Huaxi Gu, Xiaoshan Yu, Yunfeng Lu, Hong Zou, Shuo Li
{"title":"X-NEST+:用于分布式机器学习和高性能计算的高带宽和可重构光互连","authors":"Huaxi Gu, Xiaoshan Yu, Yunfeng Lu, Hong Zou, Shuo Li","doi":"10.23919/OFC49934.2023.10117103","DOIUrl":null,"url":null,"abstract":"We propose X-NEST+, a scalable and high-bandwidth optical interconnects capable of reconfiguring both intra- and inter- cluster topology based on traffic demands. The experiment results indicate up to 8%~36% reduction in completion time for HPC and ML applications compared with Helios and RotorNet.","PeriodicalId":355445,"journal":{"name":"2023 Optical Fiber Communications Conference and Exhibition (OFC)","volume":"156 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"X-NEST+: A High Bandwidth and Reconfigurable Optical Interconnects for Distributed Machine Learning and High-Performance Computing\",\"authors\":\"Huaxi Gu, Xiaoshan Yu, Yunfeng Lu, Hong Zou, Shuo Li\",\"doi\":\"10.23919/OFC49934.2023.10117103\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"We propose X-NEST+, a scalable and high-bandwidth optical interconnects capable of reconfiguring both intra- and inter- cluster topology based on traffic demands. The experiment results indicate up to 8%~36% reduction in completion time for HPC and ML applications compared with Helios and RotorNet.\",\"PeriodicalId\":355445,\"journal\":{\"name\":\"2023 Optical Fiber Communications Conference and Exhibition (OFC)\",\"volume\":\"156 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2023-03-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2023 Optical Fiber Communications Conference and Exhibition (OFC)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.23919/OFC49934.2023.10117103\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2023 Optical Fiber Communications Conference and Exhibition (OFC)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.23919/OFC49934.2023.10117103","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
X-NEST+: A High Bandwidth and Reconfigurable Optical Interconnects for Distributed Machine Learning and High-Performance Computing
We propose X-NEST+, a scalable and high-bandwidth optical interconnects capable of reconfiguring both intra- and inter- cluster topology based on traffic demands. The experiment results indicate up to 8%~36% reduction in completion time for HPC and ML applications compared with Helios and RotorNet.