ggpu,zxp,joycy,qzy￥ Computer aided hardware/software partitioning is one of the key challenges in hardware/software co-design. This paper describes a new approach to hardware/software partitioning for multiple hardware-devices model. The partitioning is t

AnApproachtoHardware/SoftwarePartitioning

forMultipleHardwareDevicesModel

PuGeguang,ZhaoXiangpeng,WangShuling,QiuZongyan,HeJifeng,andWangYi

LMAMandDepartmentofInformatics,SchoolofMathematicalSciences,

PekingUniversity,Beijing,P.R.China

InternationalInstituteforSoftwareTechnologyUnitedNationsUniversity,Macau.

DepartmentofComputerSystems,UppsalaUniversity,Sweden.

ggpu,zxp,joycy,qzy@math.pku.edu.cn,hjf@iist.unu.edu,yi@docs.uu.se

Abstract

Computeraidedhardware/softwarepartitioningisoneofthekeychallengesinhardware/softwareco-design.Thispaperdescribesanewapproachtohardware/softwarepar-titioningformultiplehardware-devicesmodel.Theparti-tioningistransformedintoareachabilityproblemoftimedautomata,andtheoptimalsolutioncanbeobtainedbymeansofanoptimalreachibilityalgorithm.Torelaxtheinitialconditionofpartitioningforoptimization,twoal-gorithmsaredesignedtoexplorethedependencyrelationsamongprocessesinthesequentialspeci cation.Someex-perimentsareconductedwithmodelcheckerUPPAALtoshowourapproachisbotheffectiveandef cient.

1Introduction

Oneoftheforemostproblemsinhardware/softwareco-design[23]isHardware/SoftwarePartitioning.Thegoalofhardware/softwarepartitioningistode netheboundaryofthehardwareandsoftwareimplementationofthesys-temspeci cation.Essentially,thepartitioningprocessistosearchareasonablecompositionofhardwareandsoftwarecomponentswhichnotonlysatis estheconstraintssuchastiming,butalsooptimizeddesiredqualitymetrics,suchascommunicationcost,powerconsumptionandsoon.

Severalapproachesbasedonalgorithmshavebeende-veloped,asdescribed,forexample,in[1,16,17,22,24].Alltheseapproachesemphasizethealgorithmicaspects,forinstance,integerprogramming[16,24],evolutional-gorithm[22]andsimulatedannealingalgorithm[17]arerespectivelyintroducedtothepartitioningprocessinthe