Scalable Algorithms for Molecular Dynamics Simulations on Commodity Clusters

TLDR

MD simulations of biomolecular systems often require days to months, yet many scientifically and pharmaceutically relevant events occur on longer timescales that are currently inaccessible. The study introduces new algorithms and implementation techniques to accelerate parallel MD simulations beyond current state‑of‑the‑art codes. The authors employ a novel parallel decomposition and message‑passing scheme, new communication primitives, and single‑precision numerical techniques to reduce communication overhead while maintaining accuracy. Desmond, implementing these methods, delivers unprecedented throughput and scalability on commodity clusters, outperforming all prior codes, including surpassing IBM Blue Gene/L's Blue Matter on a benchmark with 2 K Opteron processors versus 32 K Blue Gene/L processors.

Abstract

Although molecular dynamics (MD) simulations of biomolecular systems often run for days to months, many events of great scientific interest and pharmaceutical relevance occur on long time scales that remain beyond reach. We present several new algorithms and implementation techniques that significantly accelerate parallel MD simulations compared with current state-of-the-art codes. These include a novel parallel decomposition method and message-passing techniques that reduce communication requirements, as well as novel communication primitives that further reduce communication time. We have also developed numerical techniques that maintain high accuracy while using single precision computation in order to exploit processor-level vector instructions. These methods are embodied in a newly developed MD code called Desmond that achieves unprecedented simulation throughput and parallel scalability on commodity clusters. Our results suggest that Desmond's parallel performance substantially surpasses that of any previously described code. For example, on a standard benchmark, Desmond's performance on a conventional Opteron cluster with 2K processors slightly exceeded the reported performance of IBM's Blue Gene/L machine with 32K processors running its Blue Matter MD code