SDS Cluster

The SDS high performance computing cluster is accessible to students, postdocs, and faculty in the department. The system is designed to handle high-capacity statistical simulations (e.g., parallel MCMC or other simulations) as well as train basic deep-learning models. The system consists of 96 CPU compute cores (2x AMD EPYC Milan 7643), four Nvidia A4000 16GB GPUs, 512GB of RAM, and modest storage capacity. The system is designed to give quick and easy access to students, postdocs, and faculty in SDS who may not be ready to navigate the more substantial computing resources available through other university resources.

Use Policy

To ensure that our cluster is appropriately used and resources are appropriately shared, users must adhere to the following:

1. nice your jobs. We all want our jobs to run fast, but be nice. For example, nice R CMD BATCH run-a2-sims-1.R &

2. Keep your home directory clean i.e., delete files that you no longer need.

3. Be respectful of computing resources.

Access/Use Instructions

Click here for SDS Cluster Instructions: Using R

Click here for SDS Cluster Instructions: Using Python

Please email me for any suggested edits or additions to these documents, or questions. Thank you to Yunshan Duan and Ritwik Vashistha for their help!