NVIDIA GPU Operator V2 and Out-of-Tree Operator

by mresvanis
GNU/Linux ◆ xterm-256color ◆ bash 79 views

This demo showcases the NVIDIA GPU Operator V2 alongside the Out-of-Tree Operator, deployed in a Kubernetes cluster hosted in AWS with 2 g4dn.xlarge worker nodes.

The driver and device plugin containers are managed by the Out-of-Tree Operator, while the NVIDIA GPU components, i.e. NVIDIA Container Toolkit, NVIDIA GPU Prometheus Exporter, DCGM and DCGM Prometheus exporter by the NVIDIA GPU Operator V2.

Different NVIDIA driver versions are used among the GPU-powered nodes. One of the nodes is provisioned with the NVIDIA driver version 470.103.01, while the other is provisioned with 510.47.0.

Resources: