Skip to content
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

Fixes for cross-platform operability #607

Open
wants to merge 5 commits into
base: main
Choose a base branch
from

Conversation

tensorsofthewall
Copy link

@tensorsofthewall tensorsofthewall commented Jan 17, 2025

Hello all,
I've updated the exo library as part of my attempt to bring Windows Native support. This is NOT the actual update for Windows Native support, but these changes are important as they can help fix issues related to other PRs (#139).

Fixes

1, Updated the mlx.core import that prevents exo from running on Linux/Windows nodes, since mlx isn't available at all for these systems. Linux/Windows nodes will use numpy instead. [Tested]
2. Added AMD GPU querying through pyrsmi (official Python bindings for rocm-smi-lib [AMD supported]). [Untested]
3. Added Windows Device Capabilities, along with the aforementioned AMD GPU querying. exo on Windows now reports the available GPU and VRAM/TFLops correctly. [Tested]
4. Minor fix for type hint in inference-engine.

Please review and suggest changes. Feedback is appreciated.

Thanks!

@tensorsofthewall
Copy link
Author

tensorsofthewall commented Jan 17, 2025

  1. Replaced netifaces (unmaintained, outdated) with scapy.
    (netifaces does not install sometimes in venvs/conda envs, checked on Windows and WSL)

@tensorsofthewall tensorsofthewall changed the title Fixes for cross-platform operability. Fixes for cross-platform operability Jan 17, 2025
@tensorsofthewall
Copy link
Author

  1. Added AMD GPU querying through pyrsmi (official Python bindings for rocm-smi-lib [AMD supported]). [Untested]
  2. Added Windows Device Capabilities, along with the aforementioned AMD GPU querying. exo on Windows now reports the available GPU and VRAM/TFLops correctly. [Tested]

I don't have access to an AMD machine to test GPU querying, so if anyone can assist with this, would be great.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant