{"title":"Acousto-optic tunable filter (AOTF)","authors":"T. Nakazawa","doi":"10.1109/HPSR.2002.1024628","DOIUrl":null,"url":null,"abstract":"RHiNET-3/SW is an optical network switch that enables high-performance parallel computing in a flooror building-area environment. It can provide 10-Gbps x 8-port network switching with 1-km transmission length using 12channel parallel optical links. The switch consists of eight pairs of IO-Gbps optical link modules, one-chip CMOS ASIC switch LSI and four CMOS ASIC deskew-LSIs. Each optical link module transmits 8B10B encoded 10-bit characters and a 1-bit transmission clock at a data rate of 1.25 Gbps. The deskew-LSI can adjust the skew of 256 ns in I-km multimode fiber ribbons. By the frame synchronization used in the deskew-LSI, RHiNET-3/SW provides a longtransmission-length synchronized parallel optical data transmission. RHiNET-3/S W provides topology-free, deadlockfree and reliable network communication for the high-performance parallel computing. In order to build topology-free and deadlock-flee network, the switch supports 32 virtual channels with 640-Byte buffer memory. By using large amount of embedded memory on the switch LSI, RHiNET-3/S W allows low-latency, high-bandwidth performance. To provide highly reliable network, the switch uses hop-by-hop retransmission mechanism. This mechanism does not require any software procedure such as TCP to maintain reliability. We produced a cabinet of RHiNET-3/SW as 1-unit height and a prototype system of RHiNET-3 in 19-inch rack. RHiNET-3/SW enables high-bandwidth (80 Gbps), and long-transmission-length( 1 km) network communication by means of distributed parallel computing. This work is done with Real World Computing Partnership (RWCP) Tsukuba research center and RWCP optical interconnection Hitachi laboratory.","PeriodicalId":180090,"journal":{"name":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2002-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"3","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Workshop on High Performance Switching and Routing, Merging Optical and IP Technologie","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/HPSR.2002.1024628","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 3

Abstract

RHiNET-3/SW is an optical network switch that enables high-performance parallel computing in a flooror building-area environment. It can provide 10-Gbps x 8-port network switching with 1-km transmission length using 12channel parallel optical links. The switch consists of eight pairs of IO-Gbps optical link modules, one-chip CMOS ASIC switch LSI and four CMOS ASIC deskew-LSIs. Each optical link module transmits 8B10B encoded 10-bit characters and a 1-bit transmission clock at a data rate of 1.25 Gbps. The deskew-LSI can adjust the skew of 256 ns in I-km multimode fiber ribbons. By the frame synchronization used in the deskew-LSI, RHiNET-3/SW provides a longtransmission-length synchronized parallel optical data transmission. RHiNET-3/S W provides topology-free, deadlockfree and reliable network communication for the high-performance parallel computing. In order to build topology-free and deadlock-flee network, the switch supports 32 virtual channels with 640-Byte buffer memory. By using large amount of embedded memory on the switch LSI, RHiNET-3/S W allows low-latency, high-bandwidth performance. To provide highly reliable network, the switch uses hop-by-hop retransmission mechanism. This mechanism does not require any software procedure such as TCP to maintain reliability. We produced a cabinet of RHiNET-3/SW as 1-unit height and a prototype system of RHiNET-3 in 19-inch rack. RHiNET-3/SW enables high-bandwidth (80 Gbps), and long-transmission-length( 1 km) network communication by means of distributed parallel computing. This work is done with Real World Computing Partnership (RWCP) Tsukuba research center and RWCP optical interconnection Hitachi laboratory.
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
声光可调滤波器
RHiNET-3/SW是一种光网络交换机,可在楼层或建筑区域环境中实现高性能并行计算。采用12路并行光链路,可提供10gbps × 8口网络交换,传输长度为1km。该交换机由8对IO-Gbps光链路模块、1片CMOS ASIC开关LSI和4片CMOS ASIC桌面LSI组成。每个光链路模块以1.25 Gbps的数据速率传输8B10B编码的10位字符和1位传输时钟。桌面型lsi可以在1 -km多模光纤带中调节256ns的斜度。通过在桌面lsi中使用的帧同步,RHiNET-3/SW提供了长传输长度的同步并行光数据传输。RHiNET-3/S W为高性能并行计算提供无拓扑、无死锁和可靠的网络通信。为了构建无拓扑和无死锁的网络,交换机支持32个虚拟通道和640字节的缓冲存储器。通过在开关LSI上使用大量嵌入式内存,RHiNET-3/S W实现了低延迟、高带宽的性能。为了提供高可靠性的网络,交换机采用逐跳重传机制。这种机制不需要任何软件程序(如TCP)来保持可靠性。我们生产了一个1单位高度的RHiNET-3/SW机柜和一个19英寸机架的RHiNET-3原型系统。RHiNET-3/SW通过分布式并行计算实现高带宽(80gbps)、长传输长度(1km)的网络通信。这项工作是与现实世界计算合作伙伴(RWCP)筑波研究中心和RWCP光互连日立实验室共同完成的。
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
Next generation carriers Internet backbone node architecture (MSN Type-X) Multipath packet switch using packet bundling Permutation in rearrangeable nonblocking optical MINs with zero first-order switching-element-crosstalk End-to-end delay differentiation of IP traffic aggregates using priority queuing models Delay and throughput analysis of the high speed variable length self-routing packet switch
×
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