About using ncu in win11 to collect profiling results of an AI model


I encountered an issue about using nsight compute to collect profiling results and presented my issue as below.
In win11, I ran a Stable Diffusion project in conda and wanted to get some info of the GPU code. This project used the functions and models in PyTorch.

Is there a way of collecting the GPU metrics by ncu in Conda in Windows? Is there a command line interface available for it as what can be done in Ubuntu?

Thanks for feedback and suggestion(s).




I managed to find the way to collect the GPU metrics by ncu in conda.

In the terminal, in the activated anaconda environment, the ncu command line should be used along with some arguments, including the nsys flags, the path to python, and a python script along with those arguments to be used in the script.

Hi, @xwentian

Thanks for using developer tools and thanks for letting us know your problem has been solved !

This topic was automatically closed after 20 hours. New replies are no longer allowed.