First pass cleaning the README

pull/333/head
Sean Silva 2021-09-28 00:10:47 +00:00
parent 64ce5d54d3
commit de1a8c92f7
1 changed files with 52 additions and 222 deletions

274
README.md
View File

@ -33,98 +33,30 @@ We have few paths to lower down to the Torch MLIR Dialect.
## Examples ## 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. 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.
## 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
```
## Project Communication ## Project Communication
- `#mlir-npcomp` channel on the LLVM [Discord](https://discord.gg/xS7Z362) - `#torch-mlir` channel on the LLVM [Discord](https://discord.gg/xS7Z362)
- issues/PR's on this github repo - 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 - [`torch-mlir` section](https://llvm.discourse.group/c/projects-that-want-to-become-official-llvm-projects/torch-mlir/41) of LLVM Discourse
### Architecture
The compiler is separated into:
* [Frontend importer](python/npcomp/compiler/numpy/frontend.py): Translates from
various AST levels to corresponding MLIR dialects.
* Frontend compiler: MLIR passes and conversions, mostly operating on the
[basicpy](include/npcomp/Dialect/Basicpy/IR/BasicpyOps.td) and
[numpy](include/npcomp/Dialect/Numpy/IR/NumpyOps.td) dialects.
* Backend compiler and runtime: Some effort has been taken to make this
pluggable, but right now, only the [IREE Backend](python/npcomp/compiler/generic/backend/iree.py)
exists. There is in-tree work to also build a minimal reference backend
directly targeting LLVM.
## Repository Layout ## Repository Layout
The project is roughly split into the following areas of code: The project follows the conventions of typical MLIR-based projects:
* [User-facing Python code](python/npcomp) * `include/torch-mlir`, `lib` structure for C++ MLIR compiler dialects/passes.
* C++ [include](include) and [lib](lib) trees, following LLVM/MLIR conventions * `test` for holding test code.
* LIT testing trees: * `tools` for `torch-mlir-opt` and such.
* [test](test): Lit/FileCheck tests covering core MLIR based infra * `python` top level directory for Python code
* [test/Python/Compiler](test/Python/Compiler): Lit test suite that drive the compiler
infra from Python
* [backend_test](backend_test): Lit test suites conditionally enabled for
each backend
* [tools](tools): Scripts and binaries (npcomp-opt, refback-run, etc)
## Interactive Use ## Interactive Use
The cmake configuration populates symlinks in the `build/python` directory The `build_tools/write_env_file.sh` script will output a `.env`
mirroring the source layout. This allows edit-run without rebuilding (unless
if files are added/removed).
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`
```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`
file in the workspace folder with the correct PYTHONPATH set. This allows file in the workspace folder with the correct PYTHONPATH set. This allows
tools like VSCode to work by default for debugging. tools like VSCode to work by default for debugging. This file can also be
manually `source`'d in a shell.
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
3.x (aka. "python3") on a system at a time (and we can only hope that one
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.
## 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 $WHERE_YOU_CHECKED_OUT_NPCOMP/tools/bash_helpers.sh
```
## Build Instructions ## Build Instructions
### Common prep
```shell ```shell
# From checkout directory. # From checkout directory.
git submodule init git submodule init
@ -136,101 +68,72 @@ export CC=clang-$LLVM_VERSION
export CXX=clang++-$LLVM_VERSION export CXX=clang++-$LLVM_VERSION
export LDFLAGS=-fuse-ld=$(which ld.lld-$LLVM_VERSION) export LDFLAGS=-fuse-ld=$(which ld.lld-$LLVM_VERSION)
# run write_env_file.sh to emit a .env file with needed
# PYTHONPATH setup.
./build_tools/write_env_file.sh
```
### Vanilla - numpy-only, no pytorch
```shell
# Configure npcomp.
cmake -GNinja -Bbuild -DCMAKE_BUILD_TYPE=Release .
# Build and run tests
# ./build_tools/test_all.sh runs all of these commands.
cd build
ninja
ninja check-npcomp
```
### With PyTorch integration
```shell
# Install PyTorch. We currently track and require the nighly build. # Install PyTorch. We currently track and require the nighly build.
# If a usable PyTorch package is installed, the default cmake settings will # If a usable PyTorch package is installed, the default cmake settings will
# enable the PyTorch frontend. # enable the PyTorch frontend.
pip3 install --pre torch torchvision -f https://download.pytorch.org/whl/nightly/cpu/torch_nightly.html pip3 install --pre torch torchvision -f https://download.pytorch.org/whl/nightly/cpu/torch_nightly.html
cmake ... # Invoke CMake and build. This will also run all unit tests.
ninja check-frontends-pytorch # If building with PyTorch ./build_tools/build_standalone.sh
