Run a Sample Workload#

After completing the deployment steps, verify your installation by running a basic single-GPU sample workload inside a Confidential Container.

This page intentionally uses the simplest possible manifest so that you can confirm the deployment end-to-end. For the full set of workload configuration options, including runtime class selection, resource type naming, and multi-GPU passthrough, refer to Configuring Confidential Container Workloads.

  1. Create a file named cuda-vectoradd-kata.yaml with the following sample manifest:

    apiVersion: v1
kind: Pod
metadata:
  name: cuda-vectoradd-kata
  namespace: default
spec:
  runtimeClassName: kata-qemu-nvidia-gpu-snp # or kata-qemu-nvidia-gpu-tdx
  restartPolicy: Never
  containers:
    - name: cuda-vectoradd
      image: "nvcr.io/nvidia/k8s/cuda-sample:vectoradd-cuda12.5.0-ubuntu22.04"
      resources:
        limits:
          nvidia.com/pgpu: "1"
          memory: 16Gi

    Before applying the manifest, adjust the two highlighted lines for your environment:

    • Runtime class. Use kata-qemu-nvidia-gpu-snp on AMD SEV-SNP nodes or kata-qemu-nvidia-gpu-tdx on Intel TDX nodes.

    • GPU resource type. The sample requests nvidia.com/pgpu, which is the default resource name advertised by the NVIDIA Kata sandbox device plugin. If your cluster was installed with the P_GPU_ALIAS="" setting, replace it with the model-specific name advertised on your node, for example nvidia.com/GH100_H200_141GB.

    Refer to Configuring Confidential Container Workloads for guidance on each option.

  2. Create the pod:

    $ kubectl apply -f cuda-vectoradd-kata.yaml

    Example Output:

    pod/cuda-vectoradd-kata created

  3. Optional: Verify the pod is running:

    $ kubectl get pod cuda-vectoradd-kata

    Example Output:

    NAME                  READY   STATUS    RESTARTS   AGE
cuda-vectoradd-kata   1/1     Running   0          10s

  4. View the logs from the pod after the container starts:

    $ kubectl logs -n default cuda-vectoradd-kata

    Example Output:

    [Vector addition of 50000 elements]
Copy input data from the host memory to the CUDA device
CUDA kernel launch with 196 blocks of 256 threads
Copy output data from the CUDA device to the host memory
Test PASSED
Done

  5. Delete the pod:

    $ kubectl delete -f cuda-vectoradd-kata.yaml

Next Steps#

On this page