设计高带宽密度、低延迟、高能效的片上互连

Yong Wang, Hui Wu
{"title":"设计高带宽密度、低延迟、高能效的片上互连","authors":"Yong Wang, Hui Wu","doi":"10.1109/ISLPED.2017.8009171","DOIUrl":null,"url":null,"abstract":"For future high-performance computing chips, on-chip interconnect requires large bandwidth-density, low latency, and high energy-efficiency, which pose significant design challenges. This paper presents a design space exploration of transmission line based on-chip interconnect. First, we conduct an optimization of on-chip transmission lines to minimize the size, channel loss and inter-symbol-interference (ISI), hence to maximize the bandwidth-density. Based on the result, differential coplanar waveguide (CPW) with 55-µm pitch size is chosen as the transmission line topology. Next, channel capacities of channel lengths from 2 to 8 cm are characterized based on time-domain pulse responses. Various equalizers are studied, which are used to increase the ISI-limited channel capacity. To make better use of the large equalized channel capacity, pulse amplitude modulation (PAM) is employed instead of traditional non-return-to-zero (NRZ) signaling. A link budget analysis is then conducted to find the optimal modulation format for each channel. To verify our analyses, several transceivers are designed in 28-nm CMOS technology. An 84-Gb/s PAM-8 transceiver achieves 1.5-Gb/s/μm bandwidth-density over a 4-cm channel. The unrepeated bandwidth-density is 6.1 Gb/s/μm·cm, which is almost 10 times larger compared to prior work.","PeriodicalId":385714,"journal":{"name":"2017 IEEE/ACM International Symposium on Low Power Electronics and Design (ISLPED)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2017-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"5","resultStr":"{\"title\":\"Design high bandwidth-density, low latency and energy efficient on-chip interconnect\",\"authors\":\"Yong Wang, Hui Wu\",\"doi\":\"10.1109/ISLPED.2017.8009171\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"For future high-performance computing chips, on-chip interconnect requires large bandwidth-density, low latency, and high energy-efficiency, which pose significant design challenges. This paper presents a design space exploration of transmission line based on-chip interconnect. First, we conduct an optimization of on-chip transmission lines to minimize the size, channel loss and inter-symbol-interference (ISI), hence to maximize the bandwidth-density. Based on the result, differential coplanar waveguide (CPW) with 55-µm pitch size is chosen as the transmission line topology. Next, channel capacities of channel lengths from 2 to 8 cm are characterized based on time-domain pulse responses. Various equalizers are studied, which are used to increase the ISI-limited channel capacity. To make better use of the large equalized channel capacity, pulse amplitude modulation (PAM) is employed instead of traditional non-return-to-zero (NRZ) signaling. A link budget analysis is then conducted to find the optimal modulation format for each channel. To verify our analyses, several transceivers are designed in 28-nm CMOS technology. An 84-Gb/s PAM-8 transceiver achieves 1.5-Gb/s/μm bandwidth-density over a 4-cm channel. The unrepeated bandwidth-density is 6.1 Gb/s/μm·cm, which is almost 10 times larger compared to prior work.\",\"PeriodicalId\":385714,\"journal\":{\"name\":\"2017 IEEE/ACM International Symposium on Low Power Electronics and Design (ISLPED)\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2017-07-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"5\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2017 IEEE/ACM International Symposium on Low Power Electronics and Design (ISLPED)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ISLPED.2017.8009171\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2017 IEEE/ACM International Symposium on Low Power Electronics and Design (ISLPED)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ISLPED.2017.8009171","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 5

摘要

对于未来的高性能计算芯片,片上互连需要大带宽密度、低延迟和高能效,这对设计提出了重大挑战。本文提出了基于片上互连的传输线的设计空间探索。首先,我们对片上传输线进行优化,以最小化尺寸、信道损耗和符号间干扰(ISI),从而最大化带宽密度。在此基础上,选择55µm节距的差分共面波导(CPW)作为传输线拓扑。接下来,基于时域脉冲响应表征通道长度从2到8 cm的通道容量。研究了各种各样的均衡器,用来增加isi限制的信道容量。为了更好地利用大的均衡信道容量,采用脉冲调幅(PAM)代替传统的非归零(NRZ)信令。然后进行链路预算分析,以找到每个信道的最佳调制格式。为了验证我们的分析,采用28纳米CMOS技术设计了几个收发器。84gb /s PAM-8收发器在4cm通道上实现1.5 gb /s/μm的带宽密度。无重复带宽密度为6.1 Gb/s/μm·cm,比之前的研究成果提高了近10倍。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
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
Design high bandwidth-density, low latency and energy efficient on-chip interconnect
For future high-performance computing chips, on-chip interconnect requires large bandwidth-density, low latency, and high energy-efficiency, which pose significant design challenges. This paper presents a design space exploration of transmission line based on-chip interconnect. First, we conduct an optimization of on-chip transmission lines to minimize the size, channel loss and inter-symbol-interference (ISI), hence to maximize the bandwidth-density. Based on the result, differential coplanar waveguide (CPW) with 55-µm pitch size is chosen as the transmission line topology. Next, channel capacities of channel lengths from 2 to 8 cm are characterized based on time-domain pulse responses. Various equalizers are studied, which are used to increase the ISI-limited channel capacity. To make better use of the large equalized channel capacity, pulse amplitude modulation (PAM) is employed instead of traditional non-return-to-zero (NRZ) signaling. A link budget analysis is then conducted to find the optimal modulation format for each channel. To verify our analyses, several transceivers are designed in 28-nm CMOS technology. An 84-Gb/s PAM-8 transceiver achieves 1.5-Gb/s/μm bandwidth-density over a 4-cm channel. The unrepeated bandwidth-density is 6.1 Gb/s/μm·cm, which is almost 10 times larger compared to prior work.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
期刊最新文献
A low power duobinary voltage mode transmitter Frequency governors for cloud database OLTP workloads Tutorial: Tiny light-harvesting photovoltaic charger-supplies A 32nm, 0.65–10GHz, 0.9/0.3 ps/σ TX/RX jitter single inductor digital fractional-n clock generator for reconfigurable serial I/O Monolithic 3D IC designs for low-power deep neural networks targeting speech recognition
×
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