MRI/RUI: Development of a Computing Cluster to Support the University of Richmond Nuclear Physics Research Program at Jefferson Lab

University Of Richmond, Richmond VA

Investigators

Abstract

A computer cluster is being developed to support the University of Richmond (UR) research program in electromagnetic nuclear physics at the Thomas Jefferson National Accelerator Facility (Jefferson Lab). The system will consist of 50 slave nodes, a master node, and a 3-TByte RAID file server. Each of the slave computing nodes will have two 1-GHz Pentium III processors, 512 MBytes of RAM, and a 20-GByte hard disk. The computers will run the Linux operating system and will be configured on a local area network through a high-speed switch. The master node will be used to manage the execution of batch jobs and to act as a firewall to restrict access to the system and reduce network traffic among the nodes of the cluster. The cluster is modeled after a prototype system that was developed at UR and has been running successfully for the past two years. The system will be used for data analysis and Monte-Carlo simulations for the CEBAF Large Acceptance Spectrometer (CLAS) Collaboration in Hall B at Jefferson Lab. The goal of the CLAS research program is to study baryon and meson structure and spectroscopy, gluonic degrees of freedom, and multinucleon correlations in nuclei.