期刊文献+

数据驱动并行计算的3层软件架构设计及应用 预览 被引量:2

Three-Level Hierarchical Software Architecture for Data-Driven Parallel Computing with Applications
在线阅读 下载PDF
分享 导出
摘要 数据驱动并行计算是科学与工程计算中普遍存在的一类计算,其执行通常依赖于数据流有向图.在实际应用中,结点调度、数据通信和数值计算紧耦合并发执行,较难解耦编程,这给应用软件的协同研制和代码复用带来困难.借助于统一形式的数据流有向图并行算法框架,分无环有向图调度、无环有向图建模和数值计算3个层次,设计了软件体系结构,实现于并行自适应结构网格应用支撑软件(J parallel adaptive structured mesh applications infrastructure,JASMIN)框架的通量扫描积分构件中,有力地支持了结点调度、数据通信和数值计算的解耦编程.研究成果成功应用于科学计算中典型的中子输运计算,典型的代码开销测试和2 048个处理器核的并行性能测试表明,软件架构及其构件化实现是有效的. Data-driven parallel computing is widely used in scientific and engineering computation. Most of these computations are based on data dependency diagraphs. In real word applications, vertex scheduling, data communication and numerical computation are executed concurrently in a tightly coupled way, and it is hard to implement in a decoupled manner, which imposes difficulties for both application software co-design and code reuse. To address these problems, in this paper, we propose a hierarchical software architecture and implement it in sweeping integrator component, which is the part of the J adaptive structured mesh infrastructure (JASMIN). The hierarchical architecture is based on a unified algorithm framework for data dependency diagraph computation. It consists of three levels, including directed acyclic graph (DAG) scheduling level, DAG modeling level and numerical computation level. This design provides strong support for decoupled implementation of vertex scheduling, data communication and numerical computation, which are essential in data-driven parallel computing. We apply this result in typical scientific computing applications such as neutron transportation. The sequential implementation overhead and parallel performance results are obtained on a parallel computer with 2048 CPU cores. These results suggest that our hierarchical software architecture and component-based implementation are both effective and efficient.
作者 张爱清 莫则尧 杨章 Zhang Aiqing, Mo Zeyao, and Yang Zhang (Institute of Applied Physics and Computational Mathematics, Beijing 100094) (National Key Laboratory of Science and Technology on Computational Physics, Beijing 100094)
出处 《计算机研究与发展》 EI CSCD 北大核心 2014年第11期2538-2546,共9页 Journal of Computer Research and Development
基金 国家自然科学基金重点项目(61033009) 国家“九七三”重点基础研究发展计划基金项目(2011CB309702) 国家“八六二三”高技术研究发展计划基金项目(2012AA01A309)
关键词 数据驱动 并行计算 无环有向图 JASMIN框架 软件体系结构 解耦 data driven parallel computing directed acyclic graph J adaptive structured meshinfrastructure (JASMIN) software architecture decoupling
作者简介 Zhang Aiqing, born in 1976. PhD and associate professor. Her main research interests include parallel computing.zhang_aiqing@iapcm. ac. cn Mo Zeyao, born in 1971. PhD and professor. His research interests include parallel computing. Yang Zhang, born in 1984. PhD and assistant professor. His research interests include parallel computing.
  • 相关文献

参考文献20

  • 1Bey J, Wittum G. Downwind numbering: A robust multigrid method for convection diffusion problems on unstructured grids [J]. Applied Numerical Mathematics, 1997, 23 (1) 177-192. 被引量:1
  • 2Hackbusch W, Probst T. Downwind Ga-Seidel smoothing for convection dominated problems [J]. Numerical Linear Algebra with Applications, 1997, 4(2): 85-102. 被引量:1
  • 3Han H, Ilin V, Kellogg R, et al. Analysis of flow directed iterations[J]. Journal of Computational Mathematics, 1992, 10(1): 57-76. 被引量:1
  • 4Adams M L, Larsen E W. Fast iterative methods for discrete-ordinates particle transport calculations [J]. Progress in Nuckor Energy, 2002, 40(1): 3-159. 被引量:1
  • 5Balsara D. Fast and accurate discrete ordinates methods for multidimensional radiative transfer: Part I, basic methods [J]. Journal of Quantitative Spectroscopy and Radiative Transfer, 2001, 69(6): 671-707. 被引量:1
  • 6Valiant L. Abridging model for parallel computation [J]. Communications of the ACM, 1990, 33(8): 108-111. 被引量:1
  • 7张宝林 谷同祥 莫则尧.数值并行计算原理与方法[M].北京:国防工业出版社,1999.. 被引量:9
  • 8Mo Zeyao, Zhang Aiqing, Cao Xiaolin. Towards a parallel framework of grid-based numerical algorithms on DAGs [C] //Proc of the 20th Int Parallel and Distributed Processing Syrup (IPDPS'06). Piscataway, NJ: IEEE, 2006:310-317. 被引量:1
  • 9Mo Zeyao, Zhang Aiqing, Wittum G. Scalable heuristic algorithms for the parallel execution of data flow acyclic digraphs [J]. SIAM Journal on Scientific Computing, 2009, 32(5) : 3626-3642. 被引量:1
  • 10张爱清,莫则尧.有向图并行计算中一种新的结点调度算法[J].计算机学报,2009(11):2178-2186. 被引量:6

