摘要
现有的高性能计算机光互连网络大都需要对数据报文进行光-电-光转换并缓存,或者需要预先申请并建立从源节点到目的节点的光链路,这在一定程度上限制了网络性能.提出了一种既不需要对光数据报文进行缓存排队,也不需要预先申请并建立光链路的新型的光互连网络结构BOIN.给出了BOIN网络上的无死锁和无活锁路由算法,推导出BOIN网络中数据传输的延时上界,并通过模拟实验对比了BOIN网络与其它两种典型的网络的性能.实验结果表明,BOIN光互连网络具有良好的延时和吞吐率特性,能够很好地满足高性能计算机系统的要求.
Most of the revealed optical interconnect networks within the high performance computers require either the buffering and O-E-O conversion of the data packets or the pre-assigning of the optical path from the source to destination, which to a certain extent influence the performance metrics such as latency and throughput. Aiming at the limitation mentioned above, a novel Bufferless Optical Interconnection Network (BOIN) for high performance computer is brought forward. Together with the data link protocol, deadlock freedom and livelock freedom routing algorithms, the upper bound of the data transmission latency in BOIN are also presented. The simulation is based on the comparison of BOIN and the other two similar networks. The results show that BOIN has the advantage over the rest that can deliver high throughput at low latency, which can well satisfy the need of high performance computing systems.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2008年第11期2171-2177,共7页
Acta Electronica Sinica
基金
国家自然科学基金(No.60633050No.60603061)
关键词
光互连网络
无缓存
无死锁路由
无活锁路由
网络性能
optical interconnect networks
bufferless
livelock freedom routing
deadlock freedom routing
network performance