Partitions
On SLURM, queues where you have to submit your jobs are called partitions
On curie, there are the following partitions :
- CLUSTER : maxtime 7 days, shared nodes
- xeonv1 : maxtime 14 days, exclusive nodes
- xeonv2 : maxtime 14 days, exclusive nodes
- xeonv3 : maxtime 14 days, exclusive nodes
- xeonv4 : maxtime 14 days, exclusive nodes
- xeonv5 : maxtime 14 days, exclusive nodes
- xeonv6 : maxtime 14 days, exclusive node
- xeonv3_mono : nomaxtime, shared nodes, 15000 MB per cpu
- xeonv4_mono : nomaxtime, shared nodes, 30000 MB per cpu
- xeonv5_mono : nomaxtime, shared nodes, 40000 MB per cpu
- q-chem : maxtime 14 days, shared node, 8000 MB per cpu
- epycv1 : maxtime 14 days, exclusive node
- napab : nomaxtime, shared nodes, 62000 MB per cpu
- actipnmr : nomaxtime, shared nodes, 5200 MB per cpu
- molqed : nomaxtime, shared nodes, 5200 MB par cpu
- zen2 : nomaxtime, shared nodes, 5200 MB par cpu
CLUSTER is the default partition if you don’t specify one in your script.
CLUSTER and xeonv1 shares the same nodes.
napab partition is dedicated to Nadia/Sophie/Fabienne crew, ask her before submit.
If you want to use a specific partition, example with xeonv3 :
#SBATCH -p xeonv3Partitions hardware specs are available on the Curie/About page.