UNION: A unified inter/intra-chip optical network for chip multiprocessors

Xiaowen Wu, Yaoyao Ye, Wei Zhang, Weichen Liu, M. Nikdast, Xuan Wang, Jiang Xu
{"title":"UNION: A unified inter/intra-chip optical network for chip multiprocessors","authors":"Xiaowen Wu, Yaoyao Ye, Wei Zhang, Weichen Liu, M. Nikdast, Xuan Wang, Jiang Xu","doi":"10.1109/NANOARCH.2010.5510930","DOIUrl":null,"url":null,"abstract":"As modern computing systems become increasingly complex, communication efficiency among and inside chips has become as important as the computation speeds of individual processor cores. Traditionally, inter-chip and intra-chip communication architectures are separately designed to maximize design flexibility under different constraints. However, jointly designing communication architectures for both inter-chip and intra-chip communication could potentially yield better solutions. In this paper, we present a unified inter/intra-chip optical network, called UNION, for chip multiprocessors (CMP). UNION is based on recent progress in nano-photonic technologies. It connects not only processors on a single CMP but also multiple CMPs in a system. UNION employs a hierarchical optical network to separate inter-chip communication traffic from intra-chip communication traffic. It fully utilizes a single optical network to transmit both payload packets and control packets. The network controller on each CMP not only manages intra-chip communications but also collaborate with each other to facilitate inter-chip communications. We compared CMPs using UNION with those using a matched electronic counterpart in 45 nm process. Based on eight applications, simulation results show that on average UNION improves CMP performance by 3.1X while reducing 92% of network energy consumption and 52% of communication delay.","PeriodicalId":306717,"journal":{"name":"2010 IEEE/ACM International Symposium on Nanoscale Architectures","volume":"563 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2010-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"21","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2010 IEEE/ACM International Symposium on Nanoscale Architectures","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/NANOARCH.2010.5510930","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 21

Abstract

As modern computing systems become increasingly complex, communication efficiency among and inside chips has become as important as the computation speeds of individual processor cores. Traditionally, inter-chip and intra-chip communication architectures are separately designed to maximize design flexibility under different constraints. However, jointly designing communication architectures for both inter-chip and intra-chip communication could potentially yield better solutions. In this paper, we present a unified inter/intra-chip optical network, called UNION, for chip multiprocessors (CMP). UNION is based on recent progress in nano-photonic technologies. It connects not only processors on a single CMP but also multiple CMPs in a system. UNION employs a hierarchical optical network to separate inter-chip communication traffic from intra-chip communication traffic. It fully utilizes a single optical network to transmit both payload packets and control packets. The network controller on each CMP not only manages intra-chip communications but also collaborate with each other to facilitate inter-chip communications. We compared CMPs using UNION with those using a matched electronic counterpart in 45 nm process. Based on eight applications, simulation results show that on average UNION improves CMP performance by 3.1X while reducing 92% of network energy consumption and 52% of communication delay.
UNION:用于芯片多处理器的统一片间/片内光网络
随着现代计算系统变得越来越复杂,芯片之间和芯片内部的通信效率已经变得与单个处理器核心的计算速度一样重要。传统上,为了在不同约束条件下最大限度地提高设计灵活性,芯片间和芯片内的通信架构是分开设计的。然而,联合设计芯片间和芯片内通信的通信架构可能会产生更好的解决方案。本文提出了一种用于芯片多处理器(CMP)的统一的片间/片内光网络,称为UNION。UNION是基于纳米光子技术的最新进展。它不仅可以连接单个CMP上的处理器,还可以连接系统中的多个CMP。UNION采用分层光网络将片间通信流量与片内通信流量分离。它充分利用单光网络传输有效载荷报文和控制报文。每个CMP上的网络控制器不仅管理芯片内的通信,而且还相互协作以促进芯片间的通信。我们比较了在45纳米工艺中使用UNION的cmp与使用匹配电子对应的cmp。基于八个应用的仿真结果表明,UNION平均将CMP性能提高了3.1倍,同时降低了92%的网络能耗和52%的通信延迟。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信