Particle-Mesh-Ewald(PME)算法在GPU上的实现

分子动力学模拟(MD)是分子模拟的一类常用方法,为生物体系的模拟提供了重要途径。由于计算强度大,目前MD可模拟的时空尺度还不能满足真实物理过程的需要。作为CPU的加速设备,近年来,GPU为提高MD计算能力提供了新的可能。GPU编程难点主要在于如何将计算任务分解并映射到GPU端并合理组织线程及存储器,细致地平衡数据传输和指令吞吐量以发挥GPU的最大计算性能。静电效应是长程作用,广泛存在于生物现象的各个方面,对其精确模拟是MD的重要组成部分。Particle-Mesh-Ewald(PME)方法是公认的精确处理静电作用的算法之一。本文介绍在本实验室已建立的GPU加速分子动力学模拟程序GMD的基础上,基于NVIDIACUDA,采用GPU实现PME算法的策略,针对算法中组成静电作用的三个部分即实空间、傅立叶空间和能量修正项,分别采用不同的计算任务组织策略以提升整体性能。使用事实上的标准算例dhfr进行的测试结果表明,实现PME的GMD程序,性能分别是Gromacs4.5.3版单核CPU的3.93倍,8核CPU的1.5倍,基于OpenMM2.0加速的Gromacs4.5.3GPU版本的1.87倍。

GPU-enabled implementations of Particle-Mesh-Ewald method

Molecular dynamics (MD) is a basic method in molecular simulation that offers a computational approach to study the behavior of biomolecules at atomic detail, but such simulations are still quite limited in size and timescale to meet the spatio-temporal of real world physical process because MD is computational intensive. Owing to the recent advances in the hardware and software architecture, the graphics processing unit (GPU) has shown its potential to accelerate MD simulation. Electrostatic effects play an important role in various biological processes, such as the polypeptide chain folding, enzyme activity and protein self-assembly. Accurate simulation of electrostatic interactions is essential for MD. Using the truncated method, electrostatic interactions can get considerable acceleration as van der Waals interactions. But, the cut-off method is approximate and for biological systems, people tend to select more accurate algorithms. Particle-Mesh-Ewald (PME) algorithm is proven to be an accurate calculation of the electrostatic interactions. This paper presents a GPU-enabled implementation of PME, which is the extension to GMD, a GPU based molecular dynamics program. With the implemented PME, the program GMD has been tested using dhfr, the de facto standard benchmark, and obtains speedup of 3.93, 1.5 and 1.87 in computing performance compared to the GROMACS 4.5.3 CPU (single core) , Gromacs 4.5.3 CPU(8 cores) and GPU version (OpenMM 2.0 based) respectively.