Molecular Dynamics Simulation of Macromolecules Using Graphics Processing Unit

TL;DR

This paper presents GPU-based molecular dynamics algorithms for macromolecules, achieving significant speed-ups over CPU-based methods and validated through polymer crystallization simulations, offering an efficient alternative for large-scale studies.

Contribution

The paper introduces GPU-optimized MD algorithms for macromolecules that run entirely on GPU, providing about 10 times faster performance than CPU-based simulations.

Findings

Achieved approximately 10x speed-up over CPU simulations.

Validated GPU MD results with polymer crystallization experiments.

Provided an inexpensive and scalable alternative for large-scale MD simulations.

Abstract

Molecular dynamics (MD) simulation is a powerful computational tool to study the behavior of macromolecular systems. But many simulations of this field are limited in spatial or temporal scale by the available computational resource. In recent years, graphics processing unit (GPU) provides unprecedented computational power for scientific applications. Many MD algorithms suit with the multithread nature of GPU. In this paper, MD algorithms for macromolecular systems that run entirely on GPU are presented. Compared to the MD simulation with free software GROMACS on a single CPU core, our codes achieve about 10 times speed-up on a single GPU. For validation, we have performed MD simulations of polymer crystallization on GPU, and the results observed perfectly agree with computations on CPU. Therefore, our single GPU codes have already provided an inexpensive alternative for macromolecular simulations on traditional CPU clusters and they can also be used as a basis to develop parallel GPU programs to further speedup the computations.