中国科学院软件研究所并行计算实验室,北京,100190
网络首发:2009-10-10,
纸质出版:2009
移动端阅览
王婷 1, 孙相征 1, 张云泉 1, 等. 曙光5000A天体大规模数值模拟软件性能测试[J]. 西安交通大学学报, 2009,43(10):71-75.
Performance Evaluation of the Simulation Software on Dawning 5000A for Large Scale Celestial Bodies[J]. 2009, 43(10): 71-75.
在国产百万亿次超级计算机曙光5000A上进行了天体大规模数值模拟软件的性能和可扩展测试实验
详细介绍了软件中的测试程序以及测试环境和过程
并对测试结果进行了分析.对于80×80×50的网格规模
采用每节点4进程测试了16~128个处理器核
每节点8进程、16进程分别测试了16~512个处理器核
相对加速比最终分别达到5.33、10.48和12.57
并行效率分别达到66.66%、32.58%和32.29%.对于160×160×100的网格规模
测试了每节点16进程的64~8 192个核的性能
最大相对加速比为12.46
并行效率为9.73%.测试结果表明
曙光5000A具有良好的性能
测试结果对软件下一步的优化研究具有重要的指导意义.
The performance and scalability evaluation of the large scale simulation software on planetary fluid dynamics are investigated by the experiments on the domestic 100 TFlops supercomputer
Dawning 5000A. The details of the software testing procedures
the environment and the process of the testing are introduced. For the mesh size of 80×80×50
the performance testing on 16 to 128 processor cores with mapping 4 processes on each node and the performance testing on 16 to 512 processor cores with mapping 16 or 8 processes on each node are performed
respectively. The relative speedups reach 5.33
10.48 and 12.57
respectively
and the corresponding parallel efficiencies of these 3 testing cases are 66.66%
32.58% and 32.29%
respectively. For the mesh size of 160×160×100
the performance testing on 64 to 8 192 processor cores with mapping 16 processes on each node gives the results that the relative speedup is 12.46
and its parallel efficiency is 9.73%. The experimental results reveal the good performance of Dawning 5000A and show important clues of the software optimization.
张云泉,孙家昶,袁国兴,等. 2008年中国高性能计算机TOP100排行榜分析与展望 [J]. 科研信息化技术与应用,2008,1(3): 71-78.
ZHANG Yunquan, SUN Jiachang, YUAN Guoxing, et al.Analysis of 2008 China HPC TOP100 rank list and perspectives [J]. e-Science Technology and Application, 2008,1(3): 71-78.
中国软件行业协会数学软件分会,国家863高性能计算机评测中心,中国计算机学会高性能计算专业委员会.2008年中国高性能计算机性能TOP100排行榜[EB/OL].(2009-03-13)[2009-03-28]. http:∥www.samss.org.cn/sites/shuxue/ndhyC.jsp?contentId=2473512102846.
CHANA K H, LI Ligang, LIAO Xinhao. Modelling the core convection using finite element andfinite difference methods [J]. Phys Earth Planet Interiors, 2006, 157(2): 124-138.
LI Ligang, LIAO Xinhao, ZHANG Keke. Linear and nonlinear instabilities in rotating cylindrical Rayleigh-Benard convection[J/OL]. Physical Review: E, 2008, 78(5): 12[2009-02-20]. http:∥link. aps. org/doi/10.1103/PhysRevE. 78. 056303.
LI Ligang, LIAO Xinhao, ZHANG Keke. Countertraveling waves in rotating Rayleigh-Benard convection [J/OL]. Physical Review: E, 2008, 77(2):4[2009-02-18]. http:∥link. aps. org/doi/10.1103/PhysRevE. 77. 027301.
李力刚. 球壳内行星流体动力学方程组的有限差分法[R]∥国家863高效能计算机及网格服务环境项目技术报告.北京:中国科学院,2008: 10-20.
杨超. 球壳对流的有限差分程序注记[R]∥国家863高效能计算机及网格服务环境项目技术报告. 北京:中国科学院,2008: 30-40.
DUKOWICZ J K, DVINSKY A S. Approximate factorization as a high order splitting for the implicit incompressible flow equations [J]. Journal of Computational Physics, 1992, 102(2): 336-347.
TUMINARO R S, HEROUX M, HUTCHINSON S A, et al. Official AZTEC users guide: version 2.1 [DB/OL]. [2009-02-20].http:∥www.CS.sundia. gov/CRF/pspapers/Aztec_og_2.1.ps
张云泉,孙家昶,唐志敏,等.数值计算程序的存储复杂性分析[J].计算机学报,2000,23(4):363-373.
ZHANG Yunquan, SUN Jiachang, TANG Zhimin, et al. Memory complexity analysis on numerical programs [J].Chinese Journal of Computers, 2000,23(4):363-373.
张云泉.面向高性能数值计算的并行计算模型DRAM(h)[J].计算机学报, 2003, 26(12): 1660-1670.
ZHANG Yunquan. DRAM(h): a parallel computation model for high performance numerical computing [J]. Chinese Journal of Computers, 2003, 26(12): 1660-1670.
CHEN Jing, ZHANG Yunquan, ZHANG Linbo, et al. Performance evaluation of allgather algorithms on terascale Linux cluster with fast ethernet [C]∥Proceedings of Eighth International Conference on High-Performance Computing in Asia-Pacific Region. Los Alamitos, CA, USA: IEEE Computer Society, 2005:437-442.
薛正华,董小社,伍卫国,等.自适应大规模服务器集群监控系统的构建 [J].西安交通大学学报, 2008, 42(4): 399-403.
XUE Zhenghua, DONG Xiaoshe,WU Weiguo, et al. Design and implementation of an adaptive monitoring system for large scale server clusters [J]. Journal of Xi'an Jiaotong University, 2008, 42(4): 399-403.
0
浏览量
14
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010802024621