期刊文献+

Optimal algorithms for scheduling large-scale application on heterogeneous systems

Optimal algorithms for scheduling large-scale application on heterogeneous systems
下载PDF
导出
摘要 This paper studies optimal algorithms for scheduling large-scale application on heterogeneous systems using Divisible Load Theory.A more realistic and general model,i.e.,both processors and communication links may have different speeds and arbitrary start-up costs,and communication is in non-blocking mode,is introduced.Under such environment,the following results are obtained: ①Mathematic model and closed-form expressions both for the processing time and the fraction of load for each processor are derived;②the influence of start-up costs on the optimal processing time is analyzed;③for a given heterogeneous systems and a large-scale computing problem,optimal algorithms are proposed. This paper studies optimal algorithms for scheduling large-scale application on heterogeneous systems using Divisible Load Theory. A more realistic and general model, i. e. , both processors and communication links may have different speeds and arbitrary start-up costs, and communication is in non-blocking mode, is introduced. Under such environment, the following results are obtained:①Mathematic model and closed-form expressions both for the processing time and the fraction of load for each processor are derived ;②the influence of start-up costs on the optimal processing time is analyzed;③for a given heterogeneous systems and a large-scale computing problem, optimal algorithms are proposed.
出处 《重庆邮电大学学报(自然科学版)》 2008年第4期440-446,共7页 Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)
基金 the National Natural Scienee Foundation of China under Grant(10467007) Foundation of"863"(2006AA01Z414)
关键词 自动技术 数据通信 计算方法 程序设置 heterogeneous computing divisible load theory non-blocking mode of communication start-up costs schedu-ling algorithm
  • 相关文献

参考文献12

  • 1Veeravalli Bharadwaj,Debasish Ghose,Thomas G. Robertazzi.Divisible Load Theory: A New Paradigm for Load Scheduling in Distributed Systems[J].Cluster Computing.2003(1)
  • 2Hyoung-Joong Kim.A Novel Optimal Load Distribution Algorithm for Divisible Loads[J].Cluster Computing.2003(1)
  • 3BHAT P B,,RAGHAVENDRA C S,PRASANNA V K.Efficient collective communication in distributed hetero-geneous systems[].Journal of Parallel and DistributedComputing.2003
  • 4SAIF T,PARASHAR M.Understanding the behaviorand performance of non-blocking communications inMPI[].I Proceedings of Euro-Par :ParallelProcessing.2004
  • 5DROZDOWSKI M.Selected Problems of SchedulingTasks in Multiprocessor Computer Systems. http://www.cs.put.poznan.pl/~ma-ciejd/h.ps . 2008
  • 6SURESHS,MMANI V,MKAR S N.The effect of start-up delays in scheduling divisible load on bus networks:An alternate approach[].Computers and MathwithAppl.2003
  • 7Siegel HJ,,Ali S.Techniques for Mapping Tasks to Machines in Heterogeneous Computing Systems[].Special Issue on Hetero-geneous Distributed and Parallel Architectures:HardwareSoftware and Design Tools.2000
  • 8Beaumont O,Casanova H,Legrand A,et al.Scheduling divisi-ble loads on star and tree networks:Results and open problems[].IEEE Transon Parallel and Distributed Systems.2005
  • 9Bharadwaj V,Ghose D,Robertazzi T G.Divisible load theory:A new paradigm for load scheduling in distributed systems[].Cluster Computing.2003
  • 10Robertazzi T G.Ten reasons to use divisible load theory[].Computer.2003

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