GPU

This feature is available for Business Edition only.

The GPU view is available for environments where GPU nodes are detected. It displays three tables: GPU, GPU Nodes, and GPU Workloads.

GPU

The GPU table shows the readiness status of each node (Ready / Degraded / Not Available), and a breakdown of GPU capacity, allocatable, allocated, and available counts per nvidia.com/* resource type. The total GPU node count and degraded node count are shown at the bottom, along with GPU Operator detected and Device plugin detected badges where applicable.

GPU Nodes

The GPU Nodes table lists each node by name with its GPU capacity, allocatable GPU count, allocated GPU count, and a status badge. If the status is not Ready, a reason is displayed alongside the badge.

GPU Workloads

The GPU Workloads table lists pods with GPU requests, showing the namespace, pod name, owner workload, GPU requested, scheduled node, workload status, and any scheduling issues.