Skip to content

Bump LLVM commit to 20ed5b1f4587 #2799

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 16 commits into from
May 29, 2024
Merged

Bump LLVM commit to 20ed5b1f4587 #2799

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
16 commits
Select commit Hold shift + click to select a range
9011df7
Bump LLVM commit to 1e6ce5e284f5c0e8d64eee21af727bb164eb3caf and
Apr 15, 2024
1e0298c
Bump LLVM to 20ed5b1f4587 and stablehlo to fd0c20a10
flemairen6 Apr 18, 2024
4a4c568
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 Apr 18, 2024
7ef1b58
Remove Sequence_with_dealloc.mlir from the ran tests because it's
flemairen6 Apr 19, 2024
e16c813
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 Apr 23, 2024
a9708e1
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 Apr 25, 2024
c38e5a8
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 May 1, 2024
d95c187
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 May 10, 2024
6d43ba7
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 May 13, 2024
dea4e10
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 May 14, 2024
2dd1d31
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 May 14, 2024
7bbf15a
Merge branch 'main' into ferdinand.update_llvm_april_2024
chentong319 May 17, 2024
60000d4
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 May 20, 2024
f58cad9
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 May 21, 2024
25df99b
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 May 23, 2024
458bc31
Merge branch 'main' into ferdinand.update_llvm_april_2024
hamptonm1 May 29, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion docs/BuildOnLinuxOSX.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@ Firstly, install MLIR (as a part of LLVM-Project):
``` bash
git clone -n https://github.com/llvm/llvm-project.git
# Check out a specific branch that is known to work with ONNX-MLIR.
cd llvm-project && git checkout 7ac7d418ac2b16fd44789dcf48e2b5d73de3e715 && cd ..
cd llvm-project && git checkout 1e6ce5e284f5c0e8d64eee21af727bb164eb3caf && cd ..
```

[same-as-file]: <> (utils/build-mlir.sh)
Expand Down
2 changes: 1 addition & 1 deletion docs/BuildOnWindows.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,7 +52,7 @@ Install MLIR (as a part of LLVM-Project):
```shell
git clone -n https://github.com/llvm/llvm-project.git
# Check out a specific branch that is known to work with ONNX-MLIR.
cd llvm-project && git checkout 7ac7d418ac2b16fd44789dcf48e2b5d73de3e715 && cd ..
cd llvm-project && git checkout 1e6ce5e284f5c0e8d64eee21af727bb164eb3caf && cd ..
```

[same-as-file]: <> (utils/build-mlir.cmd)
Expand Down
3 changes: 1 addition & 2 deletions test/mlir/conversion/onnx_to_krnl/Sequence/Sequence_with_dealloc.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,6 @@ func.func @test_sequence_erase(%arg0: !onnx.Seq<tensor<?x4x5xf32>>) -> tensor<3x
