GPUs showing as 0 in the GPU operator

[vbudi000]

Just installed the GPU operator dashboard per the instruction in https://rh-ecosystem-edge.github.io/console-plugin-nvidia-gpu/
Openshift 4.16
GPUs: 8x2 H100 80GB GPUs

Console is showing the GPU plugin being loaded

the label on nvidia-gpu-operator has openshift.io/cluster-monitoring=true

The servicemonitor is showing nvidia-dcgm-exporter and the in the dcgm pod - I can see the metrics DCGM_FI_*

But the Compute > GPU showing that no GPU are found

[duritong]

I am running into the same issue and with H100 or H200 GPUs there are no metrics for DCGM_FI_PROF_GR_ENGINE_ACTIVE which is used to query all active GPUs: https://github.com/rh-ecosystem-edge/console-plugin-nvidia-gpu/blob/main/src/hooks/use-gpus-info.ts#L23