SuperStar:一种可扩展高阶互连拓扑结构

J4 ›› 2014, Vol. 36 ›› Issue (06): 1034-1041.

SuperStar:一种可扩展高阶互连拓扑结构

雷斐,董德尊,廖湘科

（国防科学技术大学计算机学院,湖南长沙 410073）

收稿日期:2013-07-10 修回日期:2013-11-12 出版日期:2014-06-25 发布日期:2014-06-25
基金资助:
国家自然科学基金资助项目（61272483,61272482,61003301）

SuperStar：A scalable high-radix topology in
high performance interconnection network

LEI Fei，DONG Dezun，LIAO Xiangke

（College of Computer,National University of Defense Technology,Changsha 410073,China）

Received:2013-07-10 Revised:2013-11-12 Online:2014-06-25 Published:2014-06-25

摘要/Abstract

摘要：

随着高速信号传输技术和VLSI技术的发展，使用高阶路由器来应对因高性能计算机峰值性能不断攀升给高性能互连网络带来的新挑战已是发展需求；同时，如何利用高阶特性减少互连网络延迟和成本开销，以支持更大的网络规模是设计高性能互连网络拓扑结构的关键和突破点。针对目前基于高阶路由器的典型拓扑结构进行了分析，并在此基础上提出一个新的高阶拓扑架构SuperStar，其不仅具有较短的网络直径而且具有良好的可扩展性；通过在一个基于OMNeT++平台自主开发的高阶互连网络性能测评模拟器上设定不同的通信负载，测评各种拓扑结构在通信系统下实际的网络延迟和吞吐率的走势，以分析SuperStar的通信开销。

关键词: 高阶路由器, 高性能互连网络, SuperStar

Abstract:

The development of the transmission of high-speed signals and the VLSI technology motivate the use of high-radix routers to deal with the new challenge of interconnection networks. How to reduce the latency and cost of the interconnection networks is crucial to designing the highradix topology. Several typical high-radix topologies such as fat tree, flattened butterfly, and dragonfly are studied. The performance and cost of those highradix networks is analyzed theoretically and a variation of those highradix topologies, the SuperStar, is considered, whose diameter and cost are effective. Meanwhile, SuperStar takes advantages of high-radix routers to approve of larger systems. A high-radix interconnection network simulator, named xNetSim, is explored to evaluate these topologies. xNetSim is built in the OMNeT++ platform. In the simulation, the network loads are varied, and the throughput and latency of different topologies are verified. The performance of SuperStar is briefly analyzed in comparison to other kinds of highradix interconnection network topologies.

Key words: high-radix router;high performance interconnection network;SuperStar

雷斐,董德尊,廖湘科. SuperStar:一种可扩展高阶互连拓扑结构[J]. J4, 2014, 36(06): 1034-1041.

LEI Fei，DONG Dezun，LIAO Xiangke. SuperStar：A scalable high-radix topology in
high performance interconnection network [J]. J4, 2014, 36(06): 1034-1041.

参考文献

［1］Yang X, Liao X, Lu K, et al. The Tianhe1A supercomputer:Its hardware and software［J］. Journal of Computer Science and Technology, 2011,26(3):344351.
［2］Kim J, Dally W J, Towles B, et al. Microarchitecture of a highradix router［C］∥Proc of ISCA’05,2005:420431.
［3］Scott S, Abts D, Kim J, et al. The BlackWidow highradix clos network［C］∥Proc of ISCA’06, 2006:1628.
［4］Leiserson C E. Fattrees:Universal networks for hardware efficient supercomputing［J］.IEEE Transactions on Computers, 1985,34(10):892901.
［5］Kim J, Dally W J, Abts D. Flattened butterfly:A costefficient topology for highradix networks［C］∥Proc of ISCA’07, 2007:126137.
［6］Kim J, Dally W J, Scott S, et al. Technologydriven, highlyscalable dragonfly topology［C］∥Proc of ISCA’08,2008:7788.
［7］Binkert N,Davis A,Jouppi N,et al. Optical highradix switch design［J］. IEEE Micro, 2012,32(3):100109.
［8］Faanes A B G, Roweth D, Court T, et al. Cray cascade:A scalable HPC system based on a dragonfly network［C］∥Proc of International Conference on High Performance Conmputing, Networking, Storage and Analysis, 2012:19.
［9］Rajamony R, Arimilli L B, Gildea K. PERCS:The IBM POWER7IH highperformance computing system［J］. IBM Journal of Research and Development, 2011,55(3):112.
［10］Ahn J H, Binkert N, Davis A, et al. HyperX:Topology, routing, and packaging of efficient largescale networks［C］∥Proc of International Conference on High Performance Conmputing, Networking, Storage and Analysis, 2009:111.
［11］Valiant L G. A scheme for fast parallel communication［J］.SIAM Journal on Computing, 1982(11):350361.
［12］Minkenberg C, Herrera G R. Tracedriven cosimulation of highperformance computing systems using OMNeT++［C］∥Proc of the 2nd International Conference on Simulation Tools and Technology, 2009:1.

[1]	梁崇山, 戴艺, 徐炜遐. 一种基于Chiplet集成技术的超高阶路由器设计[J]. 计算机工程与科学, 2022, 44(02): 207-213.
[2]	王超, 曹继军, 罗章, 赖明澈, 徐炜遐. 面向HPC互连网络的低延迟前向纠错编码研究与实现[J]. 计算机工程与科学, 2020, 42(11): 1965-1972.
[3]	蒋句平, 董德尊, 唐虹, 齐星云, 常俊胜, 庞征斌. 大规模高性能互连拓扑性能分析[J]. 计算机工程与科学, 2020, 42(10高性能专刊): 1730-1736.
[4]	路文斌，宋新亮，卢宏生，丁亚军. 一种高阶混合互连网络拓扑结构[J]. 计算机工程与科学, 2017, 39(10): 1781-1787.
[5]	杨文祥，董德尊，李存禄，雷斐，孙凯旋，吴际. 一种多级无缓存高阶路由器的设计与实现[J]. 计算机工程与科学, 2017, 39(02): 245-251.
[6]	杨文祥，董德尊，雷斐，李存禄，吴际，孙凯旋. 高阶路由器结构研究综述[J]. 计算机工程与科学, 2016, 38(08): 1517-1523.
[7]	王永庆1,王克非1,肖立权1,刘路1，庞征斌2. 非对称交叉开关优化与设计[J]. J4, 2013, 35(11): 42-47.
[8]	雷斐，董德尊，柴燕涛，王克非，李存禄. 高阶互连网络拓扑结构性能分析与研究[J]. J4, 2013, 35(11): 111-118.