当前位置:文库下载 > 所有分类 > Empirical Evaluation of Distributed Mutual Exclusion Algorithms
免费下载此文档侵权投诉

Empirical Evaluation of Distributed Mutual Exclusion Algorithms

In this paper, we evaluated various distributed mutual exclusion algorithms on the IBM SP2 machine and the Intel iPSC/860 system. The empirical results are compared in terms of such criteria as the number of message exchanges and the response time. Our res

EmpiricalEvaluationofDistributedMutualExclusionAlgorithms

ShiwaS.Fu,Nian-FengTzeng,andZhiyuanLi

CenterforAdvancedComputerStudiesUniversityofSouthwesternLouisiana

Lafayette,LA70504

Abstract

exclusionInthisiPSC/860algorithmspaper,weonevaluatedtheIBMvariousSP2distributedmutualtermsandofsuchsystem.TheempiricalresultsmachinearecomparedandtheIntelinalgorithmtheresponsecriteriatime.asOurthenumberresultsofindicatemessagethatexchangescasessystem,among[2]allachievesthealgorithmstheshortestonaresponsesmalltomediumtimetheinsizedmostStarmanywhenprocessorsrequestforthecriticalsectionOnsectionthetimesotherbeforehand,involvingifeveryanyprocessorbarrierenterssynchronization.theprovedunderRingonlyoncealgorithmbefore[4]encounteringisabarrier,criticaltheim-algorithmaheavyload;buttheStarfoundalgorithmtooutperformandtheothersCSLTheorybestsolution[3]prevailtomutualwhentheexclusionrequestinratebecomeslight.eratesystemstheirmutualisdeterminedexclusionbyrequests.

howparticipatingdistributedsitesmem-gen-1Introduction

suresMutualofsivecodeinvolvedexclusion(calledsitesisachievedbyamechanismthaten-thecriticaltogetaccesssectionto)inadesignatedasectiondistributedway.Thechines,numbercomparingmutualfocusexclusionofthispaperisonevaluatingmutuallycurrentexclu-theirbehaviorsalgorithmsintermsontwooftherealmeanma-areNMEofmessagesexchangedtheresponsepercriticalsectionentryNeilsencompared,(calledandMizuno’sincludingalgorithmtheRaymondtime.Fouralgorithms[2]algorithm[1],the[4],inofshort)andthetheStar[3].Chang,algorithm),theimprovedwithRingstaralgorithmtopologyOurimprovedSinghal,andRingLiu’salgorithmalgorithmisa(i.e.,CSLformancethatdescribedasduetotheearlierin[5]butexhibitsimprovedvariationper-evaluatedetailedinSec.2.4.eliminationInaddition,ofunnecessarywealsointroducemessages,andtobetteraperformancemodi edRaymondthanthealgorithm,Raymondwhichalgorithm.

givesrisegorithms,InorderPOWERparallelwetoimplementedobservethethemactualonbehaviorstheoftheseal-Weupthetocarried64,andoutonourSystem2(SP2)andtheIBM’sInteliPSCScalable/860.thestudyInteloniPSCtheIBMSP2machineofsizedistributed-memorylargestsubcubeavailabletousers)./860ofOursizeresults16(whichonisMizuno’srithmswithStarrespectalgorithmmachinestothe[2]suggestresponseoutperformsthatthetimeforallNeilsenbothmostotherandcases,

algo-inpartbytheNSFunderGrantsMIP-9201308andCCR-9300075.WewishtothankArgonneNationalLaboratoryandERCforComputationalFieldSimulation,Mis-sissippiStateUniversity,forallowingouraccesstotheIBMSP2machineandtheInteliPSC/860system,respectively.TheIBMSP2machinebe-longstotheArgonneHigh-PerformanceComputingResearchFacility,whichisfundedprincipallybytheU.S.DepartmentofEnergy,Mathe-matical,InformationandComputationalSciencesDivision(ER-31).

whenedlyduethetoandthecriticalthenofollowingbarriersectionistwoinvolvedisrequestedbyprocessorsrepeat-facts:(1)inthethemeantime.StaralgorithmThisisandlowestinitially,(2)whileNMEallprocessorsunlesstherequestcontendrateforistheextremelycriticalsectionhigh,hasgradually,fewertheasandfewerprocessorsexperiencecontentiondoescriticalsectiononlyoneataprocessortimeinsequenceispermittedtoenter/leaveearliernotsorsrequestgeneratehasanothercriticalsectionandrequestaprocessoruntilitstionnojustbarrierrequestsgradually“serialize”beenserved.theirgenerationConsequently,ofcriticaltheproces-sec-isinvolvedandsooninavoidthemeantime.mostcontention,aslongasbarrieronerequesttothecriticalsectionbeforeIfeveryinvolvingsiteissuesinhibitsStarthesynchronization,bestperformanceourforimprovedahighrequestRingalgorithmrate;ex-lowandrequesttherate.

CSLalgorithmsaresuperiortoothersbutfortheaJohnsonAsimulationscribed[6]tocontraststudymutualwasperformedexclusionalgorithmspreviouslybysimulationin[1,algorithmcally,[2],study2,3],mainlyonthebasisofNME.Thatde-withexamineditslogicaladifferentstructureconstructchangedofdynami-theStarthealgorithmrootasinopposedtoa xedstructurewitha xedsiteaschangedisour xed,implementation.itrequiresnoWhenmorethantherootoftheStarmanceically.thanperacriticalconstructsectionentry,exhibiting3messagesbetterperfor-ex-studyintroducesisThelimitedRaymondalgorithmwithitsstructurestudiedinchangedthatsimulationdynam-paresamodi edtoitsversionoriginalofform,thebutourworkherepresenttheempiricalthebehaviorsresultsofofourbothimprovedRaymondalgorithmRingalgorithms.andalgorithm.Wecom-alsoOurthantheearlierevaluationsimulationgivesstudyrisetoperformedadifferentinconclusion[6]does.

2DistributedMutualExclusionAlgorithms

videdDistributedmutualwhereintoexclusionalgorithmscanbedi-ceivesalltwo(2)theinvolvedclasses:(1)permission-basedalgorithms,permissionsitestovoteentertotheselectcriticalonesection,whichre-thetoken-basedpermission-basedtokenmayenteralgorithms,theinwhichonlythesitewithandtionthereforetraf coverheadalgorithmcriticalsection.Ingeneral,athanainvolveshighercommunica-ourreviewedempiricalfocusbrie ystudy.ourattentiontoken-basedalgorithm.Weinsequence.

Thealgorithmstotoken-basedunderthisalgorithmsstudyarein2.1RaymondAlgorithm

tainsTheRaymondalgorithmexample,astatic[1]determinesandmain-lifetime,dynamicallybutaspanninglogicalstructure.Thelogicalstructure(forthedirectionstree)isofkeptedgesunchangedinthethroughoutitspointedgestowardasinthestructurethethepossibletokenmigratesamongstructuresites,inorderchangetoalwaystokenpointholder.totheThepossibledirectionstoken

of

第1页

免费下载Word文档免费下载:Empirical Evaluation of Distributed Mutual Exclusion Algorithms

(下载1-5页,共5页)

我要评论

返回顶部