[GH-ISSUE #171] [BUG] Shared GPU Job - listed CPU % #106

Closed
opened 2026-05-05 03:25:40 -06:00 by gitea-mirror · 2 comments
Owner

Originally created by @StefanSander3 on GitHub (Jul 24, 2025).
Original GitHub issue: https://github.com/XuehaiPan/nvitop/issues/171

Originally assigned to: @XuehaiPan on GitHub.

Required prerequisites

  • I have read the documentation https://nvitop.readthedocs.io.
  • I have searched the Issue Tracker that this hasn't already been reported. (comment there if it has.)
  • I have tried the latest version of nvitop in a new isolated virtual environment.

What version of nvitop are you using?

1.5.0

Operating system and version

Ubuntu 24.04.02

NVIDIA driver version

550.163.01

NVIDIA-SMI


Python environment

/

Problem description

Working with nvitop-exporter, but the issue I'm facing can also be seen at nvitop level.

Here a job is spread over multiple gpus, the CPU % should be the same for every gpu, right?

Image

Steps to Reproduce

The Python snippets (if any):


Command lines:


Traceback


Logs


Expected behavior

CPU% should be the very same, as it is only one process.

Additional context

If this is not a bug: I'm using exporters data in Grafana.

I should be fine, either way, right?

  • summing all job cpu % / amount of gpus involved
  • or
  • just taking one involved gpu cpu %

Over a few minutes that should be the same pretty much?

Thanks again for this tool!

Originally created by @StefanSander3 on GitHub (Jul 24, 2025). Original GitHub issue: https://github.com/XuehaiPan/nvitop/issues/171 Originally assigned to: @XuehaiPan on GitHub. ### Required prerequisites - [x] I have read the documentation <https://nvitop.readthedocs.io>. - [x] I have searched the [Issue Tracker](https://github.com/XuehaiPan/nvitop/issues) that this hasn't already been reported. (comment there if it has.) - [x] I have tried the latest version of nvitop in a new isolated virtual environment. ### What version of nvitop are you using? 1.5.0 ### Operating system and version Ubuntu 24.04.02 ### NVIDIA driver version 550.163.01 ### NVIDIA-SMI ```text ``` ### Python environment / ### Problem description Working with nvitop-exporter, but the issue I'm facing can also be seen at nvitop level. Here a job is spread over multiple gpus, the CPU % should be the same for every gpu, right? <img width="1302" height="744" alt="Image" src="https://github.com/user-attachments/assets/a995cb98-398a-433a-ad1f-b6e6356a47d5" /> ### Steps to Reproduce The Python snippets (if any): ```python ``` Command lines: ```bash ``` ### Traceback ```pytb ``` ### Logs ```text ``` ### Expected behavior CPU% should be the very same, as it is only one process. ### Additional context If this is not a bug: I'm using exporters data in Grafana. I should be fine, either way, right? - summing all job cpu % / amount of gpus involved - or - just taking one involved gpu cpu % Over a few minutes that should be the same pretty much? Thanks again for this tool!
gitea-mirror 2026-05-05 03:25:40 -06:00
Author
Owner

@XuehaiPan commented on GitHub (Jul 24, 2025):

Hi @StefanSander3, thanks for raising this. I opened a PR to fix this. Could you try it with the following command to validate that fix?

uvx --from git+https://github.com/XuehaiPan/nvitop.git@fix-cpu-snapshot nvitop
# or
pipx run --spec git+https://github.com/XuehaiPan/nvitop.git@fix-cpu-snapshot nvitop
<!-- gh-comment-id:3113755749 --> @XuehaiPan commented on GitHub (Jul 24, 2025): Hi @StefanSander3, thanks for raising this. I opened a PR to fix this. Could you try it with the following command to validate that fix? ```bash uvx --from git+https://github.com/XuehaiPan/nvitop.git@fix-cpu-snapshot nvitop # or pipx run --spec git+https://github.com/XuehaiPan/nvitop.git@fix-cpu-snapshot nvitop ```
Author
Owner

@StefanSander3 commented on GitHub (Jul 24, 2025):

Wow, that's fast. Works for me!

<!-- gh-comment-id:3113772170 --> @StefanSander3 commented on GitHub (Jul 24, 2025): Wow, that's fast. Works for me!
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference: github-starred/nvitop#106
No description provided.