download0 view68
twitter facebook

공공누리This item is licensed Korea Open Government License

Title
Reducing Communication Costs in Collective I/O in Multi-core Cluster Systems with Non-exclusive Scheduling
Author(s)
차광호
Publication Year
2012-09-03
Abstract
Asthenumberofnodesinhighperformancecomputing(HPC)systemsincreases,collectiveI/ObecomesanimportantissueandI/OaggregatorsarethekeyfactorsinimprovingtheperformanceofcollectiveI/O.WhenanHPCsystemusesnon-exclusivescheduling,adifferentnumberofCPUcorespernodecanbeassignedforMPIjobsthus,I/Oaggregatorsexperienceadisparityintheirworkloadsandcommunicationcosts.BecausethecommunicationbehaviorsareinfluencedbythesequenceoftheI/OaggregatorsandbythenumberofCPUcoresinneighbornodes,changingtheorderofthenodesaffectsthecommunicationcostsincollectiveI/O.Therearefewstudies,however,thatseektoincorporatestepstoadequatelydeterminethenodesequence.Inthisstudy,itwasfoundthataninappropriateorderofnodesresultsinanincreaseinthecollectiveI/Ocommunicationcosts.Inordertoaddressthisproblem,weproposetheuseofspecificheuristicmethodstoregulatethenodesequence.WealsodevelopapredictionfunctioninordertoestimatetheMPI-IOperformancewhenusingtheproposedheuristicfunctions.TheperformancemeasurementsindicatedthattheproposedschemeachieveditsgoalofpreventingtheperformancedegradationofthecollectiveI/Oprocess.Forinstance,inamulti-coreclustersystemwiththeLustrefilesystem,thereadbandwidthofMPI-Tile-IOwasimprovedby7.61%to17.21%andthewritebandwidthofthebenchmarkwasalsoincreasedby17.05%to26.49%.
Keyword
CollectiveI/O; Clustersystem; MPI-IO; ParallelI/O
Journal Title
TheJournalofsupercomputing
Citation Volume
61
ISSN
0920-8542
Files in This Item:
There are no files associated with this item.
Appears in Collections:
7. KISTI 연구성과 > 학술지 발표논문
URI
https://repository.kisti.re.kr/handle/10580/14063
http://www.ndsl.kr/ndsl/search/detail/article/articleSearchResultDetail.do?cn=NART63794291
Export
RIS (EndNote)
XLS (Excel)
XML

Browse