{"title":"用于千兆多计算机通信的光子互连","authors":"J.R. Sauer;D.J. Blumenthal;A.V. Ramanan","doi":"10.1109/80.166999","DOIUrl":null,"url":null,"abstract":"The ways in which photonic and optoelectronic technologies could play an important role in future highly scalable and flexible interconnects for multicomputer parallel processing systems are discussed. For electronic interconnect implementation, the primary limitations arise from transmission drive power requirements, limited bandwidth, and the crosstalk-limited length. It is shown that photonic interconnects can relieve these bottlenecks in order to allow systems to scale to large numbers of nodes without degrading the interconnect performance. As an example, a network architecture capable of interconnecting thousands of processors with multigigabit average access rate per user, and peak access rates an order of magnitude higher is presented. The network topology is a shuffle-exchange, multihop, multipath, wraparound direct interconnect that utilizes self-routing and a deflection flow control technique to simplify and speed the processing. An experimental 2*2 photonic switching node based on the presented techniques is described.<\n<ETX>></ETX>","PeriodicalId":100626,"journal":{"name":"IEEE LTS","volume":"3 3","pages":"12-19"},"PeriodicalIF":0.0000,"publicationDate":"1992-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/80.166999","citationCount":"10","resultStr":"{\"title\":\"Photonic interconnects for gigabit multicomputer communications\",\"authors\":\"J.R. Sauer;D.J. Blumenthal;A.V. Ramanan\",\"doi\":\"10.1109/80.166999\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"The ways in which photonic and optoelectronic technologies could play an important role in future highly scalable and flexible interconnects for multicomputer parallel processing systems are discussed. For electronic interconnect implementation, the primary limitations arise from transmission drive power requirements, limited bandwidth, and the crosstalk-limited length. It is shown that photonic interconnects can relieve these bottlenecks in order to allow systems to scale to large numbers of nodes without degrading the interconnect performance. As an example, a network architecture capable of interconnecting thousands of processors with multigigabit average access rate per user, and peak access rates an order of magnitude higher is presented. The network topology is a shuffle-exchange, multihop, multipath, wraparound direct interconnect that utilizes self-routing and a deflection flow control technique to simplify and speed the processing. An experimental 2*2 photonic switching node based on the presented techniques is described.<\\n<ETX>></ETX>\",\"PeriodicalId\":100626,\"journal\":{\"name\":\"IEEE LTS\",\"volume\":\"3 3\",\"pages\":\"12-19\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"1992-08-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://sci-hub-pdf.com/10.1109/80.166999\",\"citationCount\":\"10\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"IEEE LTS\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://ieeexplore.ieee.org/document/166999/\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE LTS","FirstCategoryId":"1085","ListUrlMain":"https://ieeexplore.ieee.org/document/166999/","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Photonic interconnects for gigabit multicomputer communications
The ways in which photonic and optoelectronic technologies could play an important role in future highly scalable and flexible interconnects for multicomputer parallel processing systems are discussed. For electronic interconnect implementation, the primary limitations arise from transmission drive power requirements, limited bandwidth, and the crosstalk-limited length. It is shown that photonic interconnects can relieve these bottlenecks in order to allow systems to scale to large numbers of nodes without degrading the interconnect performance. As an example, a network architecture capable of interconnecting thousands of processors with multigigabit average access rate per user, and peak access rates an order of magnitude higher is presented. The network topology is a shuffle-exchange, multihop, multipath, wraparound direct interconnect that utilizes self-routing and a deflection flow control technique to simplify and speed the processing. An experimental 2*2 photonic switching node based on the presented techniques is described.<
>