Currently we manually build because building with CUDA/ROCM/OpenCL support doesn't work without custom CI workers. Turn this into an automated flow, rather than manually building. need runners for win/osx/linux https://github.com/actions/runner. will accept volunteer runners. maybe use beta ? github/roadmap#505
Currently we manually build because building with CUDA/ROCM/OpenCL support doesn't work without custom CI workers. Turn this into an automated flow, rather than manually building. need runners for win/osx/linux https://github.com/actions/runner. will accept volunteer runners. maybe use beta ? github/roadmap#505