Colgate University Computer Science Department Computing Cluster

Colgate University's Computer Science Department has a computational cluster managed by Toshiro K. Ohsumi. Its primary focus is on interdisciplinary simulations, mainly as a development test bed and prototype for a larger cluster Colgate University hopes to obtain through a NSF MRI grant in 2007. Faculty from many departments within Colgate have an account on the cluster. The cluster will also be used for the computer science course, COSC 445, Parallel and Distributed Computing, which will be taught in the Spring of 2007.

Hardware

The cluster's hardware is supplied by PSSC Labs and comprises of 5 nodes (1 master and 4 slaves), with each node outfitted with 2 AMD Opteron 265 (1.8 GHz dual core 64-bit) processors and 4GB of RAM, for a total of 20 processing cores. A dedicated 1 Gigabit Ethernet network connects the nodes.

Projects

Below are some of the projects currently [or at one time] running on the cluster:

Molecular replacement protein crystallography using a parallel genetic algorithm approach. This is a collaborative project between Roger S. Rowlett and T. K. Ohsumi.
Ephraim Woods III is using the cluster to do molecular dynamics using Gaussian, among other software.
A new distributed data structure library was developed by Pieter Hooimeijer (Class of 2006 - Highest Honors in Computer Science) and T. K. Ohsumi. It is optimized for storing and querying locations of clustered data efficiently.
A parallel algorithm for computing Van der Waerden numbers is being collaboratively developed by Aaron Robertson and T. K. Ohsumi.
There are projects which will utilize the cluster being developed by Daniel Schult and Damhnait McHugh. Details will be posted when available.

Accounts for Colgate Faculty and Students

Any Colgate faculty (and their collaborators) or student having a project requiring a cluster's computational capability may request an account by emailing Toshiro K. Ohsumi. Please see his web site for contact information. There is no limit on computational time. Since there are no batch processing queues, users are expected to share resources fairly.

News

None at this time.