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.
Multiple Vendors use MLIR as the middle layer, mapping from platform frameworks like PyTorch, JAX, and TensorFlow into MLIR and then progressively lowering down to their target hardware. We have seen half a dozen custom lowerings from PyTorch to MLIR. Having canonical lowerings 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 yet another PyTorch frontend for MLIR. The goal is to be similar to current hardware vendors adding LLVM target support instead of each one also implementing Clang / a C++ frontend.
This is the most tested path down to Torch MLIR Dialect, and the PyTorch ecosystem is converging on using TorchScript IR as a lingua franca.
- LazyTensorCore (Based on the PyTorch [`lazy_tensor_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.
Two setups are possible to build: in-tree and out-of-tree. The in-tree setup is the most straightforward, as it will build LLVM dependencies as well.
### Building torch-mlir in-tree
The following command generates configuration files to build the project *in-tree*, that is, using llvm/llvm-project as the main build. This will build LLVM as well as torch-mlir and its subprojects.
If you have built llvm-project separately in the directory `$LLVM_INSTALL_DIR`, you can also build the project *out-of-tree* using the following command as template:
```shell
cmake -GNinja -Bbuild \
-DCMAKE_C_COMPILER=clang \
-DCMAKE_CXX_COMPILER=clang++ \
-DPython3_FIND_VIRTUALENV=ONLY \
-DMLIR_DIR="$LLVM_INSTALL_DIR/lib/cmake/mlir/" \
-DLLVM_DIR="$LLVM_INSTALL_DIR/lib/cmake/llvm/" \
-DMLIR_ENABLE_BINDINGS_PYTHON=ON \
-DLLVM_TARGETS_TO_BUILD=host \
.
```
The same QoL CMake flags can be used to enable ccache and lld. Be sure to have built LLVM with `-DLLVM_ENABLE_PROJECTS=mlir`.
Be aware that the installed version of LLVM needs in general to match the committed version in `externals/llvm-project`. Using a different version may or may not work.
[Example IR](https://gist.github.com/silvasean/e74780f8a8a449339aac05c51e8b0caa) for a simple 1 layer MLP to show the compilation steps from TorchScript.
