You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is cool idea, right now in the screenshot I only see CPU but not GPU. Is there any specific reason for that. Because more and more LLMs are being used on consumer machines GPU usage and even power consumption makes important metric to show.
Your thoughts?
The text was updated successfully, but these errors were encountered:
This is a good point. The simple answer is that GPU information needs to be handled differently based on the manufacturer (NVIDIA, AMD, Intel, etc) since each has their own proprietary APIs and monitoring tools:
NVIDIA GPUs use NVML (NVIDIA Management Library)
AMD GPUs use ROCm/HIP APIs
Intel GPUs use oneAPI Level Zero
This makes it more complex than CPU monitoring which has standardised interfaces. The next step would be implementing support for the most common GPU vendors, starting with NVIDIA, followed by AMD and Intel.
This is cool idea, right now in the screenshot I only see CPU but not GPU. Is there any specific reason for that. Because more and more LLMs are being used on consumer machines GPU usage and even power consumption makes important metric to show.
Your thoughts?
The text was updated successfully, but these errors were encountered: