2 docs tagged with "scheduling"

The HAMi DRA driver is young and moving fast. This lab installs the exact DaemonSet manifests that were verified live on a Tesla T4 cluster (driver projecthami/k8s-dra-driver:v0.1.0). The driver repository has since added a Helm chart for the same v0.1.0 driver (in-repo at chart/hami-dra-driver, with a nvidiaDriverRoot value covering GPU Operator clusters); this lab will switch to the chart once that path has been verified. The consumable capacity feature also remains behind a Kubernetes feature gate.

This lab uses NVIDIA's nvml‑mock library to simulate a high‑end GPU node — 8 fake A100 GPUs — inside a local kind cluster. You will build HAMi directly from the main branch, then verify GPU scheduling features: sharing, memory/core limits, percentage-based memory requests, and multi‑GPU allocation — all without physical hardware.