This week, I attended Supercomputing 2019, a huge conference dedicated to high performance computing, and learned about the latest advances in research computing hardware and software. The timing was perfect because Swarthmore just purchased a large cluster computer that will be available for use by the campus community. Delivery is scheduled for the end of December 2019 and we anticipate that it will be up and running by the start of the spring 2020 semester.
The system has:
- 14 CPUs with a total of 280 cores for computation
- 8 NVIDIA RTX 2080 Ti GPUs (Graphical Processing Units)
- Nearly 100TB of storage
- High speed Infiniband networking
- Special high-memory nodes with up to 768GB of RAM
The purchase was made possible from a coalition: physicists Tristan Smith and Jesse Rivera, engineer Vidya Ganapati, ITS, and physicist Daniel Grin at Haverford College. Funders will have priority access to the on the components that they purchased, but all faculty and students will have access to the system for research and teaching use.
The new cluster will allow faculty from all disciplines at the college to accelerate their computational-based research. Academic Technology also supports high performance computing through work with XSEDE, an NSF sponsored project to provide supercomputer access to researchers, and cloud computing services such as Amazon Web Services.
If you have computational work that could be accelerated with a more powerful computer, we’d love to discuss possible uses for your teaching and research. Get in touch with your Academic Technologist to learn more.