Acquisition of Large-Memory, Many-Core Compute Node for Mathematical Science Research

Clemson University, Clemson SC

Investigators

Abstract

The purchase of a large-memory, many-core computing system and its integration into a large, distributed, high-performance computing cluster provides the opportunity for researchers at Clemson to increase the size of problems that can be solved in several research areas in the mathematical sciences and to scale up shared-memory, parallel algorithms in anticipation of new, many-core computing systems under development by chip manufacturers. In addition, it creates the possibility of developing innovative, heterogeneous, shared-distributed algorithms for problems where that approach is promising. Among the important areas where large-memory and heterogeneous computing systems promise to be effective are: computational optimization, simulation of 3-D filter systems, statistical models of reliability and survival, computational cell biology, biochemical modeling, quantum molecular dynamics, and combinatorial number theory. Large-memory, many-core systems are not widely available to researchers in the US. This resource will be a valuable addition to the pool of resources for scientific computing in the United States.