Skip to content

Replace NVML device affinity with XPUML#3

Merged
erichudev merged 2 commits into
kunlunfrom
feat/xpuml-device-affinity
Jul 13, 2026
Merged

Replace NVML device affinity with XPUML#3
erichudev merged 2 commits into
kunlunfrom
feat/xpuml-device-affinity

Conversation

@chaonanD

Copy link
Copy Markdown
Collaborator

Summary

  • replace Device handle and name queries with XPUML bindings exposed by torch_xmlir._XMLIRC
  • replace NVML CPU affinity lookup with per-device CPU core allocation
  • update distributed initialization to use the XPUML-backed Device affinity

Validation

  • python -m py_compile cosmos_framework/utils/device.py cosmos_framework/utils/distributed.py

Notes

  • this PR targets kunlun and remains open for peer review
  • follow-up commits can be pushed to feat/xpuml-device-affinity

chaonanD added 2 commits July 10, 2026 12:07
Replace NVML calls in conftest, device utils, inference args, and the
device monitor callback with torch_xmlir XPUML bindings.
@erichudev
erichudev merged commit 4b037c5 into kunlun Jul 13, 2026
1 check passed
@chaonanD
chaonanD deleted the feat/xpuml-device-affinity branch July 16, 2026 07:48
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.

2 participants