2021-09-24 02:27:03 +08:00
# torch-mlir
The Torch-MLIR project aims to provide first class support from the [Pytorch ](https://pytorch.org ) ecosystem to the MLIR ecosystem.
2020-08-01 11:53:09 +08:00
> This project is participating in the LLVM Incubator process: as such, it is
not part of any official LLVM release. While incubation status is not
necessarily a reflection of the completeness or stability of the code, it
does indicate that the project is not yet endorsed as a component of LLVM.
2021-09-24 02:27:03 +08:00
[Pytorch ](https://pytorch.org )
An open source machine learning framework that accelerates the path from research prototyping to production deployment.
[MLIR ](https://mlir.llvm.org )
The MLIR project is a novel approach to building reusable and extensible compiler infrastructure. MLIR aims to address software fragmentation, improve compilation for heterogeneous hardware, significantly reduce the cost of building domain specific compilers, and aid in connecting existing compilers together.
2021-09-24 03:29:35 +08:00
[Torch-MLIR ](https://github.com/llvm/torch-mlir )
2021-09-24 02:27:03 +08:00
Multiple Vendors use MLIR as the middle layer mapping from Platform Frameworks like Pytorch, JAX, Tensorflow onto MLIR and then progressively lower down to their target hardware. We have seen half a dozen custom lowerings from PyTorch to MLIR. Having a canonical lowering from the Pytorch ecosystem to the MLIR ecosystem would provide much needed relief to Hardware Vendors to focus on their unique value rather than implementing another Pytorch frontend for MLIR. It would be similar to current hardware vendors adding LLVM target support instead of each one also implementing the Clang/C++ frontend.
## All the roads from PyTorch to Torch MLIR Dialect
We have few paths to lower down to the Torch MLIR Dialect.
![Torch Lowering Architectures ](Torch-MLIR.png )
- Torchscript
This is the most tested path down to Torch MLIR Dialect.
- TorchFX
This provides a path to lower from TorchFX down to MLIR. This a functional prototype that we expect to mature as TorchFX matures
- Lazy Tensor Core (Based on lazy-tensor-core [staging branch ](https://github.com/pytorch/pytorch/tree/lazy_tensor_staging/lazy_tensor_core ))
This path provides the upcoming LTC path of capture. It is based of an unstable devel branch but is the closest way for you to adapt any existing torch_xla derivatives.
- “ACAP” - Deprecated torch_xla based capture Mentioned here for completeness.
## Examples
There are few examples of lowering down via path from PyTorch to MLIR and using the “mlir-cpu-runner” to target a CPU backend. Obviously this is just a starting point and you can import this project into your larger MLIR projects to continue lowering to target GPUs and other Accelerators.
2020-08-01 11:53:09 +08:00
2021-07-30 01:43:18 +08:00
## Quick Build and Install with PyTorch
Instructions below go in to detail about how to get a functioning development
setup. But to just build and install into a local python instance, the
instructions are simple. This support is new and we still need to get
dependencies pinned and make these packages distributable. This should work
locally, though:
```
python -m pip install --pre torch torchvision -f https://download.pytorch.org/whl/nightly/cpu/torch_nightly.html
./build_tools/build_python_wheels.sh
```
2021-04-03 04:00:47 +08:00
## Project Communication
- `#mlir-npcomp` channel on the LLVM [Discord ](https://discord.gg/xS7Z362 )
- issues/PR's on this github repo
- [`mlir-npcomp` section ](https://llvm.discourse.group/c/projects-that-want-to-become-official-llvm-projects/mlir-npcomp/41 ) of LLVM Discourse
2020-08-01 11:53:09 +08:00
### Architecture
2020-06-21 02:56:13 +08:00
The compiler is separated into:
2021-01-27 02:02:25 +08:00
* [Frontend importer ](python/npcomp/compiler/numpy/frontend.py ): Translates from
2020-06-21 02:56:13 +08:00
various AST levels to corresponding MLIR dialects.
* Frontend compiler: MLIR passes and conversions, mostly operating on the
2021-01-27 01:52:46 +08:00
[basicpy ](include/npcomp/Dialect/Basicpy/IR/BasicpyOps.td ) and
[numpy ](include/npcomp/Dialect/Numpy/IR/NumpyOps.td ) dialects.
2020-06-21 02:56:13 +08:00
* Backend compiler and runtime: Some effort has been taken to make this
2021-01-27 02:02:25 +08:00
pluggable, but right now, only the [IREE Backend ](python/npcomp/compiler/generic/backend/iree.py )
2020-06-21 02:56:13 +08:00
exists. There is in-tree work to also build a minimal reference backend
directly targeting LLVM.
## Repository Layout
The project is roughly split into the following areas of code:
* [User-facing Python code ](python/npcomp )
* C++ [include ](include ) and [lib ](lib ) trees, following LLVM/MLIR conventions
* LIT testing trees:
* [test ](test ): Lit/FileCheck tests covering core MLIR based infra
2020-08-05 07:59:40 +08:00
* [test/Python/Compiler ](test/Python/Compiler ): Lit test suite that drive the compiler
2020-06-21 02:56:13 +08:00
infra from Python
* [backend_test ](backend_test ): Lit test suites conditionally enabled for
each backend
2021-08-04 08:53:31 +08:00
* [tools ](tools ): Scripts and binaries (npcomp-opt, refback-run, etc)
2020-04-30 01:55:35 +08:00
2020-05-07 13:44:12 +08:00
## Interactive Use
2020-05-03 10:54:49 +08:00
2020-05-07 13:44:12 +08:00
The cmake configuration populates symlinks in the `build/python` directory
mirroring the source layout. This allows edit-run without rebuilding (unless
if files are added/removed).
2020-04-27 06:50:23 +08:00
2021-07-29 00:52:13 +08:00
For using with any interactive tooling (`python3`, Jupyter/Colab, etc) it should be
sufficient to add `python_packages/npcomp_torch/` and `python_packages/npcomp_core/`
directories under `build` directory to `PYTHONPATH`
2020-04-27 06:50:23 +08:00
2021-07-29 00:52:13 +08:00
```shell
export PYTHONPATH=$(cd build & & pwd)/python_packages/npcomp_torch:$(cd build & & pwd)/python_packages/npcomp_core
```
Note that running the `build_tools/write_env_file.sh` script will output a `.env`
2020-05-07 14:25:04 +08:00
file in the workspace folder with the correct PYTHONPATH set. This allows
tools like VSCode to work by default for debugging.
2020-04-27 06:50:23 +08:00
Notes:
* Python sources are symlinked to the output directory at configure time.
Adding sources will require a reconfigure. Editing should not.
* It is a very common issue to have both python 2.7 (aka. "python") and python
2020-06-20 01:38:51 +08:00
3.x (aka. "python3") on a system at a time (and we can only hope that one
2020-04-27 06:50:23 +08:00
day this ends). Since the native library at development time binds to a
specific version, if you try to run with a different python, you will get
an error about the "native" module not being found.
2020-04-27 11:14:20 +08:00
2020-05-30 07:39:00 +08:00
## Compiler development
For bash users, adding the following to your `.bashrc` defines some aliases
that are useful during compiler development, such as shortcuts for builing
and running `npcomp-opt` .
```
source $WHERE_YOU_CHECKED_OUT_NPCOMP/tools/bash_helpers.sh
```
2020-09-17 12:57:46 +08:00
## Build Instructions
### Common prep
```shell
# From checkout directory.
git submodule init
git submodule update
# Use clang and lld to build (optional but recommended).
LLVM_VERSION=10
export CC=clang-$LLVM_VERSION
export CXX=clang++-$LLVM_VERSION
export LDFLAGS=-fuse-ld=$(which ld.lld-$LLVM_VERSION)
2021-07-29 00:52:13 +08:00
# run write_env_file.sh to emit a .env file with needed
# PYTHONPATH setup.
./build_tools/write_env_file.sh
2020-09-17 12:57:46 +08:00
```
### Vanilla - numpy-only, no pytorch
```shell
2021-07-28 07:10:10 +08:00
# Configure npcomp.
cmake -GNinja -Bbuild -DCMAKE_BUILD_TYPE=Release .
2020-09-17 12:57:46 +08:00
# Build and run tests
# ./build_tools/test_all.sh runs all of these commands.
cd build
ninja
ninja check-npcomp
```
2021-07-28 07:10:10 +08:00
### With PyTorch integration
2020-11-04 05:46:46 +08:00
```shell
2021-07-10 03:22:45 +08:00
# Install PyTorch. We currently track and require the nighly build.
# If a usable PyTorch package is installed, the default cmake settings will
# enable the PyTorch frontend.
pip3 install --pre torch torchvision -f https://download.pytorch.org/whl/nightly/cpu/torch_nightly.html
2021-09-11 02:44:38 +08:00
cmake ...
