Open
Description
Description
This project has a lot of GPU-related issues and multiple different GPU-acceleration variants.
To help organize this work, tonight I added 3 new labels, and tried to add them to all of the relevant issues in this project.
gpu (CUDA)
- for the CUDA version, supporting NVIDIA GPUs (
-DUSE_CUDA=ON
) - 32 open, 51 closed (link)
- for the CUDA version, supporting NVIDIA GPUs (
gpu (OpenCL)
- for the OpenCL version, supporting a wide range of GPUs (
-DUSE_GPU=ON
) - 22 open, 93 closed (link)
- for the OpenCL version, supporting a wide range of GPUs (
gpu (ROCm)
- for the soon-to-be-added ROCm version, supporting AMD GPUs (
-DUSE_ROCM=ON
) - coming in [ROCm] add support for ROCm/HIP device #6086
- 0 open, 0 closed (link)
- for the soon-to-be-added ROCm version, supporting AMD GPUs (
Benefits of this work
- helps contributors to find GPU-related issues to help with
- helps to separate the different GPU variants (e.g. some contributors may want to work only on the CUDA variant, not OpenCL)
Notes
Opening this to document it for users, contributors, and other maintainers (especially those who can't access the private maintainer chat).
I'll mark this awaiting response
so it's auto-closed after 30 days.