R. C. G. N. Ewo, A. Pinna, B. Granado, M. Mbouenda, H. Fotsin
{"title":"A Hardware MPI Spawn for Distributed Multiprocessing Reconfigurable System on Chip (MP-RSoC)","authors":"R. C. G. N. Ewo, A. Pinna, B. Granado, M. Mbouenda, H. Fotsin","doi":"10.1109/FCCM.2014.73","DOIUrl":null,"url":null,"abstract":"In this paper we describe a hardware implementation of the MPI Spawn function of MPI-2 Remote Memory Access (RMA) communication library primitive, devoted to a distributed Multi Processing Reconfigurable System on Chip (MP-RSoC). This function enhances the MPI Hardware Communication Library (MPI-HCL) we realized in previous work. Designers can activate or deactivate hardware tasks on runtime, using MPI functions in MPRSoC environment. The advantages are more scalability, efficient power consumption and easier deployment of the parallel application. Our hardware primitives have been implemented and tested on a Xilinx Spartan6 FPGA board.","PeriodicalId":93352,"journal":{"name":"Proceedings ... Annual IEEE Symposium on Field-Programmable Custom Computing Machines. FCCM (Symposium)","volume":"7 6","pages":"238"},"PeriodicalIF":0.0000,"publicationDate":"2013-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings ... Annual IEEE Symposium on Field-Programmable Custom Computing Machines. FCCM (Symposium)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/FCCM.2014.73","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 2
Abstract
In this paper we describe a hardware implementation of the MPI Spawn function of MPI-2 Remote Memory Access (RMA) communication library primitive, devoted to a distributed Multi Processing Reconfigurable System on Chip (MP-RSoC). This function enhances the MPI Hardware Communication Library (MPI-HCL) we realized in previous work. Designers can activate or deactivate hardware tasks on runtime, using MPI functions in MPRSoC environment. The advantages are more scalability, efficient power consumption and easier deployment of the parallel application. Our hardware primitives have been implemented and tested on a Xilinx Spartan6 FPGA board.