基于OpenCL的Lammps短程力算法优化研究  

Short-range force algorithm optimization in Lammps based on OpenCL

作　　者：赵成龙[1] 施慧彬[1] 俞忻峰 

机构地区：[1]南京航空航天大学计算机科学与技术学院,江苏南京210016 [2]南京理工大学计算机科学与技术学院,江苏南京210094

出　　处：《计算机工程与科学》2015年第9期1614-1620,共7页Computer Engineering & Science

摘　　要：Lammps是用于分子动力学模拟及其相关问题的一款开源软件,可利用其了解固体、液体性质,应用广泛。支持使用CUDA及OpenCL进行GPU加速。因OpenCL具有跨平台特性,将其作为研究重点。总结了OpenCL内核编程中需要注意的设计原则并阐述了一种改进的阿姆达尔定律用于衡量异构平台理论加速性能。测试了Lammps短程力计算在Y485P平台下的性能参数。通过对短程力计算中的关键部分如邻接表的建立及短程力计算部分的内核代码进行优化,使其取得了更好的加速效果。Lammps is a widely used open source software for molecular dynamics simulations and re- lated issues, through which we can understand solid and liquid nature. It enables the CUDA and the OpenCL to accelerate calculation. Since the OpenCL has cross-platform features, it becomes a research focus. We summarize the OpenCL kernel programming principles and propose an improved Amdahl＇s law which can measure the acceleration performance on heterogeneous platforms. The short-range force al- gorithm in Lammps on Y485P platform is tested. We achieve better acceleration effect by optimizing the kernel source codes of the key parts, such as neighbor-list building and short-range force calculation.

关 键 词：GPU Lammps OPENCL 分子动力学 短程力 内核优化 

分 类 号：TP302[自动化与计算机技术—计算机系统结构]