32x16xf64, #sparse_tensor.encoding<{{{.*}}}>>
func.func @mul_affine_sparse2d(%arga: tensor<32x16xf64, #CSR>,
%argb: tensor<34x19xf64, #CSR>) -> tensor<32x16xf64, #CSR> {
- %argx = bufferization.alloc_tensor() : tensor<32x16xf64, #CSR>
+ %argx = tensor.empty() : tensor<32x16xf64, #CSR>
%0 = linalg.generic #trait3
ins(%arga, %argb: tensor<32x16xf64, #CSR>, tensor<34x19xf64, #CSR>)
outs(%argx: tensor<32x16xf64, #CSR>) {
@@ -383,9 +382,9 @@ func.func @mul_affine_sparse2d(%arga: tensor<32x16xf64, #CSR>,
}
// CHECK-LABEL: func.func @mul_affine_dense_dim_2d(
-// CHECK-SAME: %[[VAL_0:.*]]: tensor<34x16xf64, #sparse_tensor.encoding
-// CHECK-SAME: %[[VAL_1:.*]]: tensor<32x19xf64, #sparse_tensor.encoding<{{{.*}}}>>,
-// CHECK-SAME: %[[VAL_2:.*]]: tensor<32x16xf64>) -> tensor<32x16xf64> {
+// CHECK-SAME: %[[VAL_0:.*]]: tensor<34x16xf64, #sparse_tensor.encoding
+// CHECK-SAME: %[[VAL_1:.*]]: tensor<32x19xf64, #sparse_tensor.encoding<{{{.*}}}>>,
+// CHECK-SAME: %[[VAL_2:.*]]: tensor<32x16xf64>) -> tensor<32x16xf64> {
// CHECK-DAG: %[[VAL_3:.*]] = arith.constant 19 : index
// CHECK-DAG: %[[VAL_4:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[VAL_5:.*]] = arith.constant 1 : index
@@ -447,9 +446,9 @@ func.func @mul_affine_dense_dim_2d(%arga: tensor<34x16xf64, #CSR>,
}
// CHECK-LABEL: func.func @mul_const_affine_dense_dim_2d(
-// CHECK-SAME: %[[VAL_0:.*]]: tensor<34x16xf64,
-// CHECK-SAME: %[[VAL_1:.*]]: tensor<32x19xf64, #sparse_tensor.encoding<{{{.*}}}>>,
-// CHECK-SAME: %[[VAL_2:.*]]: tensor<32x16xf64>) -> tensor<32x16xf64> {
+// CHECK-SAME: %[[VAL_0:.*]]: tensor<34x16xf64,
+// CHECK-SAME: %[[VAL_1:.*]]: tensor<32x19xf64, #sparse_tensor.encoding<{{{.*}}}>>,
+// CHECK-SAME: %[[VAL_2:.*]]: tensor<32x16xf64>) -> tensor<32x16xf64> {
// CHECK-DAG: %[[VAL_3:.*]] = arith.constant 19 : index
// CHECK-DAG: %[[VAL_4:.*]] = arith.constant 2 : index
// CHECK-DAG: %[[VAL_5:.*]] = arith.constant 0 : index
diff --git a/mlir/test/Dialect/SparseTensor/sparse_broadcast.mlir b/mlir/test/Dialect/SparseTensor/sparse_broadcast.mlir
index ae5b941259f6515..1a5f79d23cba29a 100644
--- a/mlir/test/Dialect/SparseTensor/sparse_broadcast.mlir
+++ b/mlir/test/Dialect/SparseTensor/sparse_broadcast.mlir
@@ -16,7 +16,7 @@
// CHECK-DAG: %[[TMP_c3:.*]] = arith.constant 3 : index
// CHECK-DAG: %[[TMP_c0:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[TMP_c1:.*]] = arith.constant 1 : index
-// CHECK: %[[TMP_0:.*]] = bufferization.alloc_tensor()
+// CHECK: %[[TMP_0:.*]] = tensor.empty()
// CHECK: %[[TMP_1:.*]] = sparse_tensor.positions %[[TMP_arg0]] {level = 0 : index}
// CHECK: %[[TMP_2:.*]] = sparse_tensor.coordinates %[[TMP_arg0]] {level = 0 : index}
// CHECK: %[[TMP_3:.*]] = sparse_tensor.positions %[[TMP_arg0]] {level = 1 : index}
@@ -44,7 +44,7 @@
// CHECK: return %[[TMP_8]]
module @func_sparse {
func.func public @main(%arg0: tensor<4x5xi32, #DCSR>) -> tensor<4x3x5xi32, #SparseTensor> {
- %0 = bufferization.alloc_tensor() : tensor<4x3x5xi32, #SparseTensor>
+ %0 = tensor.empty() : tensor<4x3x5xi32, #SparseTensor>
%1 = linalg.generic #trait
ins(%arg0 : tensor<4x5xi32, #DCSR>) outs(%0 : tensor<4x3x5xi32, #SparseTensor>) {
^bb0(%in: i32, %out: i32):
diff --git a/mlir/test/Dialect/SparseTensor/sparse_expand.mlir b/mlir/test/Dialect/SparseTensor/sparse_expand.mlir
index ee3613a268def5e..d19d7fe2871d674 100644
--- a/mlir/test/Dialect/SparseTensor/sparse_expand.mlir
+++ b/mlir/test/Dialect/SparseTen...
[truncated]
|
yinying-lisa-li
approved these changes
Sep 18, 2023
This was referenced Sep 19, 2023
This was referenced Sep 20, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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 also allows tensor.empty in the "conversion" path of the sparse compiler, further paving the way to
deprecate the bufferization.allocated_tensor() op.