The RCE servers are divided into two categories:
- Login servers: These servers provide a graphical remote desktop, or command line interface to the RCE. They operate like a personal desktop environment, and do not have the resources to run intensive jobs. Intensive jobs are run on the back-end computation cluster, and display their output on the Login node.
Cluster servers: Jobs that need a lot of memory, CPU cores, or disk I/O are best run on the computation cluster as "RCE Powered" jobs. The exact hardware specs of the cluster servers will vary as we replace aging componenets, but in general the RCE cluster is composed of rackmount servers with:
- CPUs: Intel Xeon CPUs running 24-64 threads per server (2-4 CPU sockets, 6-8 CPUs per socket, and 2 threads per CPU) @ 2.2 - 2.3 GHz
- Memory: 128 - 1024 GB each
Also see running multi-threaded jobs for more information.
Disk space options are as follows:
- Home directories are allocated a small amount of personal storage space. We do not increase this space, but you may request a project directory.
- Project directories are available in storage sizes suited to each particular researcher. Please contact us to request storage space. Fees may be applicable to sizes over a certain amount. To read about using your storage space, please see:
- 1 TB of shared scratch space is available to all users on our cluster. Top level scratch space is world-writeable and -readable (Unix 1777 permissions). User created directories are only owner writeable/readable (1700) or owner/group (2770) if you are a member of a research group. Do not use the scratch space for permanent storage, or for confidential data.