Use the NVIDIA GPU Operator
You can use the NVIDIA GPU Operator to provision worker nodes for running GPU-accelerated virtual machines (VMs) in OpenShift Virtualization.
The NVIDIA GPU Operator manages NVIDIA GPU resources in an OpenShift Container Platform cluster and automates tasks when preparing nodes for GPU workloads. The NVIDIA GPU Operator can also facilitate provisioning complex artificial intelligence and machine learning (AI/ML) workloads.
|
|
You can get support for the NVIDIA GPU Operator through NVIDIA. For more information, see "Obtaining Support from NVIDIA" in the Red Hat Knowledgebase. |
-
Configure your
ClusterPolicymanifest. YourClusterPolicymanifest must match the provided example:apiVersion: nvidia.com/v1 kind: ClusterPolicy metadata: name: gpu-cluster-policy spec: daemonsets: updateStrategy: RollingUpdate dcgm: enabled: true dcgmExporter: {} devicePlugin: {} driver: enabled: false kernelModuleType: auto gfd: {} mig: strategy: single migManager: enabled: true nodeStatusExporter: enabled: true operator: defaultRuntime: crio initContainer: {} runtimeClass: nvidia use_ocp_driver_toolkit: true sandboxDevicePlugin: enabled: true sandboxWorkloads: defaultWorkload: vm-vgpu enabled: true toolkit: enabled: true installDir: /usr/local/nvidia validator: plugin: env: - name: WITH_WORKLOAD value: "true" vfioManager: enabled: true vgpuDeviceManager: config: default: default name: vgpu-devices-config enabled: true vgpuManager: enabled: true image: <vgpu_image_name> repository: <vgpu_container_registry> version: <nvidia_vgpu_manager_version>where:
<vgpu_image_name>-
Specifies the vGPU image name.
<vgpu_container_registry>-
Specifies the vGPU container registry value.
<nvidia_vgpu_manager_version>-
Specifies the version of the vGPU driver you have downloaded from the NVIDIA website and used to build the image.
-
Use the NVIDIA GPU Operator to configure mediated devices. For more information see NVIDIA GPU Operator with OpenShift Virtualization.