deploy_gpu.md

Create Intel GPU Device Plugin CR

Create CR via web console

1. Go to Operator -> Installed Operators.
2. Open Intel Device Plugins Operator.
3. Navigate to tab Intel GPU Device Plugin.
4. Click Create GpuDevicePlugin -> set correct parameters -> Click Create.
5. Optional: If you want to make any customizations, select YAML view and edit the details. Once you are done, click Create.

Verify via web console

1. Verify CR by checking the status of Workloads -> DaemonSet -> intel-gpu-plugin.
2. Now GpuDevicePlugin is created.

Create CR via CLI

Apply the CR yaml file:

$ oc apply -f  https://raw.githubusercontent.com/intel/intel-technology-enabling-for-openshift/main/device_plugins/gpu_device_plugin.yaml

Verify via CLI

Verify that the device plugin CR is ready:

$ oc get GpuDevicePlugin 

Output:

NAME		        DESIRED	  READY	  NODE SELECTOR	                                    AGE
gpudeviceplugin-sample  1         1       {"intel.feature.node.kubernetes.io/gpu":"true"}   3m12s

Using Intel Data Center GPU resource exclusively

In this release, we only verified and support the single Intel GPU i915 resource dedicated to the single workload pod. To achieve this, we set sharedDevNum: 1 and preferredAllocationPolicy: none as default options.
As the cluster administrator, use the gpu_device_plugin.yaml provided from the previous section Create CR via CLI or use the default options from Create CR via web Console.
As the application owner, when claiming the i915 resource, make sure the resource limits and requests are set as shown below:

spec:
  containers:
    - name: gpu-pod
      resources:
        limits:
          gpu.intel.com/i915: 1
        requests:
          gpu.intel.com/i915: 1

For more details, please refer to this issue.