// CHECK-DAG: [[MAP_0_:#.+]] = affine_map<()[s0] -> (s0 - 1)>
// CHECK-DAG: [[MAP_1_:#.+]] = affine_map<()[s0, s1] -> (s1 + s0)>
// CHECK-DAG: [[MAP_2_:#.+]] = affine_map<()[s0, s1] -> (s0)>
// CHECK-DAG: [[MAP_3_:#.+]] = affine_map<()[s0, s1] -> (s0 + s1)>
// CHECK-LABEL: func.func @test_sequence_erase
// CHECK-SAME: ([[PARAM_0_:%.+]]: memref<?xmemref<?x4x5xf32>>) -> memref<3xi64> {
// CHECK-DAG: [[VAR_0_:%.+]] = "krnl.global"() {name = "constant_{{[0-9]+}}", shape = [], value = dense<0> : tensor<i64>} : () -> memref<i64>
Expand Down Expand Up @@ -55,7 +54,7 @@ func.func @test_sequence_erase(%arg0: !onnx.Seq<tensor<?x4x5xf32>>) -> tensor<3x
// CHECK-DAG: [[CST_0_4_:%.+]] = arith.constant 0 : index
// CHECK-NOT: separator of consecutive DAGs
// CHECK-DAG: [[VAR_13_:%.+]] = arith.cmpi slt, [[VAR_12_]], [[CST_0_4_]] : index
// CHECK-DAG: [[VAR_14_:%.+]] = affine.apply [[MAP_3_]](){{.}}[[VAR_12_]], [[VAR_dim_4_]]{{.}}
// CHECK-DAG: [[VAR_14_:%.+]] = affine.apply [[MAP_1_]](){{.}}[[VAR_dim_4_]], [[VAR_12_]]{{.}}
// CHECK: [[VAR_15_:%.+]] = arith.select [[VAR_13_]], [[VAR_14_]], [[VAR_12_]] : index
// CHECK-DAG: [[VAR_16_:%.+]] = "krnl.seqextract"([[VAR_2_]], [[VAR_15_]]) {copy = 1 : ui1} : (memref<?xmemref<?x4x5xf32>>, index) -> memref<?x4x5xf32>
// CHECK-DAG: [[CST_3_:%.+]] = arith.constant 3 : index
Expand Down
2 changes: 1 addition & 1 deletion test/mlir/conversion/onnx_to_stablehlo/NN/Normalization.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@ func.func @test_batch_normalization(%arg0 : tensor<1x3x10x10xf32>) -> tensor<1x3
// CHECK-NEXT: [[VAR_1_:%.+]] = stablehlo.constant dense<1.000000e+00> : tensor<3xf32>
// CHECK-NEXT: [[VAR_2_:%.+]] = stablehlo.constant dense<1.000000e+00> : tensor<3xf32>
// CHECK-NEXT: [[VAR_3_:%.+]] = stablehlo.constant dense<1.000000e+00> : tensor<3xf32>
// CHECK-NEXT: [[VAR_4_:%.+]] = "stablehlo.batch_norm_inference"(%arg0, %0, %1, %2, %3) {epsilon = 1.00000007E-5 : f32, feature_index = 1 : i64} : (tensor<1x3x10x10xf32>, tensor<3xf32>, tensor<3xf32>, tensor<3xf32>, tensor<3xf32>) -> tensor<1x3x10x10xf32>
// CHECK-NEXT: [[VAR_4_:%.+]] = "stablehlo.batch_norm_inference"(%arg0, %0, %1, %2, %3) <{epsilon = 1.00000007E-5 : f32, feature_index = 1 : i64}> : (tensor<1x3x10x10xf32>, tensor<3xf32>, tensor<3xf32>, tensor<3xf32>, tensor<3xf32>) -> tensor<1x3x10x10xf32>
// CHECK-NEXT: return [[VAR_4_]] : tensor<1x3x10x10xf32>
// CHECK-NEXT: }
}
68 changes: 34 additions & 34 deletions test/mlir/conversion/onnx_to_stablehlo/NN/Pooling.mlir
View file
Open in desktop

Large diffs are not rendered by default.

8 changes: 4 additions & 4 deletions test/mlir/conversion/onnx_to_stablehlo/RNN/LSTM-loop.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -145,10 +145,10 @@ func.func @test_lstm_loop(%arg0 : tensor<128x16x512xf32>, %arg1 : tensor<2x2048x
// CHECK: [[VAR_119_:%.+]] = stablehlo.multiply [[VAR_117_]], [[VAR_118_]] : tensor<16x256xf32>
