k8s device plugin
FIPS 140-3Machine learning & AINVIDIA device plugin for Kubernetes that exposes GPUs to the Kubernetes scheduler, provides GPU health monitoring, and supports MIG, MPS, time-slicing, and CDI.
NVIDIA device plugin for Kubernetes that exposes GPUs to the Kubernetes scheduler, provides GPU health monitoring, and supports MIG, MPS, time-slicing, and CDI.
The NVIDIA k8s-device-plugin exposes NVIDIA GPUs to Kubernetes as attachable resources. It runs as a DaemonSet and
registers devices with the kubelet so that GPU-accelerated workloads can request GPU resources using the
nvidia.com/gpu resource name. Typical use cases include ML training and inference, HPC, and other GPU-accelerated
workloads that rely on CUDA or other NVIDIA libraries.
k8s-device-plugin supports features such as GPU time-slicing, Multi-Instance GPU (MIG), Multi-Process Service (MPS), GPU Feature Discovery (GFD), and the Container Device Interface (CDI), enabling flexible GPU partitioning and improved utilization in shared cluster environments.