MoCSYS: A Multi-Clock Hybrid Two-Layer Router Architecture and Integrated Topology Synthesis Framework for System-Level Design of FPGA Based On-Chip Networks
{"title":"MoCSYS: A Multi-Clock Hybrid Two-Layer Router Architecture and Integrated Topology Synthesis Framework for System-Level Design of FPGA Based On-Chip Networks","authors":"A. Janarthanan, K. Tomko","doi":"10.1109/VLSI.2008.79","DOIUrl":null,"url":null,"abstract":"Complex system-on-chip designs targeted for FPGAs merit sophisticated communication architectures to support a host of high performance applications. In this research we implement a hybrid two-layer router architecture for FPGA based NoCs and quantify its area and performance tradeoffs by characterizing a network component library (Mo-Clib). Results from the VHDL and SystemC models of the advanced router architecture show an average improvement of 20.4% in NoC bandwidth (maximum of 24% compared to a traditional NoC). As a part of the CAD flow, we develop an algorithm that utilizes the above NoC framework and includes bandwidth capacity and area as a cost during an automatic NoC topology synthesis phase. Experimental results for a set of real applications and synthetic benchmarks show an average reduction of 21.6% in FPGA area (maximum of 26%) for equivalent bandwidth constraints when compared with a baseline approach.","PeriodicalId":143886,"journal":{"name":"21st International Conference on VLSI Design (VLSID 2008)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2008-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"6","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"21st International Conference on VLSI Design (VLSID 2008)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/VLSI.2008.79","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 6
Abstract
Complex system-on-chip designs targeted for FPGAs merit sophisticated communication architectures to support a host of high performance applications. In this research we implement a hybrid two-layer router architecture for FPGA based NoCs and quantify its area and performance tradeoffs by characterizing a network component library (Mo-Clib). Results from the VHDL and SystemC models of the advanced router architecture show an average improvement of 20.4% in NoC bandwidth (maximum of 24% compared to a traditional NoC). As a part of the CAD flow, we develop an algorithm that utilizes the above NoC framework and includes bandwidth capacity and area as a cost during an automatic NoC topology synthesis phase. Experimental results for a set of real applications and synthetic benchmarks show an average reduction of 21.6% in FPGA area (maximum of 26%) for equivalent bandwidth constraints when compared with a baseline approach.