期刊文献+

一种改进的机群SPMD编程模型静态任务分配算法

Improved Static Load Balabcing Strategy for SPMD Programming Model on Cluster
下载PDF
导出
摘要 机群系统中,结点机差异及动态运行环境是导致系统运行状态不均衡的主要原因.提出了结点机差异及负载测量指标 和方法.以SPMD编程模型中的单一并行任务为测试程序,在机群系统性能测试、分析的基础上,进行运行时间相同原则基础 上的静态任务分配;在应用运行过程中,通过对并发线程运行状态测试,用已完成任务的高速设备重复执行低速机器未完成任 务,选取曩快运行结果,从而缩小不均衡因素的影响.以空间划分大规模M0nte Carlo仿真问题为例进行算法实际测试,取得良 好运行效果. System heterogeneity and load measure metrics and method are given. Using SPMD programming model's single procedure as benchmark to test nodes dynimic performance. Eased on the principle of having same runtime to static distribute load to each node at begining. During the process, checking concurrent threads' running-state to repeat the uncompleted task on slow node by faster machine, so as to control the influence of imbalance factor. Using large size Monte Carlo simulation as application, nice speedup is achieved.
出处 《小型微型计算机系统》 CSCD 北大核心 2005年第10期1778-1780,共3页 Journal of Chinese Computer Systems
基金 国家"八六三"计划项目(2001AA111081)资助.
关键词 SPMD编程模型 机群任务分配策略 MONTE Carlo仿真 SPMD programming model load balancing strageics Monte Carlo simulation
  • 相关文献

参考文献9

  • 1Leonidas Georgiadis, Christos Nikolaou, Alexander Thomasian. A fair workload allocation policy for heterogeneous systems[J]. Parallel Distrib. , Comput. ,2004,64:507-519. 被引量:1
  • 2Antonis K. Garofalakis J, Mourtos I et al. A hierarchical adaptive distributed algorithm for load balancing [J] Parallel Distrib. , Comput. ,2004,64:151-162. 被引量:1
  • 3Plastino, Ribeiro C C, Rodriguez N. Developing SPMD applications with load balancing[J]. Parallel Computing, 2003,29 : 743-766. 被引量:1
  • 4Michael O Neary, Bernd O Christiansen, Peter Cappello et al.Javalin: Parallel computing on the Internet[J]. Future Generation Computer Systems, 1999,15: 659-674. 被引量:1
  • 5Luis F G Sarmenta, Satoshi Hirano. Bayanihan: Building and studying web-based volunteer computing systems using Java [J]. Future Generation Computer Systems, 1999,15: 675-686. 被引量:1
  • 6Robert Leslie, Sati Mckenzie. Evaluation of loadsharing algorithms for heterogeneous distributed systems [J]. Computer Communications, 1999,22: 376-389. 被引量:1
  • 7Marc H Willebeek-LeMair. Strategies for dynamic load balancing on highly parallel computers[J]. IEEE Transaction on Parallel and Distributed Systems, 1993,4(9): 979-993. 被引量:1
  • 8Stavros A Zenios. High-performance computing in finance: The last 10 years and the next [J]. Parallel Computing, 1999,25:2149-2175. 被引量:1
  • 9Les Clewlow and Chris Stricklan. Implementing Derivatives Models[M]. England: John Wiley & Sons Ltd, 1998. 被引量:1

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