HPC resources
General information about the resources available through the Slurm workload management system on the CCB cluster.
Introduction
The CCB cluster is a high-performance computing system shared by many users who run programs using a pool of computational resources to process data simultaneously.
However, the resources that may be used at any point in time are limited by the total amount of resources available on the CCB cluster.
To ensure fair access to all users, the CCB cluster uses the Slurm workload manager. Users are required to submit their work as job scripts that describe the commands they wish to execute, and the amount of resources that they wish to grant to those commands during their execution. In turn, Slurm controls the execution of submitted jobs when sufficient resources are available, and manages a queue of pending work while waiting for sufficient resources to become available.
Resources
The table below summarises the types of compute nodes available.
| Nodes | Cores (per node) | RAM (per node) | |
|---|---|---|---|
| Large nodes | 1 | 40 | 900GB |
| Batch nodes | 20 | 24 | 240GB |
The full list of compute nodes and associated information can be obtained as follows.
sinfo --Node --long
More information about individual compute nodes can be obtained using commands detailed in the page Slurm commands.