JESPP: Joint Experimentation on Scalable Parallel Processors Supercomputers

摘要

This research has been manifestly successful in advancing the state of the art in large-scale military battlespace simulations and in conceiving, developing, implementing and testing trans-continentally distributed high performance computing for training, analysis and evaluation. Significant achievements include increasing the scale of simulations by one order of magnitude, championing a 256- node General Purpose Graphics Processing Unit (GPGPU) cluster, facilitating compute environment stability, developing an effective Distributed Data Grid, enabling vital advances required to meet Joint Forces Command's (JFCOM's) mission goals and designing the JLogger system for improved logging. The research has resulted in the acceptance of 37 peer- reviewed research papers. In performing this research, ISI developed needed capabilities for JFCOM experiments, developed a fault tolerant router network, successfully proposed a 256-node GPGPU cluster, taught a course on its use, created the JLogger system, enabled faster analysis and investigated review techniques. All of this has had a significant salutary impact on the defense posture of the nation, resulting in direct benefits to the warfighter.