// CHECK-DAG: [[VAR_120_:%.+]] = stablehlo.dynamic_reshape [[VAR_119_]], [[VAR_3_]] : (tensor<16x256xf32>, tensor<4xindex>) -> tensor<1x1x16x256xf32>
// CHECK-DAG: [[VAR_121_:%.+]] = stablehlo.dynamic_reshape [[VAR_iterArg_]], [[VAR_2_]] : (tensor<1xi64>, tensor<2xindex>) -> tensor<1x1xi64>
// CHECK: [[VAR_122_:%.+]] = "stablehlo.scatter"([[VAR_iterArg_0_]], [[VAR_121_]], [[VAR_120_]]) ({
// CHECK: [[VAR_122_:%.+]] = "stablehlo.scatter"([[VAR_iterArg_0_]], [[VAR_121_]], [[VAR_120_]]) <{indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<update_window_dims = [1, 2, 3], inserted_window_dims = [0], scatter_dims_to_operand_dims = [0], index_vector_dim = 1>, unique_indices = false}> ({
// CHECK: ^bb0([[arg4_:%.+]]: tensor<f32>, [[arg5_:%.+]]: tensor<f32>):
// CHECK: stablehlo.return [[arg5_]] : tensor<f32>
// CHECK: }) {indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<update_window_dims = [1, 2, 3], inserted_window_dims = [0], scatter_dims_to_operand_dims = [0], index_vector_dim = 1>, unique_indices = false} : (tensor<128x1x16x256xf32>, tensor<1x1xi64>, tensor<1x1x16x256xf32>) -> tensor<128x1x16x256xf32>
// CHECK: }) : (tensor<128x1x16x256xf32>, tensor<1x1xi64>, tensor<1x1x16x256xf32>) -> tensor<128x1x16x256xf32>
// CHECK-DAG: [[VAR_123_:%.+]] = stablehlo.dynamic_broadcast_in_dim [[VAR_iterArg_]], [[VAR_1_]], dims = [0] : (tensor<1xi64>, tensor<1xindex>) -> tensor<1xi64>
// CHECK-DAG: [[VAR_124_:%.+]] = stablehlo.dynamic_broadcast_in_dim [[VAR_10_]], [[VAR_1_]], dims = [0] : (tensor<1xi64>, tensor<1xindex>) -> tensor<1xi64>
// CHECK: [[VAR_125_:%.+]] = stablehlo.add [[VAR_123_]], [[VAR_124_]] : tensor<1xi64>
Expand Down Expand Up @@ -225,10 +225,10 @@ func.func @test_lstm_loop(%arg0 : tensor<128x16x512xf32>, %arg1 : tensor<2x2048x
// CHECK: [[VAR_119_1_:%.+]] = stablehlo.multiply [[VAR_117_1_]], [[VAR_118_1_]] : tensor<16x256xf32>
// CHECK-DAG: [[VAR_120_1_:%.+]] = stablehlo.dynamic_reshape [[VAR_119_1_]], [[VAR_3_]] : (tensor<16x256xf32>, tensor<4xindex>) -> tensor<1x1x16x256xf32>
// CHECK-DAG: [[VAR_121_1_:%.+]] = stablehlo.dynamic_reshape [[VAR_iterArg_1_]], [[VAR_2_]] : (tensor<1xi64>, tensor<2xindex>) -> tensor<1x1xi64>
// CHECK: [[VAR_122_1_:%.+]] = "stablehlo.scatter"([[VAR_iterArg_0_1_]], [[VAR_121_1_]], [[VAR_120_1_]]) ({
// CHECK: [[VAR_122_1_:%.+]] = "stablehlo.scatter"([[VAR_iterArg_0_1_]], [[VAR_121_1_]], [[VAR_120_1_]]) <{indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<update_window_dims = [1, 2, 3], inserted_window_dims = [0], scatter_dims_to_operand_dims = [0], index_vector_dim = 1>, unique_indices = false}> ({
// CHECK: ^bb0([[arg4_:%.+]]: tensor<f32>, [[arg5_:%.+]]: tensor<f32>):
// CHECK: stablehlo.return [[arg5_]] : tensor<f32>
// CHECK: }) {indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<update_window_dims = [1, 2, 3], inserted_window_dims = [0], scatter_dims_to_operand_dims = [0], index_vector_dim = 1>, unique_indices = false} : (tensor<128x1x16x256xf32>, tensor<1x1xi64>, tensor<1x1x16x256xf32>) -> tensor<128x1x16x256xf32>