2021-07-28 07:10:10 +08:00
ninja check-frontends-pytorch # If building with PyTorch
2020-11-04 05:46:46 +08:00
```
### PyTorch Frontend (via docker container)
2020-09-17 12:57:46 +08:00
2020-10-08 09:31:24 +08:00
Create docker image (or follow your own preferences):
2020-09-17 12:57:46 +08:00
2020-10-08 09:31:24 +08:00
* Mount the (host) source directory to `/src/mlir-npcomp` (in the container).
* Mount the `/build` directory (in the container) appropriately for your case.
2020-09-17 12:57:46 +08:00
2020-10-08 09:31:24 +08:00
```shell
2020-10-16 09:28:30 +08:00
docker build docker/pytorch-nightly --tag local/npcomp:build-pytorch-nightly
2020-10-08 09:31:24 +08:00
docker volume create npcomp-build
```
Shell into docker image:
2020-09-17 12:57:46 +08:00
```shell
docker run \
2021-08-04 05:54:11 +08:00
--mount type=bind,source=path/to/mlir-npcomp,target=/src/mlir-npcomp \
2020-10-08 09:31:24 +08:00
--mount source=npcomp-build,target=/build \
2020-10-16 09:28:30 +08:00
--rm -it local/npcomp:build-pytorch-nightly /bin/bash
2020-09-17 12:57:46 +08:00
```
2020-10-08 09:31:24 +08:00
Build/test npcomp (from within docker image):
2020-09-17 12:57:46 +08:00
```shell
# From within the docker image.
2020-10-08 09:31:24 +08:00
cd /src/mlir-npcomp
2021-09-11 02:44:38 +08:00
cmake -GNinja -B/build/npcomp -DCMAKE_BUILD_TYPE=Release .
2020-10-08 09:31:24 +08:00
cmake --build /build/npcomp --target check-npcomp check-frontends-pytorch
2020-09-17 12:57:46 +08:00
```
2021-07-01 05:13:21 +08:00
### IREE Backend (from IREE packages)
```shell
# We currently track and require the latest snapshot.
pip3 install iree-compiler-snapshot iree-runtime-snapshot -f https://github.com/google/iree/releases
# Run TorchScript E2E tests targeting IREE.
# Make sure to run "PyTorch Frontend" setup instructions first.
2021-07-03 01:56:42 +08:00
tools/torchscript_e2e_test.sh --config=iree
2021-07-01 05:13:21 +08:00
```
### IREE Backend (from local IREE build)
This configuration is useful for iterating locally, as you can
poke/debug/rebuild things in IREE.
```shell
# Locally build IREE.
# See https://google.github.io/iree/building-from-source/getting-started/
# Make sure IREE is configured with `-DIREE_BUILD_PYTHON_BINDINGS=ON`.
2021-07-03 01:56:42 +08:00
# Add IREE's Python bindings to PYTHONPATH.
2021-07-01 05:13:21 +08:00
echo 'PYTHONPATH="${PYTHONPATH}:/path/to/iree-build/bindings/python"' >> .env
# Run TorchScript E2E tests targeting IREE.
# Make sure to run "PyTorch Frontend" setup instructions first.
2021-07-03 01:56:42 +08:00
tools/torchscript_e2e_test.sh --config=iree
2021-07-01 05:13:21 +08:00
```
2021-07-10 03:22:45 +08:00
### Additional end-to-end tests with heavy dependencies (heavydep tests)
Some end-to-end tests require additional dependencies which don't make sense to
include as part of the default npcomp setup. Additionally, these dependencies
often don't work with the same HEAD PyTorch version that npcomp builds against
at the C++ level.
