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

Update out of date docs (#2602) 

Some of docs referred to old file paths that no longer exists. This
patch updates some of the instructions that I happened to notice were
out of date. This is not a full update
pull/2579/merge
srcarroll 2023-12-01 16:29:37 -06:00 committed by GitHub
parent e568f7e999
commit 7d0f5cc5a8
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 10 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
build_tools/write_env_file.sh
View File

@ -13,7 +13,7 @@ portable_realpath() {
td="$(portable_realpath "$(dirname "$0")"/..)"
build_dir="$(portable_realpath "${TORCH_MLIR_BUILD_DIR:-$td/build}")"
python_packages_dir="$build_dir/tools/torch-mlir/python_packages"
python_packages_dir="$build_dir/python_packages"
write_env_file() {
echo "Updating $build_dir/.env file"

2
docs/adding_an_e2e_test.md
View File

@ -5,7 +5,7 @@
Adding support for a Torch operator in Torch-MLIR should always be accompanied
by at least one end-to-end test to make sure the implementation of the op
matches the behavior of PyTorch. The tests live in the
`torch-mlir/python/torch_mlir_e2e_test/test_suite/` directory. When adding a new
`torch-mlir/projects/pt1/python/torch_mlir_e2e_test/test_suite` directory. When adding a new
test, choose a file that best matches the op you're testing, and if there is no
file that best matches add a new file for your op.

15
docs/development.md
View File

@ -20,6 +20,7 @@ source mlir_venv/bin/activate
python -m pip install --upgrade pip
# Install latest PyTorch nightlies and build requirements.
python -m pip install -r requirements.txt
python -m pip install -r torchvision-requirements.txt
```
## CMake Build
@ -108,25 +109,25 @@ cmake --build build
### Linux and macOS
```shell
export PYTHONPATH=`pwd`/build/tools/torch-mlir/python_packages/torch_mlir:`pwd`/examples
export PYTHONPATH=`pwd`/build/python_packages/torch_mlir:`pwd`/projects/pt1/examples
```
### Windows PowerShell
```shell
$env:PYTHONPATH = "$PWD/build/tools/torch-mlir/python_packages/torch_mlir;$PWD/examples"
$env:PYTHONPATH = "$PWD/build/python_packages/torch_mlir;$PWD/projects/pt1/examples"
```
## Testing MLIR output in various dialects
To test the compiler's output to the different MLIR dialects, you can use the example `examples/torchscript_resnet18_all_output_types.py`.
To test the compiler's output to the different MLIR dialects, you can use the example `projects/pt1/examples/torchscript_resnet18_all_output_types.py`.
Make sure you have activated the virtualenv and set the `PYTHONPATH` above
(if running on Windows, modify the environment variable as shown above):
```shell
source mlir_venv/bin/activate
export PYTHONPATH=`pwd`/build/tools/torch-mlir/python_packages/torch_mlir:`pwd`/examples
python examples/torchscript_resnet18_all_output_types.py
export PYTHONPATH=`pwd`/build/tpython_packages/torch_mlir:`pwd`/projects/pt1/examples
python projects/pt1/examples/torchscript_resnet18_all_output_types.py
```
This will display the Resnet18 network example in three dialects: TORCH, LINALG on TENSORS and TOSA.
@ -331,8 +332,8 @@ Torch-MLIR has two types of tests:
1. End-to-end execution tests. These compile and run a program and check the
result against the expected output from execution on native Torch. These use
a homegrown testing framework (see
`python/torch_mlir_e2e_test/torchscript/framework.py`) and the test suite
lives at `python/torch_mlir_e2e_test/test_suite/__init__.py`.
`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`.
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