// CHECK: }) : (tensor<128x1x16x256xf32>, tensor<1x1xi64>, tensor<1x1x16x256xf32>) -> tensor<128x1x16x256xf32>
// CHECK-DAG: [[VAR_123_1_:%.+]] = stablehlo.dynamic_broadcast_in_dim [[VAR_iterArg_1_]], [[VAR_1_]], dims = [0] : (tensor<1xi64>, tensor<1xindex>) -> tensor<1xi64>
// CHECK-DAG: [[VAR_124_1_:%.+]] = stablehlo.dynamic_broadcast_in_dim [[VAR_10_]], [[VAR_1_]], dims = [0] : (tensor<1xi64>, tensor<1xindex>) -> tensor<1xi64>
// CHECK: [[VAR_125_1_:%.+]] = stablehlo.subtract [[VAR_123_1_]], [[VAR_124_1_]] : tensor<1xi64>
Expand Down
8 changes: 4 additions & 4 deletions test/mlir/conversion/onnx_to_stablehlo/Tensor/Gather.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@ func.func @test_gather_axis0(%arg0 : tensor<3x2xf32>) -> tensor<2x2x2xf32> {
// CHECK-DAG: [[VAR_3_:%.+]] = stablehlo.compare LT, [[VAR_0_]], [[VAR_1_]], NOTYPE : (tensor<2x2xi64>, tensor<2x2xi64>) -> tensor<2x2xi1>
// CHECK-DAG: [[VAR_4_:%.+]] = stablehlo.add [[VAR_0_]], [[VAR_2_]] : tensor<2x2xi64>
// CHECK: [[VAR_5_:%.+]] = stablehlo.select [[VAR_3_]], [[VAR_4_]], [[VAR_0_]] : tensor<2x2xi1>, tensor<2x2xi64>
// CHECK: [[VAR_6_:%.+]] = "stablehlo.torch_index_select"([[PARAM_0_]], [[VAR_5_]]) {batch_dims = 0 : i64, dim = 0 : i64} : (tensor<3x2xf32>, tensor<2x2xi64>) -> tensor<2x2x2xf32>
// CHECK: [[VAR_6_:%.+]] = "stablehlo.torch_index_select"([[PARAM_0_]], [[VAR_5_]]) <{batch_dims = 0 : i64, dim = 0 : i64}> : (tensor<3x2xf32>, tensor<2x2xi64>) -> tensor<2x2x2xf32>
// CHECK: return [[VAR_6_]] : tensor<2x2x2xf32>
// CHECK: }

Expand All @@ -42,7 +42,7 @@ func.func @test_gather_dynamic_axis0(%arg0 : tensor<?x?xf32>) -> tensor<2x2x?xf3
// CHECK-DAG: [[VAR_3_:%.+]] = stablehlo.compare LT, [[VAR_0_]], [[VAR_1_]], NOTYPE : (tensor<2x2xi64>, tensor<2x2xi64>) -> tensor<2x2xi1>
// CHECK-DAG: [[VAR_4_:%.+]] = stablehlo.add [[VAR_0_]], [[VAR_2_]] : tensor<2x2xi64>
// CHECK: [[VAR_5_:%.+]] = stablehlo.select [[VAR_3_]], [[VAR_4_]], [[VAR_0_]] : tensor<2x2xi1>, tensor<2x2xi64>
// CHECK: [[VAR_6_:%.+]] = "stablehlo.torch_index_select"([[PARAM_0_]], [[VAR_5_]]) {batch_dims = 0 : i64, dim = 0 : i64} : (tensor<?x?xf32>, tensor<2x2xi64>) -> tensor<2x2x?xf32>
// CHECK: [[VAR_6_:%.+]] = "stablehlo.torch_index_select"([[PARAM_0_]], [[VAR_5_]]) <{batch_dims = 0 : i64, dim = 0 : i64}> : (tensor<?x?xf32>, tensor<2x2xi64>) -> tensor<2x2x?xf32>
// CHECK: return [[VAR_6_]] : tensor<2x2x?xf32>
// CHECK: }

Expand All @@ -63,7 +63,7 @@ func.func @test_gather_axis0neg(%arg0 : tensor<3x2xf32>) -> tensor<2x2x2xf32> {
// CHECK-DAG: [[VAR_3_:%.+]] = stablehlo.compare LT, [[VAR_0_]], [[VAR_1_]], NOTYPE : (tensor<2x2xi64>, tensor<2x2xi64>) -> tensor<2x2xi1>
// CHECK-DAG: [[VAR_4_:%.+]] = stablehlo.add [[VAR_0_]], [[VAR_2_]] : tensor<2x2xi64>
