面向超级计算机的故障预测和容错关键技术研究.pdf

《面向超级计算机的故障预测和容错关键技术研究.pdf》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、分类号TP393学号11069061UDC密级公开工学博士学位论文面向超级计算机的故障预测和容错关键技术研究博士生姓名胡维学科专业计算机科学与技术研究方向高性能计算可靠性指导教师刘光明研究员国防科学技术大学研究生院二〇一七年二月ResearchonFailurePredictionandFault-toleranceTechnologyforSupercomputerCandidate:HuWeiSupervisor:ProfessorLiuGuangmingAdissertationSubmittedinpartialfulfillmentoftherequirementsforthede

2、greeofDoctorofEngineeringinComputerScienceandTechnologyGraduateSchoolofNationalUniversityofDefenseTechnologyChangsha,Hunan,P.R.ChinaFebruary1,2017国防科学技术大学研究生院博士学位论文目录摘要........................................................................................................................iABSTRACT..

3、.............................................................................................................v第一章绪论............................................................................................................11.1课题背景....................................................................

4、...............................11.1.1超级计算机发展现状....................................................................11.1.2面临的可靠性问题........................................................................31.2容错研究基础.........................................................................................

5、..81.2.1容错基本概念................................................................................81.2.2超级计算机系统组成....................................................................91.2.3故障类型........................................................................................121.2.4并行系统故障模型.........

6、...............................................................141.3研究综述...................................................................................................141.3.1空间冗余容错技术分析................................................................141.3.2时间冗余容错技术分析..............................

7、..................................161.3.3混合冗余容错技术分析................................................................201.4本文工作..................................................................................