二级参考文献23

  • 1莫则尧,傅连祥,阳述林.非结构网格上求解中子输运方程的并行流水线Sn扫描算法[J].计算机学报,2004,27(5):587-595. 被引量:14
  • 2耿素云 屈婉玲.离散数学[M].北京:高等教育出版社,1997.. 被引量:4
  • 3Lee Y.F.,Ryder B.A comprehensive approach to parallel data flow analysis.In:Proceedings of the 1992 International Conference on Supercomputing,Washington,DC,USA,1992,236~247 被引量:1
  • 4Koutsougeras C.,Papachristou C.A.Vemuri,data flow graph partitioning to reduce communication cost.In:Proceedings of the 19th Annual Workshop on Microprogramming,New York,1986,82~91 被引量:1
  • 5Lee B.,Hurson A.R.,Feng T.Y.A vertically layered allocation scheme for data flow systems.Journal of Parallel and Distributed Computing,1991,11(3):175~187 被引量:1
  • 6Mo Ze-Yao.More efficient performance analysis for parallel codes.Journal of Numer Mathematics & Applications,2001,23(2):29~42 被引量:1
  • 7Lewis E.E.,Miller W.F.Computational Methods of Neutron Transport.New York:John Wiley & Sons Publisher,1984,401 被引量:1
  • 8Wareing T.A.,McGhee J.M.,Morel J.E.,Pautz S.D.Discontinuous finite methods Sn methods on 3-D unstructured grids.Nuclear Science and Engineering,2001,138(3):256~268 被引量:1
  • 9Mo Ze-Yao,Fu Lian-Xiang.Parallel flux sweeping algorithm for neutron transport on unstructured grid.Journal of Supercomputing,2004,30(1):5~17 被引量:1
  • 10Kirk Schloegel,George Karypis,Vipin Kumar.Graph partitioning for high performance scientific simulations.In:Dongarra J.,Foster I.,Fox G.et al.eds.Sourcebook of Parallel Computing.Morgan Kaufmann Publishers,2003,491~541 被引量:1

共引文献14

同被引文献89

  • 1莫则尧,张爱清,曹小林,左风丽.多介质辐射流体力学数值模拟中的并行计算研究[J].自然科学进展,2006,16(3):287-292. 被引量:7
  • 2高兴誉,曹小林,赵伟波,等.数万核上复杂应用程序的性能测试与分析[J].软件学报,2011,22(增刊(2)):157-162. 被引量:2
  • 3Amarasinghe S, Hall M, Lethin R, et.al. DOE Workshop on Exascale Programming Challenges Report. July 27-29, 2011, USA. 被引量:1
  • 4莫则尧.面向高性能科学与工程计算的领域编程框架研究[J].中国计算机学会通讯,2014,10(1):8-12. 被引量:1
  • 5Heroux M A. Next generation programming environment: what we need and do not need. DOE Workshop on Exascale Programming Challenges, July 27-29, 2011, USA. 被引量:1
  • 6A.Dubey, A.Almgren, J.Bell, M.Berzins, S.Brandt, G.Bayan, P.Colella, D.Graves, M.Lijewski, F.Loffler, B.Oshea, E.Schnetter, B.V.Straalen, K.Weide. A survey of high level frameworks in block-structured adaptive mesh refinement packages, Special Issue on Domain-Specific Language and High-Level Frameworks for High Performance Computing, Journal of Parallel and Distributed Computing, 74(12) (2014):3217-3227. 被引量:1
  • 7BoxLib. 2011. https://ccse.lbnl.gov/BoxLib. 被引量:1
  • 8P. Colella, D. Graves, et al.Chombo software Package for AMR Applications Design Document, Tech. Rep., LBNL, 2009. 被引量:1
  • 9Cactus developers, Cactus Computational Toolkit. 2013. http://www.cactuscode.org/. 被引量:1
  • 10Enzo developers, Enzo astrophysical AMR code. 2013. http://enzo-proj ect.org/. 被引量:1

引证文献2

二级引证文献5

投稿分析

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部 意见反馈