// CHECK: [[VAR_5_:%.+]] = stablehlo.select [[VAR_3_]], [[VAR_4_]], [[VAR_0_]] : tensor<2x2xi1>, tensor<2x2xi64>
// CHECK: [[VAR_6_:%.+]] = "stablehlo.torch_index_select"([[PARAM_0_]], [[VAR_5_]]) {batch_dims = 0 : i64, dim = 0 : i64} : (tensor<3x2xf32>, tensor<2x2xi64>) -> tensor<2x2x2xf32>
// CHECK: [[VAR_6_:%.+]] = "stablehlo.torch_index_select"([[PARAM_0_]], [[VAR_5_]]) <{batch_dims = 0 : i64, dim = 0 : i64}> : (tensor<3x2xf32>, tensor<2x2xi64>) -> tensor<2x2x2xf32>
// CHECK: return [[VAR_6_]] : tensor<2x2x2xf32>
// CHECK: }

Expand All @@ -84,6 +84,6 @@ func.func @test_gather_axis1(%arg0 : tensor<3x3xf32>) -> tensor<3x1x2xf32> {
// CHECK-DAG: [[VAR_3_:%.+]] = stablehlo.compare LT, [[VAR_0_]], [[VAR_1_]], NOTYPE : (tensor<1x2xi64>, tensor<1x2xi64>) -> tensor<1x2xi1>
// CHECK-DAG: [[VAR_4_:%.+]] = stablehlo.add [[VAR_0_]], [[VAR_2_]] : tensor<1x2xi64>
// CHECK: [[VAR_5_:%.+]] = stablehlo.select [[VAR_3_]], [[VAR_4_]], [[VAR_0_]] : tensor<1x2xi1>, tensor<1x2xi64>
// CHECK: [[VAR_6_:%.+]] = "stablehlo.torch_index_select"([[PARAM_0_]], [[VAR_5_]]) {batch_dims = 0 : i64, dim = 1 : i64} : (tensor<3x3xf32>, tensor<1x2xi64>) -> tensor<3x1x2xf32>
// CHECK: [[VAR_6_:%.+]] = "stablehlo.torch_index_select"([[PARAM_0_]], [[VAR_5_]]) <{batch_dims = 0 : i64, dim = 1 : i64}> : (tensor<3x3xf32>, tensor<1x2xi64>) -> tensor<3x1x2xf32>
// CHECK: return [[VAR_6_]] : tensor<3x1x2xf32>
// CHECK: }
2 changes: 1 addition & 1 deletion test/mlir/conversion/onnx_to_stablehlo/Tensor/GatherElements.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,6 @@ func.func @main_gather_elements(%arg0: tensor<3x2xf32>, %arg1: tensor<2x2xi64>)
// CHECK-DAG: [[VAR_6_:%.+]] = stablehlo.dynamic_reshape [[VAR_5_]], [[CST_]] : (tensor<2x2xi64>, tensor<3xindex>) -> tensor<2x2x1xi64>
// CHECK-DAG: [[VAR_7_:%.+]] = stablehlo.dynamic_iota [[CST_]], dim = 1 : (tensor<3xindex>) -> tensor<2x2x1xi64>
// CHECK-NEXT: [[VAR_8_:%.+]] = stablehlo.concatenate [[VAR_6_]], [[VAR_7_]], dim = 2 : (tensor<2x2x1xi64>, tensor<2x2x1xi64>) -> tensor<2x2x2xi64>
// CHECK-NEXT: [[VAR_9_:%.+]] = "stablehlo.gather"([[PARAM_0_]], [[VAR_8_]]) {dimension_numbers = #stablehlo.gather<collapsed_slice_dims = [0, 1], start_index_map = [0, 1], index_vector_dim = 2>, indices_are_sorted = false, slice_sizes = array<i64: 1, 1>} : (tensor<3x2xf32>, tensor<2x2x2xi64>) -> tensor<2x2xf32>
// CHECK-NEXT: [[VAR_9_:%.+]] = "stablehlo.gather"([[PARAM_0_]], [[VAR_8_]]) <{dimension_numbers = #stablehlo.gather<collapsed_slice_dims = [0, 1], start_index_map = [0, 1], index_vector_dim = 2>, indices_are_sorted = false, slice_sizes = array<i64: 1, 1>}> : (tensor<3x2xf32>, tensor<2x2x2xi64>) -> tensor<2x2xf32>
// CHECK-NEXT: return [[VAR_9_]] : tensor<2x2xf32>
}
12 changes: 6 additions & 6 deletions test/mlir/conversion/onnx_to_stablehlo/Tensor/ScatterND.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,10 +5,10 @@ func.func @test_scatternd_1(%arg0 : tensor<8xf32>, %arg1 : tensor<4x1xi64>, %arg
