CS Cluster Documentation

Welcome to the Computer Science Department’s cluster documentation. The CS Department maintains a cluster of 97 (3776 cores) machines with 200 GPUs and 48T RAM which are available to the CS members for general computing. The cluster uses Slurm for job submission.

For remote access and for general application computing, we maintain login servers (login.cs.duke.edu). These servers are intended for light use (e.g., editing files, compiling code, running small scripts, etc… we request that you do not run CPU-intensive computations on these machines. Please use the compute nodes for intensive applications.

For announcements, updates, and community messages please subscribe to the CS Cluster Announcements mailing list. For assistance, please email CS Cluster Support.

Note

The slurm updates have been completed. The following changes were made to the GPU device names:

A5000 -> a5000 A6000 -> a6000 2080rtx -> rtx_2080 titanv -> titan_v

Make sure to update any batch commands that reference the old GPU device names. Please let us know if you have any questions or experience any issues.