Cluster usage

Jump to bottom

August edited this page Dec 8, 2021 · 4 revisions

Useful resources for using the OHSU Exacloud cluster

OHSU Advanced Computing Center (ACC): https://www.ohsu.edu/advanced-computing-center/acc-and-exacloud-cluster

Tutorials

An exceptionally good tutorial is available at https://laderast.github.io/exacloud_tutorial/

Another good tutorial is available at http://docplayer.net/152843431-The-command-line-without-tears-exacloud-for-big-or-medium-tasks.html that has a lot of information on things like setting up your bash environment

Job Scheduler

The exacloud cluster uses SLURM to run jobs on the nodes. SLURM is pretty widely supported, and lots of Googleable information on it.

The first tutorial above describes how to interact with it. Any program or code you want to run on the cluster needs to go through SLURM.

Partitions

exacloud: Main partition. Many nodes, with anywhere from 28 to 44 CPUs/node. Generally not a long wait time.
gpu: Partition with fewer nodes, but they have GPUs. Depending on how many GPUs you request, the wait time can stack up on this partition.