return %0 : tensor<8xf32>
// CHECK-LABEL: func.func @test_scatternd_1
// CHECK-SAME: ([[PARAM_0_:%.+]]: tensor<8xf32>, [[PARAM_1_:%.+]]: tensor<4x1xi64>, [[PARAM_2_:%.+]]: tensor<4xf32>) -> tensor<8xf32> {
// CHECK: [[VAR_0_:%.+]] = "stablehlo.scatter"([[PARAM_0_]], [[PARAM_1_]], [[PARAM_2_]]) ({
// CHECK: [[VAR_0_:%.+]] = "stablehlo.scatter"([[PARAM_0_]], [[PARAM_1_]], [[PARAM_2_]]) <{indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<inserted_window_dims = [0], scatter_dims_to_operand_dims = [0], index_vector_dim = 1>, unique_indices = false}> ({
// CHECK: ^bb0([[arg3_:%.+]]: tensor<f32>, [[arg4_:%.+]]: tensor<f32>):
// CHECK: stablehlo.return [[arg4_]] : tensor<f32>
// CHECK: }) {indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<inserted_window_dims = [0], scatter_dims_to_operand_dims = [0], index_vector_dim = 1>, unique_indices = false} : (tensor<8xf32>, tensor<4x1xi64>, tensor<4xf32>) -> tensor<8xf32>
// CHECK: }) : (tensor<8xf32>, tensor<4x1xi64>, tensor<4xf32>) -> tensor<8xf32>
// CHECK: return [[VAR_0_]] : tensor<8xf32>
// CHECK: }
}
Expand All @@ -20,10 +20,10 @@ func.func @test_scatternd_2(%arg0 : tensor<4x4x4xi32>, %arg1 : tensor<2x1xi64>,
return %0 : tensor<4x4x4xi32>
// CHECK-LABEL: func.func @test_scatternd_2
// CHECK-SAME: ([[PARAM_0_:%.+]]: tensor<4x4x4xi32>, [[PARAM_1_:%.+]]: tensor<2x1xi64>, [[PARAM_2_:%.+]]: tensor<2x4x4xi32>) -> tensor<4x4x4xi32> {
// CHECK: [[VAR_0_:%.+]] = "stablehlo.scatter"([[PARAM_0_]], [[PARAM_1_]], [[PARAM_2_]]) ({
// CHECK: [[VAR_0_:%.+]] = "stablehlo.scatter"([[PARAM_0_]], [[PARAM_1_]], [[PARAM_2_]]) <{indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<update_window_dims = [1, 2], inserted_window_dims = [0], scatter_dims_to_operand_dims = [0], index_vector_dim = 1>, unique_indices = false}> ({
// CHECK: ^bb0([[arg3_:%.+]]: tensor<i32>, [[arg4_:%.+]]: tensor<i32>):
// CHECK: stablehlo.return [[arg4_]] : tensor<i32>
// CHECK: }) {indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<update_window_dims = [1, 2], inserted_window_dims = [0], scatter_dims_to_operand_dims = [0], index_vector_dim = 1>, unique_indices = false} : (tensor<4x4x4xi32>, tensor<2x1xi64>, tensor<2x4x4xi32>) -> tensor<4x4x4xi32>
// CHECK: }) : (tensor<4x4x4xi32>, tensor<2x1xi64>, tensor<2x4x4xi32>) -> tensor<4x4x4xi32>
// CHECK: return [[VAR_0_]] : tensor<4x4x4xi32>
// CHECK: }
}
Expand All @@ -35,10 +35,10 @@ func.func @test_scatternd_dynamic(%arg0 : tensor<1x?x32x128xf32>, %arg1 : tensor
return %0 : tensor<1x?x32x128xf32>
// CHECK-LABEL: func.func @test_scatternd_dynamic
// CHECK-SAME: ([[PARAM_0_:%.+]]: tensor<1x?x32x128xf32>, [[PARAM_1_:%.+]]: tensor<?x?x32x64x4xi64>, [[PARAM_2_:%.+]]: tensor<?x?x?x?xf32>) -> tensor<1x?x32x128xf32> {
// CHECK: [[VAR_0_:%.+]] = "stablehlo.scatter"([[PARAM_0_]], [[PARAM_1_]], [[PARAM_2_]]) ({
