Partitions

The Slurm partitions are divisions of compute nodes, sorted mainly by purpose. You can check the available partitions by running the Slurm command :

scontrol show part

This will give you all the data you need to analyze which partition is best suited for your job (especially the priority tied to the partition). The partition can be specified in your Slurm script with --partition=xxxxx. There are mainly 3 partitions that you may want to use :

standard: Standard CPU compute nodes. If your jobs goes OOM or you need more RAM, ask for more CPUs, as the default RAM/CPU core default might not be suited for you.
visu: The visu partition have graphics cards used to visualize the results on XCS.
gpu: The gpus partition is the same as the visu partition, except the cards aren’t made to visualise the results, but rather run GPU-intensive calculations.
bigmem: The nodes are the same as standard, but with more RAM