MatrixAI / Overwatch

Distributed Infrastructure Telemetry
2 stars 0 forks source link

GPU Monitoring #18

Open CMCDragonkai opened 5 years ago

CMCDragonkai commented 5 years ago

For GPUs, we have to consider things like nvidia-smi and NVML C based library.

There are NVML python based bindings.

Tools should use NVML or Python Bindings, not nvidia-smi as it is not stable.