一种无缓存片上网络交叉开关调度机制  

Crossbar Scheduling Mechanism for Bufferless Network-on-Chip

作　　者：刘亮亮[1] 韩国栋[1] 张帆[1] 陈庆强[1] 

机构地区：[1]国家数字交换系统工程技术研究中心,郑州450002

出　　处：《小型微型计算机系统》2013年第7期1579-1584,共6页Journal of Chinese Computer Systems

基　　金：国家"八六三"高技术研究发展计划项目(2009AA012201)资助

摘　　要：提出一种无缓存片上网络的交叉开关调度机制.无缓存是指将路由节点输入端口缓存以及相应的控制逻辑移除,以降低实现开销.该调度机制在每个输入端口设置一条自回环通道,并在"偏转"路由基础上采用一种基于"偏转"和自回环次数的优先级策略,以通过减小"偏转"次数提高无缓存片上网络的网络性能.实验表明,此调度机制相对于基本的无缓存机制,可保证在硬件开销与能耗不增的前提下,提高网络性能;相对于典型的有缓存机制,在负载较低时其网络性能更优,负载较高时性能略低,但可使硬件开销与能耗分别下降58.9%与40.2%,并使频率提高74.6%.A crossbar scheduling mechanism for bufferless NoC（Network-on-Chip） is proposed in this paper.The idea of bufferless is to remove the input buffer and the corresponding control logic of routers to reduce the implementation cost.In this mechanism,a loopback channel is configured at each inputport,and deflection routing is adopted as a foundation routing method.In addition,a priority schemme w hich is based on deflection and loopback counts is proposed.Thus the mechanism can improve the netw ork performance of bufferless NoC through decreasing the deflection counts.The results of experiments indicate that： the proposed mechanism could improve the netw ork performance w ithout increasing harw are or pow er consumption compared w ith basic bufferless NoC;it has better netw ork performance in light load and w orse in high load,but could respectively reduce harw are and pow er consumption by 58.9% and 40.2% and increase the frequency by 74.6% w hen compared w ith typical buffered NoC.

关 键 词：片上网络 无缓存 交叉开关调度 偏转路由 

分 类 号：TP302[自动化与计算机技术—计算机系统结构]