// CHECK: [[VAR_0_:%.+]] = "stablehlo.scatter"([[PARAM_0_]], [[PARAM_1_]], [[PARAM_2_]]) <{indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<inserted_window_dims = [0, 1, 2, 3], scatter_dims_to_operand_dims = [0, 1, 2, 3], index_vector_dim = 4>, unique_indices = false}> ({
// CHECK: ^bb0([[arg3_:%.+]]: tensor<f32>, [[arg4_:%.+]]: tensor<f32>):
// CHECK: stablehlo.return [[arg4_]] : tensor<f32>
// CHECK: }) {indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<inserted_window_dims = [0, 1, 2, 3], scatter_dims_to_operand_dims = [0, 1, 2, 3], index_vector_dim = 4>, unique_indices = false} : (tensor<1x?x32x128xf32>, tensor<?x?x32x64x4xi64>, tensor<?x?x?x?xf32>) -> tensor<1x?x32x128xf32>
// CHECK: }) : (tensor<1x?x32x128xf32>, tensor<?x?x32x64x4xi64>, tensor<?x?x?x?xf32>) -> tensor<1x?x32x128xf32>
// CHECK: return [[VAR_0_]] : tensor<1x?x32x128xf32>
// CHECK: }
}
2 changes: 1 addition & 1 deletion third_party/stablehlo
Open in desktop
Submodule stablehlo updated 92 files
+11 −0 BUILD.bazel
+2 −2 WORKSPACE.bazel
+1 −1 build_tools/github_actions/lint_markdown.sh
+1 −1 build_tools/github_actions/lint_whitespace_checks.sh
+1 −1 build_tools/llvm_version.txt
+3 −3 docs/reference.md
+40 −4 docs/spec.md
+1 −1 docs/status.md
+2 −2 examples/c++/BUILD.bazel
+1 −1 examples/python/stablehlo_add.py
+158 −0 rfcs/20231005-hybrid-quantization.md
+495 −0 stablehlo/conversions/linalg/tests/sparse.mlir
+3 −3 stablehlo/conversions/linalg/transforms/LegalizeToLinalgUtils.cpp
+52 −50 stablehlo/conversions/linalg/transforms/MapStablehloToScalarOp.h
+6 −1 stablehlo/conversions/linalg/transforms/Passes.td
+53 −2 stablehlo/conversions/linalg/transforms/StablehloLegalizeToLinalg.cpp
+2 −2 stablehlo/conversions/linalg/transforms/StablehloToLinalgRandom.cpp
+10 −10 stablehlo/conversions/tosa/transforms/StablehloLegalizeToTosa.cpp
+4 −4 stablehlo/conversions/tosa/transforms/StablehloLegalizeToTosa.pdll
+9 −9 stablehlo/dialect/AssemblyFormat.cpp
+4 −4 stablehlo/dialect/AssemblyFormat.h
+25 −25 stablehlo/dialect/Base.cpp
+40 −1 stablehlo/dialect/Base.h
+43 −15 stablehlo/dialect/Base.td
+3 −3 stablehlo/dialect/BroadcastUtils.cpp
+1 −0 stablehlo/dialect/CMakeLists.txt
+64 −94 stablehlo/dialect/ChloOps.cpp
+0 −55 stablehlo/dialect/ChloOps.td
+1 −1 stablehlo/dialect/StablehloAttrs.td
+37 −0 stablehlo/dialect/StablehloBytecode.cpp
+198 −55 stablehlo/dialect/StablehloOps.cpp
+34 −0 stablehlo/dialect/StablehloOps.h
+148 −71 stablehlo/dialect/StablehloOps.td
+435 −191 stablehlo/dialect/TypeInference.cpp
+7 −0 stablehlo/dialect/TypeInference.h
+1 −1 stablehlo/dialect/Version.h
+15 −15 stablehlo/dialect/VhloBytecode.cpp
+5 −5 stablehlo/dialect/VhloOps.cpp
+10 −10 stablehlo/dialect/VhloTypes.cpp
+5 −4 stablehlo/integrations/c/ChloAttributes.cpp
+67 −106 stablehlo/integrations/c/StablehloAttributes.cpp
+1 −1 stablehlo/integrations/c/StablehloTypes.cpp
+1 −1 stablehlo/integrations/python/StablehloModule.cpp
+3 −3 stablehlo/reference/Api.cpp
