GPUMLib An Efficient Open-Source GPU

GPUMLib An Efficient Open-Source GPU

ID:39748171

大小:725.16 KB

页数:8页

时间:2019-07-10

GPUMLib An Efficient Open-Source GPU_第1页
GPUMLib An Efficient Open-Source GPU_第2页
GPUMLib An Efficient Open-Source GPU_第3页
GPUMLib An Efficient Open-Source GPU_第4页
GPUMLib An Efficient Open-Source GPU_第5页
资源描述:

《GPUMLib An Efficient Open-Source GPU》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、InternationalJournalofComputerInformationSystemsandIndustrialManagementApplicationsISSN2150-7988Volume3(2011)pp.355-362©MIRLabs,www.mirlabs.net/ijcisim/index.htmlGPUMLib:AnEfficientOpen-SourceGPUMachineLearningLibraryNoelLopes1andBernardeteRibeiro21UDI,PolytechnicInstituteofGuarda,P

2、ortugalCISUC,UniversityofCoimbra,Portugalnoel@ipg.pt2DepartmentofInformaticsEngineeringCISUC,UniversityofCoimbra,Portugalbribeiro@dei.uc.ptAbstract:GraphicsProcessingUnits(GPUs)placedatourdis-ciatedwithMLproblemsincreases,thetrendistohavemoreposalanunprecedentedcomputational-power,

3、largelysurpass-challengingandcomputationallydemandingproblemsthatingtheperformanceofcutting-edgeCPUs(CentralProcess-canbecomeintractablefortraditionalCPU(CentralProcess-ingUnits).Thehigh-parallelisminherenttotheGPUmakesingUnit)architectures.Therefore,thepressuretoshiftdevel-thisdev

4、iceespeciallywell-suitedtoaddressMachineLearn-opmenttowardsparallelarchitectureswithhigh-throughputing(ML)problemswithprohibitivelycomputationalintensivehasbeenaccentuated.Inthiscontext,theGraphicsProcess-tasks.Nevertheless,fewMLalgorithmshavebeenimplementedingUnit(GPU)representsac

5、ompellingsolutiontoaddressontheGPUandmostarenotopenlyshared,posingdifficul-theincreasingneedsofcomputationalperformance,inpar-tiesforresearchersandengineersaimingtodevelopGPU-basedticularintheMLfield.systems.Tomitigatethisproblem,weproposethecreationofanInthelasteightyearstheperforma

6、nceandcapabilitiesofopensourceGPUMachineLearningLibrary(GPUMLib)thattheGPUshavebeensignificantlyaugmentedandtodaysaimstoprovidethebuildingblocksforthedevelopmentofef-GPUs,includedinmainstreamcomputingsystems,arepow-ficientGPUMLsoftware.Experimentalresultsonbenchmarkerful,highlyparall

7、elandprogrammabledevicesthatcanbedatasetsshowthatthealgorithmsalreadyimplementedyieldusedforgeneral-purposecomputingapplications[2].SincesignificanttimesavingsovertheCPUcounterparts.GPUsaredesignedforhigh-performancerenderingwhereKeywords:GPUComputing,machinelearningalgorithms.repea

8、tedoperationsarecommon,the

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。