SLURM

less than 1 minute read

Published:

SLURM is one of the most common job scheduler used in many high performance cluster computing severs (HPC). Here, I summarize useful SLURM commands.

  • sbatch -C "CPU_GEN:HSW|CPU_GEN:BDW|CPU_GEN:SKX":
    • AVX2 support started with Haswell on Intel CPUs. To target any CPU generation after Haswell, we can specify constraits assuming that the cluster admin configured those properties in each node.
  • scontrol hold <jid>
  • scontrol release <jid>
  • scontrol update job <jid> MinMemoryCPU=8000
  • scontrol update job <jid> Dependency=afterany:<jid>
  • scontrol update job <jid> JobName="newJobName"
  • scontrol update job <jid> partition=<job partitions>
  • scontrol show job <jid>
  • sacct --starttime 2020-08-01 --

Reference