计算机科学 ›› 2013, Vol. 40 ›› Issue (Z11): 32-36.

• 无线网络与通信 • 上一篇    下一篇

大规模并行操作系统研究

邵宗有,王昭顺,许建卫   

  1. 北京科技大学信息工程学院 北京100083;北京科技大学信息工程学院 北京100083;无锡城市云计算中心有限公司 无锡214315
  • 出版日期:2018-11-16 发布日期:2018-11-16

Study of the Large Scale Parallel Operating System

SHAO Zong-you,WANG Zhao-shun and XU Jian-wei   

  • Online:2018-11-16 Published:2018-11-16

摘要: 通过对大规模并行超级计算机模拟器上运行的操作系统进行分析,发现超级计算机内部计算节点、I/O节点和服务节点对操作系统各模块的需求是不同的,一个通用的操作系统无法满足不同类型节点的需求。提出了一种轻核心结构的操作系统并在模拟器上实现了原型系统SandOS,它包括计算节点轻核心SandPOS、I/O节点操作系统FileServer和服务节点操作系统MonitorServer。对计算节点轻核心SandPOS与通用的操作系统在内存开销、调度效率、运行效率、页表管理等方面的性能进行了分析对比,结果显示出超大规模并行系统中,SandOS比通用的操作系统具有更好的执行效率。

关键词: 大规模,轻核心,操作系统 SandOS

Abstract: By analyzing the operating system running in a supercomputer simulator,it is found that different nodes in the supercomputer use different part of an operating system.However,a traditional operating system cannot meet the demand.In such case,a heterogeneous large scale operating system is designed and a prototype named SandOS is implemented based on the simulator.The prototype includes light weight kernel operating system SandPOS for compute nodes,FileServer for I/O nodes and MonitorServer for service nodes.By comparing SandPOS with traditional operational system in memory overhead,schedule efficiency,run efficiency and address translating efficiency,result shows that in large scale parallel system,SandOS has better efficiency than traditional operational system.

Key words: Large scale,Light weight,Operating system,SandOS

[1] Moreira J,Brutman M,et al.Designing a highly-scalable opera-ting system:the blue gene/L story[C]∥Proceedings of the 2006ACM/IEEE conference on Supercomputing.Tampa,Florida,November 2006
[2] Yoshii K,Iskra K.Performance and Scalability Evaluation of‘Big Memory’ on Blue Gene Linux[J].International Journal of High Performance Computing Applications,May 2010:148-160
[3] Lange J,Pedretti K.Palacios and Kitten:New High Performance Operating Systems For Scalable Virtualized and Native Supercomputing[C]∥2010IEEE International Symposium on Parallel & Distributed Processing (IPDPS).Atlanta,USA,April 2010
[4] Yoshii K,Iskra K.Characterizing the performance of “big me-mory” on blue gene linux[C]∥Proceedings of the 2009International Conference on Parallel Processing Workshops.Vienna,Austria,September 2009
[5] Wheat S R,et al.PUMA:an operating system for massively para-llel systems[C]∥Proceedings of the 27th Hawaii International Conference on System Sciences. Hawaii,USA,January 1994
[6] Brightwell R,Maccabe A B.On the appropriateness of commodity operating systems for large-scale,balanced computing systems[C]∥Proceedings of the 17th International Symposium on Parallel and Distributed Processing. Nice,France,April 2003
[7] 黄凯,徐志伟.可扩展并行计算-技术、结构与编程[M].北京:机械工业出版社,2001:37-39
[8] 包云岗,许建卫,陈明宇,等.一种新型计算机体系结构模拟器的研究与实现[J].系统仿真学报,2007
[9] Zajcew R,Roy P.An OSF/1UNIX for Massively Parallel Multicomputers[C]∥Proceedings of the Winter USENIX Confe-rence.San Diego,USA,January 1993
[10] Adiga N R,et al.An overview of the BlueGene/L supercompu-ter[C]∥Proceedings of the 2002ACM/IEEE conference on Supercomputing.Baltimore,Maryland,November 2002
[11] Cray XT3 Datasheet [N].http://www.craysupercomputers.com/downloads/CrayXT3/CrayXT3_Datasheet.pdf
[12] Linux Networx Cluster System Climbs to Third Fastest Supercomputer in the World [N].http://www.prnewswire.com/news-releases/linux-networx-cluster-system-climbs-to-third-fastest-supercomputer-in-the-world-71374077.html
[13] Mattson T G,Henry G.An overview of the Intel TFLOPS supercomputer [J].Intel Technology Journal,1998,1
[14] ASCI White[N].http://www.thocp.net/hardware/asci_whi-te.htm,2003
[15] Ron B,Ann F L,et al.Massively parallel computing using commodity components[M].Elsevier Science Publishers B.V.26:243-266
[16] Steve N.Scaling Linux to new heights:the SGI Altix 3000system [J].Specialized Systems Consultants, 2003,3
[17] The Small Linux for Big Computers[N].http://www.mcs.anl.gov/ research/ projects/zeptoos/links,2011
[18] 沈林峰,陈明宇,许建卫.兼容Linux应用环境的多粒度全系统模拟平台-SandUPSim [J].计算机工程,2005(22)
[19] TOP500Supercomputer Sites[N].http://www.top500.org/,2013

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!