首页 | 本学科首页   官方微博 | 高级检索  
     

k-ary n-cube网络中高速开关TH-Switch的设计与路由算法
引用本文:杜毅,李三立.k-ary n-cube网络中高速开关TH-Switch的设计与路由算法[J].计算机学报,1999,22(1):16-23.
作者姓名:杜毅  李三立
作者单位:清华大学计算机科学与技术系,北京,100084
基金项目:国家攀登计划研究项目,博士点科研基金
摘    要:随着处理器速度的不断提高,互联网络逐渐成为限制网络并行计算系统性能发挥的主要瓶颈。本文中所介绍的TH-GBNet是基于动态crossbar开关的可扩展高速互联网络。TH-GBNet系统以TH-Switch为基本开关单元,具有很强的扩展能力。其中的TH-Switch开关单元具有8个双向端口,单向的数据传输率能够达到1.056Gbps。本文主要介绍了高速开关TH-Switch的结构及其在k-ary n

关 键 词:网络并行计算  互联网络  路由算法
修稿时间:1998年1月6日

DESIGN AND ROUTING ALGORITHM OF THE HIGH-SPEED TH-SWITCH IN K-ARY N-CUBE INTERCONNECTION NETWORK
DU Yi,LI San-Li.DESIGN AND ROUTING ALGORITHM OF THE HIGH-SPEED TH-SWITCH IN K-ARY N-CUBE INTERCONNECTION NETWORK[J].Chinese Journal of Computers,1999,22(1):16-23.
Authors:DU Yi  LI San-Li
Abstract:With the increasingly rapid growth of processor speed, the interconnection network for networked parallel computing (NPC) has become an essential factor influencing on the efficiency of NPC. TH GBNet in this paper is a high speed scaleable, dynamic crossbar based interconnection network. TH GBNet not only has better scalability, but also can achieve high bandwidth and low latency data transmission and provides the powerful support for the network communication of NPC. As a basic unit of TH GBNet, TH Switch has eight full duplex ports and the unidirectional bandwidth of the port can reach 1.056 Gbps, the transmission latency is only 30ns in transmitting 32 bits data. This paper illustrates the structure of TH Switch and the design of routing algorithm for k ary n cube interconnection network in TH Switch. For reducing the heavy load of the high dimension channel in Idle routing algorithm for message passing, an improved idle algorithm with introducing a pseudo random variable is described. In resolving the contention of the outport of the TH Switch, the arbitration unit with token cyclic test mode is designed. Finally, the analysis based on the simulation of improved idle routing algorithm shows that it is not only very simple and easy to implement using hardware, but also has high efficiency.
Keywords:Networked parallel computing  interconnection network  routing algorithm  
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号