This is an archive of the discontinued LLVM Phabricator instance.

Differential D134486

[mlir][sparse] Mark sparse_tensor.concatenate as NoSideEffect
ClosedPublic

Authored by Peiming on Sep 22 2022, 3:03 PM.

Details

Reviewers
aartbik
nicolasvasilache
bixia
Commits
rG1ab2b0075dfc: [mlir][sparse] Mark sparse_tensor.concatenate as NoSideEffect

Diff Detail

Event Timeline

Peiming created this revision.Sep 22 2022, 3:03 PM
Herald added a reviewer: aartbik. · View Herald TranscriptSep 22 2022, 3:03 PM
Herald added a project: Restricted Project. · View Herald Transcript
Herald added subscribers: anlunx, bzcheeseman, sdasgup3 and 19 others. · View Herald Transcript
Peiming requested review of this revision.Sep 22 2022, 3:03 PM
Harbormaster completed remote builds in B188274: Diff 462307.Sep 22 2022, 3:03 PM
Herald added a reviewer: nicolasvasilache. · View Herald TranscriptSep 22 2022, 3:03 PM
Herald added a project: Restricted Project. · View Herald Transcript
Herald added a reviewer: nicolasvasilache. · View Herald Transcript
Herald added subscribers: stephenneuendorffer, nicolasvasilache. · View Herald Transcript
Peiming added a reviewer: bixia.Sep 22 2022, 3:03 PM
aartbik added inline comments.Sep 22 2022, 3:05 PM
mlir/test/Dialect/SparseTensor/fold.mlir
36

CHECK-NOT: sparse_tensor.concatenate

;-)

Peiming updated this revision to Diff 462312.Sep 22 2022, 3:08 PM

fix typo

Harbormaster completed remote builds in B188279: Diff 462312.Sep 22 2022, 3:08 PM
Peiming marked an inline comment as done.Sep 22 2022, 3:09 PM
aartbik accepted this revision.Sep 22 2022, 3:16 PM
This revision is now accepted and ready to land.Sep 22 2022, 3:16 PM
Closed by commit rG1ab2b0075dfc: [mlir][sparse] Mark sparse_tensor.concatenate as NoSideEffect (authored by Peiming). · Explain WhySep 22 2022, 3:25 PM
This revision was automatically updated to reflect the committed changes.
Peiming added a commit: rG1ab2b0075dfc: [mlir][sparse] Mark sparse_tensor.concatenate as NoSideEffect.

Revision Contents

PathSize
mlir/
include/
mlir/
Dialect/
SparseTensor/
IR/
2 lines
test/
Dialect/
SparseTensor/
13 lines

Diff 462312

mlir/include/mlir/Dialect/SparseTensor/IR/SparseTensorOps.td

Show First 20 Lines • Show All 155 Lines • ▼ Show 20 Lines let description = [{
```mlir ```mlir
%1 = sparse_tensor.values %0 : tensor<64x64xf64, #CSR> to memref<?xf64> %1 = sparse_tensor.values %0 : tensor<64x64xf64, #CSR> to memref<?xf64>
``` ```
}]; }];
let assemblyFormat = "$tensor attr-dict `:` type($tensor) `to` type($result)"; let assemblyFormat = "$tensor attr-dict `:` type($tensor) `to` type($result)";
let hasVerifier = 1; let hasVerifier = 1;
} }
def SparseTensor_ConcatenateOp : SparseTensor_Op<"concatenate", []>, def SparseTensor_ConcatenateOp : SparseTensor_Op<"concatenate", [NoSideEffect]>,
Arguments<(ins Variadic<AnyRankedTensor>:$inputs, IndexAttr:$dimension)>, Arguments<(ins Variadic<AnyRankedTensor>:$inputs, IndexAttr:$dimension)>,
Results<(outs AnyRankedTensor:$result)> { Results<(outs AnyRankedTensor:$result)> {
let summary = "Concatenates a list of tensors into a single tensor."; let summary = "Concatenates a list of tensors into a single tensor.";
let description = [{ let description = [{
Concatenates a list input tensors and the output tensor with the same rank. Concatenates a list input tensors and the output tensor with the same rank.
The concatenation happens on the specified `dimension` (0<= dimension < rank). The concatenation happens on the specified `dimension` (0<= dimension < rank).
The resulting `dimension` size is the sum of all the input dimension sizes, The resulting `dimension` size is the sum of all the input dimension sizes,
▲ Show 20 Lines • Show All 549 Lines • Show Last 20 Lines

mlir/test/Dialect/SparseTensor/fold.mlir

Show All 26 Lines
// CHECK-NOT: sparse_tensor.values // CHECK-NOT: sparse_tensor.values
// CHECK: return // CHECK: return
func.func @sparse_dce_getters(%arg0: tensor<64xf32, #SparseVector>) { func.func @sparse_dce_getters(%arg0: tensor<64xf32, #SparseVector>) {
%0 = sparse_tensor.pointers %arg0 { dimension = 0 : index } : tensor<64xf32, #SparseVector> to memref<?xindex> %0 = sparse_tensor.pointers %arg0 { dimension = 0 : index } : tensor<64xf32, #SparseVector> to memref<?xindex>
%1 = sparse_tensor.indices %arg0 { dimension = 0 : index } : tensor<64xf32, #SparseVector> to memref<?xindex> %1 = sparse_tensor.indices %arg0 { dimension = 0 : index } : tensor<64xf32, #SparseVector> to memref<?xindex>
%2 = sparse_tensor.values %arg0 : tensor<64xf32, #SparseVector> to memref<?xf32> %2 = sparse_tensor.values %arg0 : tensor<64xf32, #SparseVector> to memref<?xf32>
return return
} }
// CHECK-LABEL: func @sparse_concat_dce(
// CHECK-NOT: sparse_tensor.concatenate
aartbikUnsubmitted
Done
ReplyInline Actions

CHECK-NOT: sparse_tensor.concatenate

;-)

aartbik: CHECK-NOT: sparse_tensor.concatenate ;-)
// CHECK: return
func.func @sparse_concat_dce(%arg0: tensor<2xf64, #SparseVector>,
%arg1: tensor<3xf64, #SparseVector>,
%arg2: tensor<4xf64, #SparseVector>) {
%0 = sparse_tensor.concatenate %arg0, %arg1, %arg2 {dimension = 0 : index}
: tensor<2xf64, #SparseVector>,
tensor<3xf64, #SparseVector>,
tensor<4xf64, #SparseVector> to tensor<9xf64, #SparseVector>
return
}