BlueGem Supercomputer Cluster
BlueGem is the BrisSynBio Supercomputer Cluster. It has over 900 Intel Haswell CPU cores, and 8 nVidia K80 GPUs. These are spread over 53 compute nodes and 4 GPU nodes, all connected together using a high-speed QDR Infiniband network, and attached to a high-performance and resilient Panasas network disk (60 TB).
Please read the pages below to get more information about how to use BlueGem.
- Disk space - information about how the disk is laid out, and where you should write your simulation data.
- Backups - information about backups on BlueGem (there are NO BACKUPS!).
- Accounts - information about the job accounting system. Every job must be tied to a project account.
- Queues - information about the job queues on BlueGem.
- Submitting Jobs - how to submit a job on BlueGem.
- Job Limits - What are the limits imposed on jobs that run on BlueGem?
- Bin - How to control where the output is written from a job script.
- Bin - How to write simple job scripts.
- Multi-node (MPI) Job Scripts - How to write multi-node (MPI) job scripts.
- Examples - A set of example job scripts for typical jobs.
BlueGem also runs its own webserver, on which you can get status information and see the jobs running on the queue. To access this webserver, use your browser to navigate to https://bluegem.acrc.bris.ac.uk. This link will only work from within the University firewall (i.e. on University computers), and you will need to log in using your BlueGem username and password. Please ignore any "invalid certificate" errors that your browser may report when you first connect to BlueGem.
Equipment Manager: Dr Christopher Woods