Submitting jobs

Whatever you read here may need to be adjusted to fit your specific case. Do not hesitate to ask for some help wen needed.

Filesystems

Some but not all partitions are available to the compute nodes. Compute nodes will not be able to access any data from filesystems that are not listed here. /work /scratch /home/

Slurm partitions

There are currently 2 partitions, normal and bigmem. The normal partition is the default partition if you submit a job without precising witch partition should be used you job will be placed in one of the normal partition. The normal partition has limited RAM of 250GB, in case you need more than that please use the bigmem partition. -p option can be used to specify the needed partition.

Example of batch script

   1 #!/bin/bash
   2 
   3 #SBATCH -J test               # Job name
   4 #SBATCH -o /work/<<UID>>/job.%j.out   # Name of stdout output file (%j expands to jobId)
   5 #SBATCH -N 1                  # Total number of nodes requested
   6 #SBATCH -n 16                 # Total number of mpi tasks requested
   7 #SBATCH -t 01:30:00           # Run time (hh:mm:ss) - 1.5 hours
   8 
   9 # Load your software/command
  10 module load CMD
  11 
  12 # Run your command
  13 CMD [OPTIONS] ARGUMENTS