intel / intel/xpumanager

xpu-smi dump stats shows incorrect GPU utilization

Open
#76 5 comments 1 reaction 0 assignees View on GitHub
Dominant language
C++
Stars
194
Forks
34
PR merge metrics
No merged PRs in 30d

Description

I am running a workload on Intel Max Series 1550. I have `xpu-smi` installed. I can see the model is training on the GPU but `xpu-smi dump -d 0 -m 0` shows `N/A`

![image](https://github.com/intel/xpumanager/assets/41214682/144576b4-063d-41c1-b30a-dd6b57579f17)

But i can see percentage GPU memory utilization `xpu-smi dump -d 0 -m 5`

![image](https://github.com/intel/xpumanager/assets/41214682/0f56b06b-6516-48fc-b97c-6f60531a79ee)

Here is the version details of xpu-smi on my machine
![image](https://github.com/intel/xpumanager/assets/41214682/02f572bd-1666-4d6b-bd68-d5292883a75d)

Contributor guide

Open the contributing guide

Research direction

Reproduce the report on Intel Max Series hardware with `xpu-smi dump -d 0 -m 0` while a GPU workload is running, then compare it with `xpu-smi dump -d 0 -m 5`. Trace the implementation of these dump metrics and verify that GPU utilization reports a value rather than `N/A`; the issue does not name source files or tests.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
cli, observability
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.