llvm
/
torch-mlir
mirror of https://github.com/llvm/torch-mlir
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
torch-mlir/test/Dialect/Torch/verify-backend-contract-err...

37 lines
1.4 KiB
MLIR
Raw Normal View History

Enable VerifyBackendContract in LTC backend (#1798) * Enable VerifyBackendContract in LTC backend * Update VerifyBackendContract pass * Move convert_scalar_implicit to jit_utils * Rename VerifyBackendContract to VerifyBackendContractNoDecompositions * Update verify-backend-contract-error.mlir test
2023-01-25 11:14:17 +08:00
// RUN: torch-mlir-opt -torch-verify-backend-contract-no-decompositions -split-input-file -verify-diagnostics %s
Clean up verification of calling conventions. The implementation at this place was a remnent of the times the pipeline was run only once. Rely instead on the backend verification, after optimizations have had an opportunity to resolve some uncertainties. (e.g. `!torch.optional`).
2023-07-18 22:32:26 +08:00
Enable VerifyBackendContract in LTC backend (#1798) * Enable VerifyBackendContract in LTC backend * Update VerifyBackendContract pass * Move convert_scalar_implicit to jit_utils * Rename VerifyBackendContract to VerifyBackendContractNoDecompositions * Update verify-backend-contract-error.mlir test
2023-01-25 11:14:17 +08:00
func.func @f(%arg0: !torch.vtensor<[?,?],f32>) -> !torch.vtensor {
// expected-error @below {{unsupported by backend contract: tensor with unknown rank}}
// expected-note @below {{this is likely due to a missing transfer function}}
%t = torch.aten.t %arg0 : !torch.vtensor<[?,?],f32> -> !torch.vtensor
return %t : !torch.vtensor
Pass op legality info to `verifyBackendContractPass` (#1705) In order to verify if a given IR satisfies the backend contract, the verifier needs to know if decompositions took place, and if so, which ops were decomposed and which were not. This commit adds two arguments to `verifyBackendContractPass` to specify if decompositions took place and which ops to consider backend legal, similar to the arguments of `LowerToBackendContractPass`.
2022-12-16 00:32:52 +08:00
}
Clean up verification of calling conventions. The implementation at this place was a remnent of the times the pipeline was run only once. Rely instead on the backend verification, after optimizations have had an opportunity to resolve some uncertainties. (e.g. `!torch.optional`).
2023-07-18 22:32:26 +08:00
// -----
// expected-error @below {{invalid dtype 'i9'}}
func.func @bad_element_type(%arg: !torch.vtensor<[?],i9>) -> !torch.vtensor<[?],i9> {
return %arg : !torch.vtensor<[?],i9>
}
// -----
// expected-error @below {{unsupported by backend contract: non-value tensor type}}
// expected-note @below {{this is likely due to a missing case in the MaximizeValueSemantics pass}}
func.func @non_value_tensor(%arg0: !torch.tensor) -> !torch.tensor {
return %arg0 : !torch.tensor
}
// -----
func.func @valid_tuple(%arg0: !torch.vtensor<[?],f32>) -> !torch.tuple<vtensor<[?],f32>> {
%0 = torch.prim.TupleConstruct %arg0 : !torch.vtensor<[?],f32> -> !torch.tuple<vtensor<[?],f32>>
return %0 : !torch.tuple<vtensor<[?],f32>>
}
// -----
func.func @valid_multiple_ret_values(%arg0: !torch.vtensor<[?],f32>) -> (!torch.vtensor<[?],f32>, !torch.vtensor<[?],f32>) {
return %arg0, %arg0 : !torch.vtensor<[?],f32>, !torch.vtensor<[?],f32>
}