首页 | 本学科首页   官方微博 | 高级检索  
     

基于Intel Xeon Phi的激光等离子体粒子模拟研究
引用本文:姚文科 杜云飞 吴 强 杨灿群. 基于Intel Xeon Phi的激光等离子体粒子模拟研究[J]. 计算机工程与科学, 2014, 36(5): 809-813
作者姓名:姚文科 杜云飞 吴 强 杨灿群
基金项目:国家863计划资助项目(2012AA010903);国家自然科学基金资助项目(61170049)
摘    要:激光等离子体粒子模拟广泛用于探索极端物质状态下的科学问题。将一种基于粒子云网格方法的三维等离子体粒子模拟程序LARED P移植到Intel Xeon Phi协处理器上。在移植的过程中,综合运用了Native和Offload两种编程模式:首先运用Native模式对LARED P程序中热点计算任务进行优化研究,通过采用SIMD扩展指令使该计算任务获得了4.61倍的加速;然后运用Offload模式将程序移植到CPU-Intel Xeon Phi异构系统上,并通过使用异步数据传输和双缓冲技术分别提升了程序性能9.8%和21.8%。

关 键 词:LARED P  Intel Xeon Phi  Native模式  Offload模式  512位SIMD扩展指令  异步数据传输  双缓冲  
收稿时间:2013-08-10
修稿时间:2014-05-25

Research of LARED-P on Intel Xeon Phi
YAO Wen ke,DU Yun fei,WU Qiang,YANG Can qun. Research of LARED-P on Intel Xeon Phi[J]. Computer Engineering & Science, 2014, 36(5): 809-813
Authors:YAO Wen ke  DU Yun fei  WU Qiang  YANG Can qun
Affiliation:(1.National Key Laboratory for Parallel and Distributed Processing,National University of Defense Technology,Changsha 410073;2.College of Computer,National University of Defense Technology,Changsha 410073,China)
Abstract:Plasma simulations have been widely used to exploit scientific problems under extremely situations. The paper ports a particle in cell based plasma code, LARED P, to Intel Xeon Phi co processor. In order to accomplish this, two modes are employed, i.e. the Native mode and the Offload mode. Firstly, the Native mode is employed to study on the hot computing tasks, which have been accelerated to 4.61 times faster by using SIMD extension instructions. Secondly, the Offload mode is employed to transplant the whole code onto a CPU Intel Xeon Phi heterogeneous system. We also adopt optimizations such as asynchronous data transferring and double buffer technique to improve the performance. And we obtain 9.8% and 21.8% improvement respectively.
Keywords:LARED-P  Intel Xeon Phi  native mode  offload mode  512bit SIMD extension  asynchronous data transferring  double buffer,
本文献已被 CNKI 等数据库收录!
点击此处可从《计算机工程与科学》浏览原始摘要信息
点击此处可从《计算机工程与科学》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号