Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Questions about gpustat display memory usage logic

See original GitHub issue

Hi, I was writing some GPU memory usage monitoring codes using nvmlDeviceGetComputeRunningProcesses. And I used gpustat for debugging. I found that gpustat reported about 5x memory usage than my program.

Did I use nvmlDeviceGetComputeRunningProcesses in the wrong way? How is the display logic of the yellow number below? (I tried to look at the source codes but did not find out the logic)

gpustat

My codes

Issue Analytics

State:
Created 3 years ago
Comments:7 (3 by maintainers)

Top GitHub Comments

1reaction

wookayincommented, Jul 9, 2020

No worries. I think your program and gpustat are using the same API so you should be able to get the exactly same result. Please let us know if there’s anything we could help with.

0reactions

ganlercommented, Jul 9, 2020

FYI, we basically use the same API (nvmlDeviceGetComputeRunningProcesses and usedGpuMemory).

https://github.com/wookayin/gpustat/blob/v0.6.0/gpustat/core.py#L305 https://github.com/wookayin/gpustat/blob/v0.6.0/gpustat/core.py#L346

Hi @wookayin , I think this is my problem. I meant to monitor my C++ program’s GPU memory usage via nvmlDeviceGetComputeRunningProcesses but could not get the result as gpustat and nvidia-smi. It seems to be my problem and I should try to debug my program.

Thank you for your patience. Closed.