installing-mlir-on-macos.md

Installing LLVM MLIR on MacOS

This gist gives instructions on how LLVM MLIR can be installed on MacOS. It is also available as a blog post on my website.

Build dependencies

On MacOS, building MLIR requires installing ninja and cmake, along with xcode's developer tools.

xcode-select --install
brew install ninja cmake

Build commands

The following commands modified from https://mlir.llvm.org/getting_started/#unix-like-compiletesting can then be run to build MLIR. The changes include only shallow copying the LLVM repo and including the commented out clang flags for CMake.

git clone --depth=1 https://github.com/llvm/llvm-project.git
mkdir llvm-project/build
cd llvm-project/build
cmake -G Ninja ../llvm \
   -DLLVM_ENABLE_PROJECTS=mlir \
   -DLLVM_BUILD_EXAMPLES=ON \
   -DLLVM_TARGETS_TO_BUILD="Native;NVPTX;AMDGPU" \
   -DCMAKE_BUILD_TYPE=Release \
   -DLLVM_ENABLE_ASSERTIONS=ON \
   -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++
cmake --build . --target check-mlir

Including built binaries in the PATH

Then, the built binaries can be moved to ~/.local/share and symlinked to ~/.local/bin to be included in the users PATH with:

mv bin ~/.local/share/mlir
ln -s $HOME/.local/share/mlir/mlir-opt mlir-opt # + whatever other tools are being used

Testing the installation

And you can test if this has worked with

mlir-opt --help

Cleaning up

Finally, the llvm-project/ directory can be deleted.