C++/CUDA/SYCL Extensions in PyTorch

An example of writing a C++/CUDA/Sycl extension for PyTorch. See here for the accompanying tutorial. This repo demonstrates how to write an example extension_cpp.ops.mymuladd custom op that has both custom CPU and CUDA/Sycl kernels.

Note: SYCL serves as the backend programming language for Intel GPUs (device label xpu). For configuration details, see: Getting Started on Intel GPUs.

The examples in this repo work with PyTorch 2.4 or later for C++/CUDA & PyTorch 2.8 or later for Sycl.

To build:

pip install --no-build-isolation -e .

To test:

python test/test_extension.py

Authors

Peter Goldsborough, Richard Zou