Helm charts for GPU metrics

To collect and visualize NVIDIA GPU metrics in a Kubernetes cluster, use the provided Helm chart to deploy DCGM-Exporter.

For full instructions on setting up Prometheus (using kube-prometheus-stack) and Grafana with DCGM-Exporter, review the documentation

Install Helm charts

First, install Helm v3 using the official script:

curl -fsSL -o && \
    chmod 700 && \

Next, setup the Helm repo:

helm repo add gpu-helm-charts \

Update the repo:

helm repo update

Install the official chart for DCGM-Exporter:

helm install \
    --generate-name \

GPU Metrics Dashboard

We provide an official dashboard on Grafana: