Efficient optical interconnections for data-center computing systems

A. Krishnamoorthy
{"title":"Efficient optical interconnections for data-center computing systems","authors":"A. Krishnamoorthy","doi":"10.1109/E3S.2013.6705884","DOIUrl":null,"url":null,"abstract":"Within the past decade, the semiconductor computing industry has developed multicore and multithreaded core processors to overcome the challenges and shrinking benefits of traditional technology scaling. Multichip systems built using these components will require immense amount of off-chip bandwidth and low latency chip-to-chip links at the lowest energy cost possible. Wavelength-division multiplexed (WDM) silicon photonics have the potential to provide a solution for this immense interconnect problem. At Oracle, we are aggressively building a portfolio of active and passive nanophotonic devices, circuits, and multichip packaging with the aim to achieve sub-picojoule per bit communication links between computing elements in a large array “Macrochip”. To achieve ultralow energy consumption will certainly require integrating best-in-breed photonic devices with electronic circuits. While the juxtaposition of silicon photonic devices and VLSI circuits on the same silicon substrate represents the most intimate integration of electronic and photonic technologies, achieving this will require an immense amount of sophistication in design and process integration. Instead, hybrid integration of aggressive components, built on individually optimized technology platforms, is a pragmatic approach to achieving peak performance. Such “photonic bridge” chip components may be used as units in a larger transmitter or receiver array or as drop-in communication physical layer elements in a multi-chip computing node.","PeriodicalId":231837,"journal":{"name":"2013 Third Berkeley Symposium on Energy Efficient Electronic Systems (E3S)","volume":"116 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2013-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2013 Third Berkeley Symposium on Energy Efficient Electronic Systems (E3S)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/E3S.2013.6705884","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

Abstract

Within the past decade, the semiconductor computing industry has developed multicore and multithreaded core processors to overcome the challenges and shrinking benefits of traditional technology scaling. Multichip systems built using these components will require immense amount of off-chip bandwidth and low latency chip-to-chip links at the lowest energy cost possible. Wavelength-division multiplexed (WDM) silicon photonics have the potential to provide a solution for this immense interconnect problem. At Oracle, we are aggressively building a portfolio of active and passive nanophotonic devices, circuits, and multichip packaging with the aim to achieve sub-picojoule per bit communication links between computing elements in a large array “Macrochip”. To achieve ultralow energy consumption will certainly require integrating best-in-breed photonic devices with electronic circuits. While the juxtaposition of silicon photonic devices and VLSI circuits on the same silicon substrate represents the most intimate integration of electronic and photonic technologies, achieving this will require an immense amount of sophistication in design and process integration. Instead, hybrid integration of aggressive components, built on individually optimized technology platforms, is a pragmatic approach to achieving peak performance. Such “photonic bridge” chip components may be used as units in a larger transmitter or receiver array or as drop-in communication physical layer elements in a multi-chip computing node.
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
数据中心计算系统的高效光互连
在过去的十年中,半导体计算行业已经开发了多核和多线程核心处理器,以克服传统技术扩展的挑战和缩小的好处。使用这些组件构建的多芯片系统将需要大量的片外带宽和低延迟的芯片到芯片链路,以尽可能低的能源成本。波分复用(WDM)硅光子学有潜力为这一巨大的互连问题提供解决方案。在Oracle,我们正在积极地构建一个有源和无源纳米光子器件、电路和多芯片封装的组合,目标是在大型阵列“Macrochip”的计算元件之间实现每比特亚皮焦耳的通信链接。要实现超低能耗,必然需要将同类最佳的光子器件与电子电路集成在一起。虽然在同一硅衬底上并置硅光子器件和超大规模集成电路代表了电子和光子技术的最密切集成,但实现这一目标将需要大量复杂的设计和工艺集成。相反,在单独优化的技术平台上构建侵略性组件的混合集成是实现最佳性能的实用方法。这种“光子桥”芯片组件可以用作较大的发射器或接收器阵列中的单元,或用作多芯片计算节点中的插入式通信物理层元件。
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 去求助
来源期刊
自引率
0.00%
发文量
0
期刊最新文献
Device design considerations for ultra-thin body non-hysteretic negative capacitance FETs Ultra-Low power neuromorphic computing with spin-torque devices Power-efficient server utilization in compute clouds Energy transparency from hardware to software Prospects for high-aspect-ratio FinFETs in low-power logic
×
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
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 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 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1