# Run write_env_file.sh to emit a .env file with needed
# PYTHONPATH setup.
./build_tools/write_env_file.sh
source .env
``` ```
### PyTorch Frontend (via docker container) ## Demos
Create docker image (or follow your own preferences): ### TorchScript
* Mount the (host) source directory to `/src/mlir-npcomp` (in the container). A lot of the prior effort on torch-mlir has gone into the TorchScript compiler,
* Mount the `/build` directory (in the container) appropriately for your case. with a path to execution via linalg-on-tensors (a commonly used representation
of tensor computations in the MLIR community). Thus, this path has the most
extensive testing and functionality.
```shell Running execution (end-to-end) tests:
docker build docker/pytorch-nightly --tag local/npcomp:build-pytorch-nightly
docker volume create npcomp-build ```
# Run E2E TorchScript tests. These compile and run the TorchScript program
# through torch-mlir with a simplified linalg-on-tensors based backend we call
# RefBackend (more production-grade backends at this same abstraction layer
# exist in the MLIR community, such as IREE).
./tools/torchscript_e2e_test.sh --filter Conv2d --verbose
``` ```
Shell into docker image: Standalone script:
```shell ```
docker run \ # Run ResNet18 as a standalone script.
--mount type=bind,source=path/to/mlir-npcomp,target=/src/mlir-npcomp \ python examples/torchscript_resnet18_e2e.py
--mount source=npcomp-build,target=/build \
--rm -it local/npcomp:build-pytorch-nightly /bin/bash
``` ```
Build/test npcomp (from within docker image): Jupyter notebook:
```
```shell python -m ipykernel install --user --name=torch-mlir --env PYTHONPATH "$PYTHONPATH"
# From within the docker image. # Open in jupyter, and then navigate to
cd /src/mlir-npcomp # `examples/resnet_inference.ipynb` and use the `torch-mlir` kernel to run.
cmake -GNinja -B/build/npcomp -DCMAKE_BUILD_TYPE=Release . jupyter notebook
cmake --build /build/npcomp --target check-npcomp check-frontends-pytorch
``` ```
### IREE Backend (from IREE packages)
```shell ### TorchFX
# We currently track and require the latest snapshot.
pip3 install iree-compiler-snapshot iree-runtime-snapshot -f https://github.com/google/iree/releases
# Run TorchScript E2E tests targeting IREE. TODO
# Make sure to run "PyTorch Frontend" setup instructions first.
tools/torchscript_e2e_test.sh --config=iree
```
### IREE Backend (from local IREE build)
This configuration is useful for iterating locally, as you can ### Lazy Tensor Core
poke/debug/rebuild things in IREE.
```shell TODO
# Locally build IREE.
# See https://google.github.io/iree/building-from-source/getting-started/
# Make sure IREE is configured with `-DIREE_BUILD_PYTHON_BINDINGS=ON`.
# Add IREE's Python bindings to PYTHONPATH. ### Additional TorchScript end-to-end tests with heavy dependencies
echo 'PYTHONPATH="${PYTHONPATH}:/path/to/iree-build/bindings/python"' >> .env
# Run TorchScript E2E tests targeting IREE. Some of the Torchscript end-to-end tests require additional dependencies which
# Make sure to run "PyTorch Frontend" setup instructions first. don't make sense to include as part of the default torch-mlir setup.
tools/torchscript_e2e_test.sh --config=iree Additionally, these dependencies often don't work with the same HEAD PyTorch
``` version that torch-mlir builds against at the C++ level (the TorchScript
importer is written in C++)
### 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 We have a self-contained script that generates all the needed artifacts from a
self-contained virtual environment. It can be used like so: self-contained virtual environment. It can be used like so:
@ -258,76 +161,3 @@ torchscript_e2e_test.sh, but the tests are added in
We rely critically on serialized TorchScript compatibility across PyTorch We rely critically on serialized TorchScript compatibility across PyTorch
versions to transport the tests + pure-Python compatibility of the `torch` versions to transport the tests + pure-Python compatibility of the `torch`
API, which has worked well so far. API, which has worked well so far.
### VSCode with a Docker Dev Image
#### Start a docker dev container based on our image
Assumes that mlir-npcomp is checked out locally under `~/src/mlir-npcomp`.
See `docker_shell_funcs.sh` for commands to modify if different.
```shell
# Build/start the container.
# Follow instructions here to allow running `docker` without `sudo`:
# https://docs.docker.com/engine/install/linux-postinstall/
source ./build_tools/docker_shell_funcs.sh
npcomp_docker_build # Only needed first time/on updates to docker files.
npcomp_docker_start
```
```shell
# Get an interactive shell to the container and initial build.
npcomp_docker_login
```
```shell
# Stop the container (when done).
npcomp_docker_stop
```
### Configure VSCode:
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`.
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)!