{"title":"芯片多处理器对外部存储器的公平访问","authors":"Shufan Yang, Qiang Wu, Xiongren Xiao, Renfa Li, Dominic Hillenbrand","doi":"10.1109/IPDPSW.2012.36","DOIUrl":null,"url":null,"abstract":"A memory arbitration scheme is required for chip-multiprocessor accessing external (off-chip) memory. A round-robin arbitration is an efficient and low-cost candidate for chip-multiprocessor. However, the adoption of simple round robin scheme raises a potential problem of uneven sharing of external memory bandwidth. One way to manage an unbalanced resource allocation resulting from the use of arbiters is to use priority arbitrations combined with a look-up table, but this is not economical since large lookup tables require a large silicon area. We propose an admission control scheme operating at the edge of on-chip interconnection. The fair-sharing service is guaranteed using a token mechanism to schedule packets onto fabric. This work shows that a fair service problem can be solved by controlling traffic of on-chip interconnect, since it avoid saturation in critical paths and to maintain equilibrium in the allocation of resource.","PeriodicalId":378335,"journal":{"name":"2012 IEEE 26th International Parallel and Distributed Processing Symposium Workshops & PhD Forum","volume":"2 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2012-05-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Fair Access to External Memory for Chip-multiprocessor\",\"authors\":\"Shufan Yang, Qiang Wu, Xiongren Xiao, Renfa Li, Dominic Hillenbrand\",\"doi\":\"10.1109/IPDPSW.2012.36\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"A memory arbitration scheme is required for chip-multiprocessor accessing external (off-chip) memory. A round-robin arbitration is an efficient and low-cost candidate for chip-multiprocessor. However, the adoption of simple round robin scheme raises a potential problem of uneven sharing of external memory bandwidth. One way to manage an unbalanced resource allocation resulting from the use of arbiters is to use priority arbitrations combined with a look-up table, but this is not economical since large lookup tables require a large silicon area. We propose an admission control scheme operating at the edge of on-chip interconnection. The fair-sharing service is guaranteed using a token mechanism to schedule packets onto fabric. This work shows that a fair service problem can be solved by controlling traffic of on-chip interconnect, since it avoid saturation in critical paths and to maintain equilibrium in the allocation of resource.\",\"PeriodicalId\":378335,\"journal\":{\"name\":\"2012 IEEE 26th International Parallel and Distributed Processing Symposium Workshops & PhD Forum\",\"volume\":\"2 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2012-05-21\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2012 IEEE 26th International Parallel and Distributed Processing Symposium Workshops & PhD Forum\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/IPDPSW.2012.36\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2012 IEEE 26th International Parallel and Distributed Processing Symposium Workshops & PhD Forum","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/IPDPSW.2012.36","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Fair Access to External Memory for Chip-multiprocessor
A memory arbitration scheme is required for chip-multiprocessor accessing external (off-chip) memory. A round-robin arbitration is an efficient and low-cost candidate for chip-multiprocessor. However, the adoption of simple round robin scheme raises a potential problem of uneven sharing of external memory bandwidth. One way to manage an unbalanced resource allocation resulting from the use of arbiters is to use priority arbitrations combined with a look-up table, but this is not economical since large lookup tables require a large silicon area. We propose an admission control scheme operating at the edge of on-chip interconnection. The fair-sharing service is guaranteed using a token mechanism to schedule packets onto fabric. This work shows that a fair service problem can be solved by controlling traffic of on-chip interconnect, since it avoid saturation in critical paths and to maintain equilibrium in the allocation of resource.