Scheduling
Scheduling
- k8s-gpu-scheduler-lab
partial
A controlled comparison of Kubernetes GPU schedulers on identical traces, on a real control plane with simulated GPU nodes. K0 and four degenerate baselines are built; Kueue, Volcano and KAI are planned.
- Slurm Scheduler Lab
built
A discrete-event model of Slurm's multifactor priority, Fair Tree fairshare, backfill and preemption, for testing scheduling policy against a trace before it reaches a live controller.
- Slurm Backfill vs. Priority Weights on a Synthetic Workload: What One Seed Showed, and What Twenty Did Not
Corrected September 2026. In my simulator backfill has a large effect on every seed. The original claim that priority weights barely matter came from one seed, and the time-limit claim was never measured.