README.md.j2

# {{ image.pretty_name }} {{ image.version }} Container Image
{% include 'badges.j2' %}

## Description

The {{ image.pretty_name }} container image provides NVIDIA GPU drivers in a
containerized environment. This image is based on
{{ image.os_version.pretty_distribution_base_name }} and includes both the
open-source and proprietary NVIDIA kernel modules {{ image.version }}, along with the necessary
user-space tools.

It is providing [precompiled binaries](https://docs.nvidia.com/datacenter/cloud-native/gpu-operator/latest/precompiled-drivers.html) for the NVIDIA GPU Operator or for manual
driver deployment on container hosts, allowing for faster deployment.

## Usage

To deploy this container image, use the NVIDIA GPU Operator Helm chart version 26.3.1 or later, following [these steps](https://docs.nvidia.com/datacenter/cloud-native/gpu-operator/latest/getting-started.html#procedure).

Add the arguments `--set driver.repository=registry.suse.com/third-party/nvidia --set driver.usePrecompiled=true` and `--set driver.version=<driver-branch>` to the `helm install` command. `<driver-branch>` is the major version of the GPU driver (such as `595`, `590` or `580`).

For k3s or RKE2 cl, add `--set toolkit.env[0].name=CONTAINERD_SOCKET --set toolkit.env[0].value=/run/k3s/containerd/containerd.sock` when [Node Resource Interface (NRI) plugin](https://docs.nvidia.com/datacenter/cloud-native/gpu-operator/latest/cdi.html#nri-plugin) is not enabled.

As an example:
```ShellSession
helm install --wait gpu-operator \
     -n gpu-operator --create-namespace \
     nvidia/gpu-operator \
     --version=v26.3.1 \
     --set driver.repository=registry.suse.com/third-party/nvidia \
     --set driver.usePrecompiled=true \
     --set driver.version="<driver-branch> \
     --set toolkit.env[0].name=CONTAINERD_SOCKET \
     --set toolkit.env[0].value=/run/k3s/containerd/containerd.sock"
```

{% include 'licensing_and_eula.j2' %}