资源描述:
《一种基于模糊聚类的网格dag任务图调度算法》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、ISSN1000-9825,CODENRUXUEWE-mail:jos@iscas.ac.cnJournalofSoftware,Vol.17,No.11,November2006,pp.2277−2288http://www.jos.org.cnDOI:10.1360/jos172277Tel/Fax:+86-10-62562563©2006byJournalofSoftware.Allrightsreserved.∗一种基于模糊聚类的网格DAG任务图调度算法1,2+1,21,21,2杜晓丽,蒋昌俊,徐国荣,丁志军1(同济大学电子与信息工程
2、学院,上海201804)2(国家高性能计算机工程技术研究中心同济分中心,上海201804)AGridDAGSchedulingAlgorithmBasedonFuzzyClustering1,2+1,21,21,2DUXiao-Li,JIANGChang-Jun,XUGuo-Rong,DINGZhi-Jun1(ElectronicsandInformationEngineeringSchool,TongjiUniversity,Shanghai201804,China)2(TonjiBranch,NationalEngineeringandT
3、echnologyCenterofHighPerformanceComputer,Shanghai201804,China)+Correspondingauthor:Phn:+86-21-69589864,E-mail:du_xiaoli@163.com,http://www.tongji.edu.cnDuXL,JiangCJ,XuGR,DingZJ.AgridDAGschedulingalgorithmbasedonfuzzyclustering.JournalofSoftware,2006,17(11):2277−2288.http://
4、www.jos.org.cn/1000-9825/17/2277.htmAbstract:Focusingontheproblemoftaskschedulingunderlarge-scale,heterogeneousanddynamicenvironmentsingridcomputing,aheuristicalgorithmbasedonfuzzyclusteringispresented.Manypreviousschedulingalgorithmsneedtosearchandcompareeveryprocessingcel
5、linthetargetsysteminordertochooseasuitableoneforatask.ThoughthosemethodscangetanapprovingMake-span,undoubtedly,itwouldincreasetheentireruntime.Agroupoffeatures,whichdescribethesyntheticperformanceofprocessingcellsinthetargetsystem,aredefinedinthispaper.Withthesedefinedfeatu
6、res,thetargetsystem,alsocalledprocessingcellnetwork,ispretreatedbyfuzzyclusteringmethodinordertorealizethereasonableclusteringofprocessornetwork.Intheschedulingstage,theclusterwithbettersyntheticperformancewillbechosenfirst.Thereisnoneedtosearcheveryprocessingcellinthetarge
7、tsystemateveryschedulingstep.Therefore,itlargelyreducesthecostonchoosingwhichprocessingcelltoexecutethecurrenttask.Thedesignofthereadytask’spriorityconsidersnotonlytheinfluencethatcomesfromtheexecutingofnodesoncriticalpath,butalsotheinfluenceinducedbyheterogeneousresource,o
8、nwhichthetaskwillbescheduled.Inthelastpart,thealgorithm’sperformanceisanalyzedandc