You can run jobs at different sizes and durations on Bridges-2 RM. The following lists the different queues that you can submit to, describing how many nodes you get, how long you can run, the type of resources you get, and the average wait time.
Jobs are submitted through Slurm. Unless you set otherwise, a job runs on one node with a one-hour walltime. Pick a queue by how much of a node you need:
- RM: whole 256 GB nodes. Request the number of nodes with -N; each node gives you all 128 cores. Up to 64 nodes per job, 72-hour maximum walltime.
- RM-shared: part of a single 256 GB node, shared with other jobs. Request up to 64 cores with --ntasks-per-node, and you are charged only for the cores you request. Each core comes with about 2 GB of memory, and a job cannot span nodes. Use RM-shared when you do not need a full node; it usually shortens your wait.
- RM-512: like RM, but on the 15 larger 512 GB nodes for memory-intensive work. These are always allocated whole, up to 2 nodes per job, 72-hour maximum walltime.
- RM-small: for interactive and small or quick jobs. Check its current limits with sinfo.
For submission commands, sample scripts, and node-sharing details, see the Bridges-2 Batch Jobs guide. Partition-specific information is on the Bridges-2 Regular Memory section of the user guide.