摘要
对芯片面积、能耗上的严格限制是片上网络与宏观网络的最大不同。片上网络路由器中的缓存占用了大量的芯片面积和功耗,因此无缓存片上网络得到广泛关注。它完全去掉路由器内的缓存,通过偏转未获得有效端口的微片,处理微片对输出端口的竞争。本文对无缓存偏转网络的原理及关键技术进行了研究,包括拓扑结构、仲裁策略等。最后,通过与有缓存网络的对比,对无缓存网络的优势与劣势进行了总结。
Compared with off-chip network, area occupancy and power consumption are two main constraints in Network-on-Chips ( NoCs ). A high proportion of them is consumed by the buffers in the input ports of router. Therefore, bufferless NoC, which eliminates in-router buffers and copes with contention by dropping or deflecting packets, has received extensive attention. Existing bufferless NoCs mainly apply deflection routing, thus this paper makes a survey on the principle and critical techniques of bufferless deflection NoCs policy, etc. Finally we summarize the advantages and disadvantages of bufferless N . including topology, compared with the arbitration traditional buffered NoC
出处
《中国集成电路》
2012年第5期48-53,56,共7页
China lntegrated Circuit
基金
国家大学生创新创业训练计划项目的支持
关键词
片上网络
无缓存
偏转
关键技术
Network-on-Chip
bufferless
deflection
key technique