gputest.yaml

apiVersion: apps/v1
kind: Deployment 
metadata:
  name: gpu-pod
  namespace: default
  labels:
    app: powerai-x86
spec:
  replicas: 1
  selector:
    matchLabels:
      app: powerai-x86
  template:
    metadata:
      labels:
        app: powerai-x86
    spec:
      containers:
      - name: cuda-container
        #image: ibmcom/powerai:1.6.1-base-ubuntu18.04-py3-x86_64 
        image: tensorflow/tensorflow:latest-gpu-jupyter 
        env:
        - name: LICENSE
          value: 'yes'
        resources:
          limits:
            nvidia.com/gpu: 1
          #livenessProbe:
          #exec:
          #  command:
          #  - bash
          #  - "-c"
          #  - nvidia-smi
          #initialDelaySeconds: 30
          #timeoutSeconds: 1
          #periodSeconds: 60
          #successThreshold: 1
          #failureThreshold: 3
          #readinessProbe:
          #exec:
          #  command:
          #  - bash
          #  - "-c"
          #  - nvidia-smi
          #initialDelaySeconds: 30
          #timeoutSeconds: 1
          #periodSeconds: 60
          #successThreshold: 1
          #failureThreshold: 3
      nodeSelector: 
        workload: mldl