Memory issues on cluster

The job type-specific RAM usage estimates are for what we consider “typical” use cases.
For larger than “typical” cases, assuming the actual availability of required RAM resources:

  • slurm must be configured to allow such jobs
  • a dedicated “large_mem” cluster lane should be added to your cryoSPARC instance (cryosparcm cluster connect) with a suitably multiplied #SBATCH --mem= parameter inside cluster_script.sh, like in this example. Adding a lane instead of replacing the existing lane has the advantage that cryoSPARC jobs with smaller (“typical”) memory requirements won’t have to “wait” for the availability of large memory resources.
1 Like