llvm
/
torch-mlir
mirror of https://github.com/llvm/torch-mlir
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Document requirements for `torch_mlir_e2e_test` (#3722) 

This documents which CMake options must be set to be able to use
`torch_mlir_e2e_test`, required e.g. for
`projects/pt1/tools/e2e_test.sh`.

Makes progress on #3696.
Closes #3719.
pull/3730/head
Marius Brehler 2024-09-24 00:06:54 +02:00 committed by GitHub
parent 99848265c3
commit e4f2bdf0db
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 11 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
docs/development.md
View File

@ -109,6 +109,15 @@ cmake -GNinja -Bbuild \
-DLLVM_ENABLE_ASSERTIONS=ON \ -DLLVM_ENABLE_ASSERTIONS=ON \
``` ```
#### Flags to run end-to-end tests:
Running the end-to-end execution tests locally requires enabling the native PyTorch extension features and the JIT IR importer, which depends on the
former and defaults to `ON` if not changed:
```shell
-DTORCH_MLIR_ENABLE_PYTORCH_EXTENSIONS=ON \
-DTORCH_MLIR_ENABLE_JIT_IR_IMPORTER=ON \
```
### Building against a pre-built LLVM ### Building against a pre-built LLVM
If you have built llvm-project separately in the directory `$LLVM_INSTALL_DIR`, you can also build the project *out-of-tree* using the following command as template: If you have built llvm-project separately in the directory `$LLVM_INSTALL_DIR`, you can also build the project *out-of-tree* using the following command as template:
@ -396,6 +405,8 @@ Torch-MLIR has two types of tests:
a homegrown testing framework (see a homegrown testing framework (see
`projects/pt1/python/torch_mlir_e2e_test/framework.py`) and the test suite `projects/pt1/python/torch_mlir_e2e_test/framework.py`) and the test suite
lives at `projects/pt1/python/torch_mlir_e2e_test/test_suite/__init__.py`. lives at `projects/pt1/python/torch_mlir_e2e_test/test_suite/__init__.py`.
The tests require to build with `TORCH_MLIR_ENABLE_PYTORCH_EXTENSIONS` (and
the dependent option `TORCH_MLIR_ENABLE_JIT_IR_IMPORTER`) set to `ON`.
2. Compiler and Python API unit tests. These use LLVM's `lit` testing framework. 2. Compiler and Python API unit tests. These use LLVM's `lit` testing framework.
For example, these might involve using `torch-mlir-opt` to run a pass and For example, these might involve using `torch-mlir-opt` to run a pass and