Available Partitions
| Name | Nodecount | Max walltime | Nodes/Threads per job (min-max) | Other rules | Purpose |
|---|---|---|---|---|---|
debug | 5 | 30 Minutes | 1-3 nodes | For debugging job scripts | |
t1small | 98 | 2 days | 1-2 nodes | For short, small jobs with quick turnover | |
t1standard | 98 | 4 days | 3-24 nodes | Default | General-purpose partition |
t2small | 98 | 4 days | 1-2 nodes | Tier 2 users only. Increased priority and walltime. | Tier 2 version of t1small |
t2standard | 98 | 7 days | 3-48 nodes | Tier 2 users only. Increased priority and walltime. | Tier 2 general-purpose partition |
gpu | 2 | 2 days | 1-2 nodes | For jobs using the GPUs | |
bio | 3 BigMem | 14 days | 1-7 threads per job | For high memory, low CPU jobs | |
analysis | 3 BigMem | 4 days | 1-7 threads per job | For serial, post-processing and data analysis | |
transfer | 1 | 1 day | 1 node | Shared use | Copy files between archival storage and scratch space |
Selecting a partition is done by adding a directive to the job submission script such as #SBATCH --partition=t1standard, or on the command line: $ sbatch -p t1standard
Please note: When running a job in the analysis or bio partition the --mem SBATCH directive must be set as to not affect other users' work. Adding: #SBATCH --mem=214G is the RCS recommendation
Anyone interested in gaining access to the higher-priority Tier 2 partitions (t2small, t2standard) by subscribing to support the cluster or procuring additional compute capacity should contact uaf-rcs@alaska.edu.
Last updated