Sweetgum's Biggest Users Recognized

During its nearly 10 year period of service, sweetgum generated over 4,400,000 CPU hours in support of approximately 150,000 computations submitted by 1300 faculty and students conducting research at, or in collaboration with, Mississippi's public universities. MCSR would like to recognize the "biggest users" of sweetgum from each of Mississippi's four research universities over the last 10 years:

JSU: Pierre Bonifassi (404,307 CPU hours). Dr. Bonifassi is currently a Visiting Researcher in the Interdisciplinary Nanotoxicity CREST Center, Jackson State University (Jerzy Leszczynski, Director.)

UM: Yuanqing Ding (132,103 CPU hours). Dr. Ding is a post-doctoral research associate at The University of Mississippi, where he conducts molecular research on natural products with Dr. Li Xing-Cong and Dr. Daneel Ferreira.

MSU: Delwar Hossain (84,053 CPU hours). Dr. Hossain is currently a post-doctoral fellow in Computational Manufacturing and Design at the Center for Advanced Vehicular Systems at Mississippi State University, where he works with Dr. Mark F. Horstemeyer. Much of his work at MCSR work was performed as a Ph.D. student in Computational Chemistry at MSU, and as a postdoctoral fellow at MSU with Dr. S. R. Gwaltney.

USM: Praveen Madasu (83,509 CPU hours.) Mr. Madasu is a Ph.D. student in the research group of Dr. Janice Phillips, Assistant Professor in the Department of Chemistry and Biochemistry at The University of Southern Mississippi.

Sweetgum was MCSR's first major parallel computing resource, and was heavily used for many years by a variety of research applications. Most of the use recently has been in support of computational chemistry. Sweetgum has been self-maintained by MCSR and IT staff for several years, in order to reduce costs. Although relatively stable, it has been increasingly difficult to maintain, with scarce parts availability. It has also been sensitive to power problems, with occasional extended downtime. User CPU utilization has declined significantly in the last year and averages about 20% currently, reflecting the system's age and modest performance when compared to other, more powerful MCSR HPC systems.