简介：Theprocessingspeedofthecommunicationbetweennodesinaparallelprocessorhasbecomethemajorbottleneckoftheprocessor'sperformance.RDMA(RemoteDirectMemoryAccess)technologyhasdrawnmoreattentionrecentlyduetoitscapabilityoftransferringalargeramountofdata,higherspeedandreliability.4DSP(4DigitalSignalProcessing)modulecomprisedofTiger-SHARC201chipisconnectedbyLVDS(LowVoltageDifferentialSignal)circuits.ThispaperproposesageneralandreconfigurableRDMAplatformanditscorrespondingcommunicationprotocolwithalltherouteslinkedbasedonthezerocopy.TheprotocoltransfersmessageofDSPbyinterruptingofDMAandisappliedonmassiveremoteimageimpression,whichreducesmemoryneedsandworkingburdenofCPU.Theexperimentresultsshowthisplatformisefficient,flexible,andexpandableofbeingintegratedtoalargerscaleinthenextdevelopmentstages.

简介：摘要：基于道路环境数据的实时动态更新策略是一种通过获取全局道路环境信息和局部道路环境信息，将它们进行融合，并实时更新车辆的行驶策略的方法。该策略可以提高车辆的行驶效率和安全性，是未来智能交通系统的重要支持。主要步骤包括数据预处理、数据融合、实时动态更新和反馈控制。其中，数据融合是关键步骤，可以采用多种融合方法。该策略的优点是可以实时获取道路环境信息，并做出相应的调整，以适应不同的交通情况。

简介：Themulticoreevolutionhasstimulatedrenewedinterestsinscalingupapplicationsonshared-memorymultiprocessors,significantlyimprovingthescalabilityofmanyapplications.Butthescalabilityislimitedwithinasinglenode;thereforeprogrammersstillhavetoredesignapplicationstoscaleoutovermultiplenodes.Thispaperrevisitsthedesignandimplementationofdistributedsharedmemory(DSM)asawaytoscaleoutapplicationsoptimizedfornon-uniformmemoryaccess(NUMA)architectureoverawell-connectedcluster.ThispaperpresentsMAGI,anefficientDSMsystemthatprovidesatransparentsharedaddressspacewithscalableperformanceonaclusterwithfastnetworkinterfaces.MAGIisuniqueinthatitpresentsaNUMAabstractiontofullyharnessthemulticoreresourcesineachnodethroughhierarchicalsynchronizationandmemorymanagement.MAGIalsoexploitsthememoryaccesspatternsofbig-dataapplicationsandleveragesasetofoptimizationsforremotedirectmemoryaccess(RDMA)toreducethenumberofpagefaultsandthecostofthecoherenceprotocol.MAGIhasbeenimplementedasauser-spacelibrarywithpthread-compatibleinterfacesandcanrunexistingmultithreadedapplicationswithminimizedmodifications.WedeployedMAGIoveran8-nodeRDMA-enabledcluster.ExperimentalevaluationshowsthatMAGIachievesupto9.25:4speedupcomparedwithanunoptimizedimplementation,leadingtoasealableperformanceforlarge-scaledata-intensiveapplications.

简介：摘要：在网络的普及下社会发展和人们的生活都得到了更为便捷化的服务，对数据中心RDMA网络的应用需求也大幅增加，如何有效的满足在远程直接数据存取方面的需求，解决网络数据传输延迟问题成为了亟待完成的任务，对此拥塞控制算法和拥塞控制器设计有着针对性的作用。基于此，本文将对面向数据中心RDMA网络的拥塞控制算法研究及控制器设计展开研究。