Enhanced molecular dynamics performance with a programmable graphics processor

Abstract

Design considerations for molecular dynamics algorithms capable of taking advantage of the computational power of a graphics processing unit (GPU) are described. Accommodating the constraints of scalable streaming-multiprocessor hardware necessitates a reformulation of the underlying algorithm. Performance measurements demonstrate the considerable benefit and cost-effectiveness of such an approach, which produces a factor of 2.5 speed improvement over previous work for the case of the soft-sphere potential.