Last active
October 25, 2022 19:14
-
-
Save AmosLewis/77647fab452c7a6a2a9f275620172d76 to your computer and use it in GitHub Desktop.
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
| func.func @torch.aten.softmax.int$cst_dim(%t: !torch.vtensor<[2,3],f32>) -> !torch.vtensor<[2,3],f32> { | |
| %none = torch.constant.none | |
| %dim = torch.constant.int 1 | |
| %ret = torch.aten.softmax.int %t, %dim, %none : !torch.vtensor<[2,3],f32>, !torch.int, !torch.none -> !torch.vtensor<[2,3],f32> | |
| return %ret : !torch.vtensor<[2,3],f32> | |
| } |
Author
AmosLewis
commented
Oct 20, 2022
(mlir_venv) nod% torch-mlir-opt -torch-backend-to-tosa-backend-pipeline /tmp/softmax.mlir | externals/llvm-project/mlir/utils/generate-test-checks.py
// NOTE: Assertions have been autogenerated by utils/generate-test-checks.py
// The script is designed to make adding checks to
// a test case fast, it is *not* designed to be authoritative
// about what constitutes a good test! The CHECK should be
// minimized and named to reflect the test intent.
// CHECK-LABEL: func.func @torch.aten.softmax.int$cst_dim(
// CHECK-SAME: %[[VAL_0:.*]]: tensor<2x3xf32>) -> tensor<2x3xf32> {
// CHECK: %[[VAL_1:.*]] = "tosa.custom"(%[[VAL_0]]) {dim = 1 : i64, identifier = "softmax"} : (tensor<2x3xf32>) -> tensor<2x3xf32>
// CHECK: return %[[VAL_1]] : tensor<2x3xf32>
// CHECK: }
With pass patch:
➜ torch-mlir git:(tosa-custom-softmax) ✗ torch-mlir-opt -convert-torch-to-tosa /tmp/softmax.mlir | externals/llvm-project/mlir/utils/generate-test-checks.py
// NOTE: Assertions have been autogenerated by utils/generate-test-checks.py
// The script is designed to make adding checks to
// a test case fast, it is *not* designed to be authoritative
// about what constitutes a good test! The CHECK should be
// minimized and named to reflect the test intent.
// CHECK-LABEL: func.func @torch.aten.softmax.int$cst_dim(
// CHECK-SAME: %[[VAL_0:.*]]: !torch.vtensor<[2,3],f32>) -> !torch.vtensor<[2,3],f32> {
// CHECK: %[[VAL_1:.*]] = torch.constant.none
// CHECK: %[[VAL_2:.*]] = torch.constant.int 1
// CHECK: %[[VAL_3:.*]] = torch.aten.softmax.int %[[VAL_0]], %[[VAL_2]], %[[VAL_1]] : !torch.vtensor<[2,3],f32>, !torch.int, !torch.none -> !torch.vtensor<[2,3],f32>
// CHECK: return %[[VAL_3]] : !torch.vtensor<[2,3],f32>
// CHECK: }
➜ torch-mlir git:(tosa-custom-softmax) ✗ torch-mlir-opt -convert-torch-to-tosa="custom-ops=torch.aten.softmax.int" /tmp/softmax.mlir | externals/llvm-project/mlir/utils/generate-test-checks.py
// NOTE: Assertions have been autogenerated by utils/generate-test-checks.py
// The script is designed to make adding checks to
// a test case fast, it is *not* designed to be authoritative
// about what constitutes a good test! The CHECK should be
// minimized and named to reflect the test intent.
// CHECK-LABEL: func.func @torch.aten.softmax.int$cst_dim(
// CHECK-SAME: %[[VAL_0:.*]]: !torch.vtensor<[2,3],f32>) -> !torch.vtensor<[2,3],f32> {
// CHECK: %[[VAL_1:.*]] = torch_c.to_builtin_tensor %[[VAL_0]] : !torch.vtensor<[2,3],f32> -> tensor<2x3xf32>
// CHECK: %[[VAL_2:.*]] = torch.constant.none
// CHECK: %[[VAL_3:.*]] = torch.constant.int 1
// CHECK: %[[VAL_4:.*]] = "tosa.custom"(%[[VAL_1]]) {dim = 1 : i64, identifier = "softmax"} : (tensor<2x3xf32>) -> tensor<2x3xf32>
// CHECK: %[[VAL_5:.*]] = torch_c.from_builtin_tensor %[[VAL_4]] : tensor<2x3xf32> -> !torch.vtensor<[2,3],f32>
// CHECK: return %[[VAL_5]] : !torch.vtensor<[2,3],f32>
// CHECK: }
➜ torch-mlir git:(tosa-custom-softmax) ✗
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment