Alauda Build of Hami

Allocate device core resource

TOC

Allocate device core to container

Allocate device core to container

To allocate a portion of device core resources, only specify nvidia.com/gpucores; no other resource fields are needed.

apiVersion: v1
kind: Pod
metadata:
  name: gpu-pod
spec:
  containers:
    - name: ubuntu-container
      image: ubuntu:18.04
      command: ["bash", "-c", "sleep 86400"]
      resources:
        limits:
          nvidia.com/gpu: 2 # requesting 2 vGPUs
          nvidia.com/gpucores: 50 # requesting 50% of each vGPU's core resources

NOTICE: HAMi implements nvidia.com/gpucores using time-slicing, so core utilization reported by nvidia-smi may fluctuate.