期刊文献+

基于事件的异构平台并行程序性能可视化研究

Research of parallel program performance visualization under heterogeneous platforms based on event
下载PDF
导出
摘要 为了便于用户快速、直观地了解到机群系统中并行应用程序的性能情况,将Linux计算机群与Windows控制显示平台相结合,提出了一种基于事件的异构平台并行程序性能可视化方法。该方法以MPI作为底层编程环境,在高层使用MPE技术,依据动态性能检测方式获取程序执行过程信息;设计C#语言及Jumpshot日志图形化分析集成工具实现并行程序性能可视化。实验结果表明,该方法可准确、直观地反映程序性能信息,有助于程序员简便、有效地对并行程序进行量化分析,对提高机群系统的可用性、改善程序性能及效率等方面具有较高的实用价值。 To understand the performance of parallel programs quickly and intuitively,combined Linux computing cluster with Windows control and display platform,a method of parallel program performance visualization under heterogeneous platforms based on event is presented.The method uses MPI as the underlying conditions,and embeds MPE in the high-level,and according to dynamic detection obtains performance information,designs C# and Jumpshot to achieve performance visualization.The results show that,this method reflects the program performance information accurately and achieves the quantitative analysis of parallel programs more conveniently and effectively.
出处 《计算机工程与设计》 CSCD 北大核心 2010年第24期5282-5284,5293,共4页 Computer Engineering and Design
基金 国家自然科学基金项目(60603047) 辽宁省教育厅高等学校科研基金项目(05L209)
关键词 异构平台 性能检测 事件 MPE 可视化 heterogeneous performance testing event MPE visualization
  • 相关文献

参考文献7

二级参考文献26

  • 1王忱 ,刘吟 ,蒋昌俊 ,张兆庆 .一种并行程序性能分析工具的体系结构设计[J].信息技术与信息化,2004(5):29-31. 被引量:3
  • 2孙建国,孔力,郑艳.水厂高压控制室模拟屏控制系统的设计[J].微计算机信息,2005,21(1):17-18. 被引量:12
  • 3许开新.可扩展集群计算机系统集成开发环境IPCE2研究与设计[D].北京:清华大学,1999.
  • 4[1]Simitci H. Pablo MPI Instrumentation Users Guide. Technical Report, University of Illinois at Urbana Champaign, 1996
  • 5[2]Vampire Users Manual. http://www. pallas.com/
  • 6Dodd P S, Racishankar C V. Monitoring and debugging distributed real-time programs [J]. Software-Practice and Experience, 1992, 22(10) : 863 - 877.
  • 7TIAN Laisheng, HUANG Lianshu, TAO Jie.The implementation of DC in a loosely-coupled distributed environment [J]. Chinese J Advanced Software Research,1995, 2(3): 227-236.
  • 8Borgeest R, Dimke B, Hansen O. A trace based performance evaluation tool for parallel real time system [J]. Parallel Computing, 1995, 21(4) : 551 - 564.
  • 9Hofmann R, Klar P, Mohr B, et al. Distributed performance monitoring: methods, tools, and applications [J]. IEEE Transactions on Parallel and Distributed Systems, 1994,5(6): 585-598.
  • 10McBryan O A. An overview of message passing environments[J]. Parallel Computing, 1994, 20(4) : 417 - 444.

共引文献16

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