+15 −17 stablehlo/reference/Element.cpp
+2 −2 stablehlo/reference/Index.cpp
+4 −2 stablehlo/reference/Index.h
+7 −7 stablehlo/reference/InterpreterOps.cpp
+1 −1 stablehlo/reference/NumPy.cpp
+286 −7 stablehlo/reference/Ops.cpp
+10 −0 stablehlo/reference/Ops.h
+11 −11 stablehlo/reference/Tensor.cpp
+2 −2 stablehlo/reference/Types.cpp
+2 −0 stablehlo/tests/BUILD.bazel
+2 −2 stablehlo/tests/CheckOps.cpp
+60 −0 stablehlo/tests/TestUtils.cpp
+11 −0 stablehlo/tests/TestUtils.td
+438 −440 stablehlo/tests/chlo/chlo_legalize_to_stablehlo.mlir
+2 −1 stablehlo/tests/infer_stablehlo.mlir
+79 −0 stablehlo/tests/interpret/convolution.mlir
+2 −2 stablehlo/tests/interpret/tanh.mlir
+0 −55 stablehlo/tests/ops_chlo.mlir
+0 −18 stablehlo/tests/ops_chlo_roundtrip.mlir
+1,239 −0 stablehlo/tests/ops_speculatability.mlir
+48 −24 stablehlo/tests/ops_stablehlo.mlir
+153 −7 stablehlo/tests/ops_stablehlo_quantized.mlir
+1 −2 stablehlo/tests/print_types_invalid.mlir
+456 −0 stablehlo/tests/stablehlo_aggressive_simplification.mlir
+8 −8 stablehlo/tests/stablehlo_canonicalize_dynamism.mlir
+1 −1 stablehlo/tests/stablehlo_refine_shapes.mlir
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_10_0.mlir.bc
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_11_0.mlir.bc
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_12_0.mlir.bc
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_13_0.mlir.bc
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_14_0.mlir.bc
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_15_0.mlir.bc
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_16_0.mlir.bc
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_17_0.mlir.bc
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_18_0.mlir.bc
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_19_0.mlir.bc
+ − stablehlo/tests/vhlo/stablehlo_legalize_to_vhlo.0_9_0.mlir.bc
+1 −1 stablehlo/tools/StablehloTranslateMain.cpp
+103 −105 stablehlo/transforms/ChloLegalizeToStablehlo.cpp
+13 −13 stablehlo/transforms/ShapeLegalizeToStablehlo.cpp
+123 −9 stablehlo/transforms/StablehloAggressiveSimplification.cpp
+2 −2 stablehlo/transforms/StablehloCanonicalizeDynamism.cpp
+2 −2 stablehlo/transforms/StablehloInstrumentWithProbe.cpp
+40 −47 stablehlo/transforms/StablehloLegalizeToVhlo.cpp
+4 −4 stablehlo/transforms/StablehloRefineArguments.cpp
+8 −8 stablehlo/transforms/StablehloRefineShapes.cpp
+90 −83 stablehlo/transforms/VhloLegalizeToStablehlo.cpp
+12 −12 stablehlo/transforms/VhloToVersion.cpp
2 changes: 1 addition & 1 deletion utils/clone-mlir.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,3 +1,3 @@
git clone -n https://github.com/llvm/llvm-project.git
# Check out a specific branch that is known to work with ONNX-MLIR.
cd llvm-project && git checkout 7ac7d418ac2b16fd44789dcf48e2b5d73de3e715 && cd ..
cd llvm-project && git checkout 1e6ce5e284f5c0e8d64eee21af727bb164eb3caf && cd ..
Loading