We have a self-contained script that generates all the needed artifacts from a
self-contained virtual environment. It can be used like so:
```shell
# Build the virtual environment in the specified directory and generate the
# serialized test artifacts in the other specified directory.
# This command is safe to re-run if you have already built the virtual
# environment and just changed the tests.
build_tools/torchscript_e2e_heavydep_tests/generate_serialized_tests.sh \
path/to/heavydep_venv \
path/to/heavydep_serialized_tests
# Add the --serialized-test-dir flag to point at the directory containing the
# serialized tests. All other functionality is the same as the normal invocation
# of torchscript_e2e_test.sh, but the serialized tests will be available.
2021-08-04 05:59:58 +08:00
tools/torchscript_e2e_test.sh --serialized-test-dir=path/to/heavydep_serialized_tests
2021-07-10 03:22:45 +08:00
```
2021-08-04 05:59:58 +08:00
The tests use the same (pure-Python) test framework as the normal
torchscript_e2e_test.sh, but the tests are added in
2021-07-10 03:22:45 +08:00
`build_tools/torchscript_e2e_heavydep_tests` instead of
`frontends/pytorch/e2e_testing/torchscript` .
We rely critically on serialized TorchScript compatibility across PyTorch
versions to transport the tests + pure-Python compatibility of the `torch`
API, which has worked well so far.
2021-07-01 05:13:21 +08:00
2020-10-08 09:31:24 +08:00
### VSCode with a Docker Dev Image
2020-09-17 12:57:46 +08:00
2020-10-08 09:31:24 +08:00
#### Start a docker dev container based on our image
2020-09-25 08:40:25 +08:00
2020-10-08 09:31:24 +08:00
Assumes that mlir-npcomp is checked out locally under `~/src/mlir-npcomp` .
See `docker_shell_funcs.sh` for commands to modify if different.
2020-09-25 08:40:25 +08:00
```shell
2020-10-08 09:31:24 +08:00
# Build/start the container.
2020-10-10 01:21:07 +08:00
# Follow instructions here to allow running `docker` without `sudo`:
# https://docs.docker.com/engine/install/linux-postinstall/
2020-10-08 09:31:24 +08:00
source ./build_tools/docker_shell_funcs.sh
npcomp_docker_build # Only needed first time/on updates to docker files.
npcomp_docker_start
2020-09-25 08:40:25 +08:00
```
```shell
2020-10-08 09:31:24 +08:00
# Get an interactive shell to the container and initial build.
npcomp_docker_login
2020-09-25 08:40:25 +08:00
```
```shell
2020-10-08 09:31:24 +08:00
# Stop the container (when done).
npcomp_docker_stop
2020-09-25 08:40:25 +08:00
```
2020-10-08 09:31:24 +08:00
### Configure VSCode:
2020-10-10 01:21:07 +08:00
First, install the [VSCode Docker
extension](https://marketplace.visualstudio.com/items?itemName=ms-azuretools.vscode-docker) and [VSCode Remote - Containers ](https://marketplace.visualstudio.com/items?itemName=ms-vscode-remote.remote-containers ) extension.
Follow instructions here to allow running `docker` without `sudo` ,
otherwise VSCode won't be able to use docker
https://docs.docker.com/engine/install/linux-postinstall/
(Note that VSCode has some daemons that you will need to kill/restart for
the instructions there to take effect; consider just rebooting your
machine)
Attach to your running container by opening the Docker extension tab (left panel), right clicking on the container name, and selecting "Attach Visual Studio code". The container name if you are using docker_shell_funcs.sh is `npcomp` .
2020-10-08 09:31:24 +08:00
Install extensions in container:
* CMake Tools
* C/C++
* C++ Intellisense
#### Add workspace folders:
* `mlir-npcomp` source folder
* `external/llvm-project` source folder
#### Configure general settings:
`Ctrl-Shift-P` > `Preferences: Open Settings (UI)`
* For `mlir-npcomp` folder:
* `Cmake: Build directory` : `/build/npcomp`
* Uncheck `Cmake: Configure On Edit` and `Cmake: Configure on Open`
* For `llvm-project` folder:
* `Cmake: Build directory` : `/build/llvm-build`
* Uncheck `Cmake: Configure On Edit` and `Cmake: Configure on Open`
#### Configure Intellisense:
`Ctrl-Shift-P` > `C/C++: Edit Configurations (UI)`
* Open C/C++ config (for each project folder):
* Under Advanced, Compile Commands:
* set `/build/npcomp/compile_commands.json` for mlir-npcomp
* set `/build/llvm-build/compile_commands.json` for llvm-project
* Open a C++ file, give it a few seconds and see if you get code completion
(press CTRL-Space).
Make sure to save your workspace (prefer a local folder with the "Use Local" button)!