17 lines (15 loc) · 500 Bytes

Kubernetes GPU Workloads & Scheduling

Architecture

Terraform Node Group: Provisioned with g4dn.xlarge instances with auto-scaling limits (0-3).
NVIDIA Device Plugin: Deployed as a DaemonSet to expose GPU hardware metrics and limits to Kubernetes pods.

Resource Allocation & Tolerations

Pod specifications requesting GPU must declare:

resources:
  limits:
    nvidia.com/gpu: 1
tolerations:
  - key: "nvidia.com/gpu"
    operator: "Exists"
    effect: "NoSchedule"