全部 标题 作者
关键词 摘要

OALib Journal期刊
ISSN: 2333-9721
费用:99美元

查看量下载量

相关文章

更多...
电子学报  2015 

结构网格CFD应用程序在天河超级计算机上的高效并行与优化

DOI: 10.3969/j.issn.0372-2112.2015.01.007, PP. 36-44

Keywords: 计算流体力学,多区结构网格,并行计算,天河计算机,CPU+MIC异构计算

Full-Text   Cite this paper   Add to My Lib

Abstract:

对多区结构网格大规模CFD流场模拟的高效并行方法进行了研究,以天河超级计算机平台的CPU同构计算环境和CPU+MIC异构计算环境为例,重点讨论了CFD应用特点与超级计算机运行环境相适应的性能优化与改进策略,发展了一系列多层次并行与性能优化方法.通过在天河2高性能计算平台上进行了多个算例的数值模拟,验证了这些优化方法的并行效果;在CPU+MIC异构平台上模拟的最大CFD问题规模达到6800亿个网格单元,共使用137.6万CPU+MIC处理器核,测试结果表明在CPU+MIC异构平台上移植优化后的程序性能提高2.6倍左右,且具有良好的可扩展性.

References

[1]  Deng Xiaogang,Maekawa H.Compact high-order accurate nonlinear schemes[J].Journal of Computational Physics,1997,130(1):77-91.
[2]  刘昕.高阶精度加权紧致非线性格式研究与其在复杂流动中的应用[D].博士学位论文,中国空气动力研究与发展中心.2004. Liu Xin.Study ofhigh-order accurate weighted compact nonlinear schemes and applications to complicated flows[D].PhD Thesis,China Aerodynamics Research and Development Center.2004.(in Chinese)
[3]  Deng Xiaogang,Mao Meiliang,Tu Guohua,et al.Geometric conservation law andapplications to high-order finite difference schemes with stationary grids[J].Journal of Computational Physics,2011,230(4):1100-1115.
[4]  王勇献,张理论,刘巍,等.结点型多区结构网格的奇点重构算法[A].第十五届全国计算流体力学会议[C].山东烟台,2012.609-614. Wang Yongxian,Zhang Lilun,Liu Wei,et al.A fast algorithm for reconstructing singular connection in the multi-block CFD applications[A].Proceedings of 15th Conference of CFD in China[C].Yantai,China,2012.609-614.(in Chinese)
[5]  Top 500 supercomputer sites[OL].http://www.top500.org,2014-03-12.
[6]  Liu Li,Liu Li,Yang Guangwen.A highly efficient GPU-CPU hybrid parallel implementation of sparse LU factorization[J].Chinese Journal of Electronics.2012,21(1):7-12.
[7]  王光学,张玉伦,等.WCNS高精度并行软件的大规模计算研究[J].计算机工程与科学,2012,34(8):125-130. Wang Guangxue,Zhang Yulun,et al.A study on massively parallel computation[J].Computer Engineering and Science,2012,34(8):125-130.(in Chinese)
[8]  Wang Yongxian,Zhang Lilun,Liu Wei,et al.Efficient parallel implementation of large scale 3D structured grid CFD applications on the Tianhe-1A supercomputer[J].Computers & Fluids,2013,80(10):244-250.
[9]  Liu Buquan,Yao Yiping,Wang huaimin.On the Technology of high-performance parallel simulation[J].Chinese Journal of Electronics,2012,21(1):1-6.
[10]  Thibault J,Senocak I.CUDA Implementation of a Navier-Stokes solver on multi-GPU desktop platforms for incompressible flows[A] Aerospace Sciences Meetings[C].USA:American Institute of Aeronautics and Astronautics,2009.758-772.
[11]  Jacobsen D A,Senocak I.Multi-level parallelism for incompressible flow computations on GPU clusters[J].Parallel Computing,2013,39 (1):1-20.
[12]  Subhash Saini,Haoqiang Jin,Dennis Jespersen,et al.An early performance evaluation of many integrated core based SGI Rackable computing system[A].Proceedings of Supercomputing[C].Denver,Colorado:SC,2013.17-22.
[13]  王勇献,张理论,车永刚,等.高阶精度CFD应用在天河2 系统上的异构并行模拟与性能优化[A].中国高性能计算学术年会[C].广西桂林,2013.1-31. Wang Yong-Xian,Zhang Li-Lun,et al.Heterogeneous computing and optimization on Tianhe-2 supercomputer system for high-order accurate CFD applications[A].Proceedings of HPC China[C].Guilin,China,2013.1-31.(in Chinese)
[14]  王勇献,张理论,刘巍,等.CFD并行计算中的多区结构网格二次剖分方法与实现[J].计算机研究与发展,2013,50(8):1762-1768. Wang Yong-xian,Zhang Li-lun,Liu Wei,et al.Grid repartitioning method of multi-block structured grid for parallel CFD simulation[J].Computer Research and Development,2013,50(8):1762-1768.(in Chinese)
[15]  Wang Yong-Xian,Zhang Li-Lun,Che Yong-Gang,et al.Improved algorithm for reconstructing singular connection in multi-block CFD Application[J].Transaction of Nanjing University of Aeronautics & Astronautics,2013,30(S):51-57.
[16]  李邦明,鲍麟,童秉纲.高超声速压缩拐角峰值热流位置预测模型研究[J].力学学报,2012,44(5):869-875. Li Bangming,Bao Lin,et al.Theoretical modeling for the prediction of thelocation of peak heat flux for hypersonic compression ramp flow[J].Chinese Journal of Theoretical and Applied Mechanics,2012,44(5):869-875.(in Chinese)

Full-Text

Contact Us

[email protected]

QQ:3279437679

WhatsApp +8615387084133