Partitions

Partition == Queue

SLURM allows the definition of different partitions. A partition is a subset of all the nodes in the cluster. Partitions are not mutually exclusive, so a node can belong to multiple partitions. Every compute node is assigned to one or more partitions.

Our cluster has 4 different partitions:

• standard
  • The default queue. Contains the 31 standard compute nodes (at least 128GB RAM, at least 16 cores). No time limit.
• gpu
  • Contains the GPU machines only. No time limit. node91 has two (old) Nvidia K20m GPUs. node94 has an Nvidia V100 with 32GB of RAM.
• bigmem
  • Contains big memory machines only (512GB or 768GB, 32 cores). No time limit.
• short
  • Contains all nodes, including gpu and bigmem, but has a 5 hour time limit, and jobs in this queue have a lower priority than jobs in the other queues. The short queue allows users with quick-running jobs to make use of all the cluster nodes without holding up users who need the special purpose (bigmem or gpu) nodes too much.

Standard, gpu, and bigmem have no time limit.