ConceptPlatform engineers

What the GPU Operator does

The GPU Operator automates drivers, the container runtime, the device plugin, and telemetry on Kubernetes.

Rather than hand-installing drivers on every node, the GPU Operator manages the whole GPU software stack as Kubernetes resources: the driver, the NVIDIA container toolkit, the device plugin that advertises nvidia.com/gpu, node feature discovery, and DCGM telemetry.

It rebuilds drivers against the running kernel, which is why kernel upgrades must stay within a validated baseline. AMD provides an equivalent AMD GPU Operator for ROCm.

See the Kubernetes section for per-distribution enablement, and Day-Two Ops for the upgrade runbook.

gpu-operatorkubernetesdrivers
Related
Guide

Mixing GPU types and vendors in one cluster