Skip to content

Backward compatible ML compute opset inspired by HLO/MHLO

License

Notifications You must be signed in to change notification settings

aartbik/stablehlo

 
 

Repository files navigation

StableHLO

Official Documentation: https://openxla.org/stablehlo

StableHLO is an operation set for high-level operations (HLO) in machine learning (ML) models. Essentially, it's a portability layer between different ML frameworks and ML compilers: ML frameworks that produce StableHLO programs are compatible with ML compilers that consume StableHLO programs.

Our goal is to simplify and accelerate ML development by creating more interoperability between various ML frameworks (such as TensorFlow, JAX and PyTorch) and ML compilers (such as XLA and IREE).

StableHLO is based on the MHLO dialect and enhances it with additional functionality, including serialization and versioning. We use MLIR bytecode as serialization format and provide backward and forward compatibility guarantees. This ensures compatibility between frameworks and compilers, even as StableHLO continues to evolve.

This repository includes the StableHLO specification along with an MLIR-based implementation in C++ and Python, which you can use to define StableHLO programs for consumption by compilers such as XLA and IREE, as well as on-device use via Google AI Edge.

Build instructions

Here's how to build the StableHLO repo on Linux or macOS:

  1. CMake is our primary build tool, so before you begin make sure that you have CMake and Ninja installed.

    If you're using Linux, we recommend installing lld as well - we have observed it to be noticeably faster than alternatives on our typical software and hardware configurations.

    # On Linux
    sudo apt install cmake ninja-build lld
    
    # On macOS
    brew install cmake ninja
  2. Set the LLVM_ENABLE_LLD shell variable depending on your preferences. We recommend setting it to ON on Linux and to OFF on macOS.

    [[ "$(uname)" != "Darwin" ]] && LLVM_ENABLE_LLD="ON" || LLVM_ENABLE_LLD="OFF"
  3. Clone the StableHLO repo and the LLVM repository:

    git clone https://github.com/openxla/stablehlo
    cd stablehlo && git clone https://github.com/llvm/llvm-project.git

    Cloning the LLVM repository may take a few minutes.

  4. Make sure you check out the correct commit in the LLVM repository:

    (cd llvm-project && git fetch && git checkout $(cat ../build_tools/llvm_version.txt))

    You need to do this every time llvm_version.txt changes.

  5. Configure and build MLIR:

    MLIR_ENABLE_BINDINGS_PYTHON=OFF build_tools/build_mlir.sh ${PWD}/llvm-project/ ${PWD}/llvm-build

    This will take a considerable amount of time. For example, on a MacBook Pro with an M1 Pro chip, building MLIR took around 10 minutes at the moment of writing.

    Again, you need to do this every time llvm_version.txt changes.

  6. Build StableHLO as a standalone library:

    mkdir -p build && cd build
    
    cmake .. -GNinja \
      -DLLVM_ENABLE_LLD="$LLVM_ENABLE_LLD" \
      -DCMAKE_BUILD_TYPE=Release \
      -DLLVM_ENABLE_ASSERTIONS=ON \
      -DSTABLEHLO_ENABLE_BINDINGS_PYTHON=OFF \
      -DMLIR_DIR=${PWD}/../llvm-build/lib/cmake/mlir
    
    cmake --build .

    If you are actively developing StableHLO, you may want the following additional CMake settings:

    cmake .. -GNinja \
      -DSTABLEHLO_ENABLE_LLD=ON \
      -DCMAKE_BUILD_TYPE=RelWithDebInfo \
      -DLLVM_ENABLE_ASSERTIONS=ON \
      -DSTABLEHLO_ENABLE_BINDINGS_PYTHON=OFF \
      -DSTABLEHLO_ENABLE_SPLIT_DWARF=ON \
      -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
      -DCMAKE_C_COMPILER_LAUNCHER=ccache \
      -DCMAKE_EXPORT_COMPILE_COMMANDS=ON \
      -DSTABLEHLO_ENABLE_SANITIZER=address \
      -DMLIR_DIR=${PWD}/../llvm-build/lib/cmake/mlir
    
    cmake --build .

    This will enable debug symbols and ccache, which can speed up incremental builds. It also creates a GDB index file in the binary to speed up debugging.

    If you build MLIR using the script above it should also set by default LLVM_USE_SPLIT_DWARF which does the majority of the size saving for the binary and should also be set.

  7. Now you can make sure it works by running some tests:

    ninja check-stablehlo-tests

    You should see results like this:

    Testing Time: 5.99s
      Passed: 47

    This runs all the tests in stablehlo/tests/.

Python

If you'd like to build the Python bindings, you'll need to install a few additional dependencies.

pip install -r ./llvm-project/mlir/python/requirements.txt

Then build StableHLO with python bindings enabled:

STABLEHLO_ENABLE_BINDINGS_PYTHON=ON ./build_tools/github_actions/ci_build_cmake.sh ${PWD}/llvm-build ${PWD}/build

After you have built the project you can import the Python bindings to begin by modifying your Python path variable

$ PYTHONPATH="./build/python_packages/stablehlo" python3
Python 3.11.6 (main, Oct  8 2023, 05:06:43) [GCC 13.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import mlir.dialects.stablehlo
>>> from mlir.ir import Context, Location
>>> import mlir.dialects.arith

You can also build a wheel yourself using the setup.py file. We also make nightly wheels available on our GitHub Releases page.

pip install stablehlo -f https://github.com/openxla/stablehlo/releases/expanded_assets/dev-wheels

StableHLO to TensorFlow SavedModel

This repository offers tooling for the conversion of a StableHLO program, including its metadata (representing trained weights and biases), into a TensorFlow SavedModel. Please refer to README.md for details.

Community

Building an amazing portability layer between ML frameworks and ML compilers requires collaboration across the whole ML industry, so we're happy to have your help on the StableHLO project.

We're using GitHub issues / pull requests to organize development and openxla-discuss to have longer discussions. We also have a #stablehlo channel on the OpenXLA Discord server.

About

Backward compatible ML compute opset inspired by HLO/MHLO

Resources

License

Code of conduct

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • MLIR 70.7%
  • C++ 26.0%
  • Python 1.1%
  • Starlark 0.8%
  • CMake 0.6%
  • Shell 0.4%
  • C 0.4%