-
-
Notifications
You must be signed in to change notification settings - Fork 606
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Intel Arc GPU support. #2345
Comments
The GPU support in general is handled through other libraries. The Arc GPU is handled through https://github.com/JuliaGPU/oneAPI.jl which can be used similarly to CUDA interface with some changes. Looking at the documentation it seems that the current version does not support the full set of operations that Flux needs to run. |
I'm also very interested to play around with Flux on my Intel Arc GPU. However, despite oneAPI.jl existing and looking quite mature, it is currently not among the supported GPU backends for Flux. Are there any concrete plans to add support in the future? |
Given that none of the deep learning specific oneAPI libraries are wrapped yet in oneAPI.jl, I would not call it quite mature for ML use cases. Were someone to do that, we could then think about writing extensions for it in Flux and NNlib like was done for AMDGPU.jl and Metal.jl. |
I am just learning about these things, but are you sure that oneAPI would be the best solution here? There is also OpenVINO (Quick Start Guide) which seems to have APIs from other programming languages. |
My understanding is that OpenVINO requires a very declarative setup-and-run API. It seems more designed for inference of already trained models instead of the define-by-run workflow most modern DL libraries use. |
Does Flux.jl support Intel Arc GPU?
The text was updated successfully, but these errors were encountered: