You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Nvidia's push to ARM64 architecture (e.g. GH200, NVL32/NV72) is driving a lot more projects to support this architecture. While downstream engines such as VLLM support aarch64, it would be more difficult to support official container distributions if they need to compile dependencies such as bitsandbytes from source for each build (vllm-project/vllm#10499).
Your contribution
I believe this would need to be done directly by the maintainers.
The text was updated successfully, but these errors were encountered:
Hi @drikster80! Thank you for the suggestion. Distributing the package for aarch64 is something we'd certainly want to do in the future. We're a bit bandwidth constrained at the moment so I cannot promise an ETA, but I'm making sure to put this on our backlog.
Feature request
Requesting to distribute a Pypi whl for aarch64.
Motivation
Nvidia's push to ARM64 architecture (e.g. GH200, NVL32/NV72) is driving a lot more projects to support this architecture. While downstream engines such as VLLM support aarch64, it would be more difficult to support official container distributions if they need to compile dependencies such as bitsandbytes from source for each build (vllm-project/vllm#10499).
Your contribution
I believe this would need to be done directly by the maintainers.
The text was updated successfully, but these errors were encountered: