Forum Discussion

Jasonkeck-GDIT's avatar
2 months ago
Solved

Monitoring GPU's

Good morning ScienceLogic Community, with the rapid adoption of Artificial Intelligence I am seeing a lot of GPU's being deployed and since these GPU's are not cheap there are lots of people asking if they can be monitored. Anyone monitoring GPU's already?

Thanks. 

Jason. 

  • We have an NVIDIA PowerPack already released and we also have an AMD PowerPack coming in the next week or so. 

    We have released a vLLM powerpack for users of those model libraries for specific AI workflows. Definitely share if you have some other use cases to look at. 

4 Replies

  • We have an NVIDIA PowerPack already released and we also have an AMD PowerPack coming in the next week or so. 

    We have released a vLLM powerpack for users of those model libraries for specific AI workflows. Definitely share if you have some other use cases to look at. 

  • The AMD GPU monitoring powerpack was released yesterday and we'd love feedback on it, NVIDIA, and the vLLM integration. 

  • TaylorJohnson​ I downloaded installed and have the NVIDIA GPU's monitored fantastically on my server that hosts my pool of GPU, however when I try to monitor a virtual server which I have GPU resources mapped to I am unable to get anything monitored in ScienceLogic, I know that not everything will come through on a virtual GPU especially things like Serial numbers and hardware side info but I was hoping to get some utilization stats in order to see which of my virtual servers may be running hot,, its not easy to map them to the main host and make it hard for end users to understand. 

  • Taylor, right now I only have direct access to NVIDIA GPU's I will go on the hunt for some AMD GPU's but no promises.