coa復(fù)習(xí)第十三章簡化版

WilliamStallings

ComputerOrganization

andArchitecture

6thEditionChapter13ReducedInstructionSetComputers(RISC)

精簡指令集計算機(jī)MajorAdvancesinComputers(1)Thefamilyconcept

系列概念I(lǐng)BMSystem/3601964DECPDP-8SeparatesarchitecturefromimplementationMicroporgrammedcontrolunit

微程序式控制器IdeabyWilkes1951ProducedbyIBMS/3601964CachememoryCache

存儲器IBMS/360model851969Page462MajorAdvancesinComputers(2)SolidStateRAM靜態(tài)RAM(Seememorynotes)Microprocessors微處理器Intel40041971Pipelining流水線IntroducesparallelismintofetchexecutecycleMultipleprocessors多個處理器Page462TheNextStep(趨勢)-RISCReducedInstructionSetComputerKeyfeaturesLargenumberofgeneralpurposeregistersoruseofcompilertechnologytooptimizeregisteruseLimitedandsimpleinstructionsetEmphasisonoptimisingtheinstructionpipelinePage463DrivingforceforCISCSoftwarecostsfarexceedhardwarecostsIncreasinglycomplexhighlevellanguagesSemanticgap語義間隙ThedifferencebetweentheoperationsprovidedinHLLsandthoseprovidedincomputerarchitecture.Leadsto:LargeinstructionsetsMoreaddressingmodesHardwareimplementationsofHLLstatementse.g.CASE(switch)onVAXPage464IntentionofCISCEasecompilerwritingImproveexecutionefficiencyComplexoperationsinmicrocodeSupportmorecomplexHLLsPage464§13.1ExecutionCharacteristicsOperationsperformedOperandsusedExecutionsequencing

—StudieshavebeendonebasedonprogramswritteninHLLs

—DynamicstudiesaremeasuredduringtheexecutionoftheprogramPage4641.Operations Dynamic MachineInstruction MemoryReference Occurrence (Weighted) (Weighted) Pascal C(%) Pascal C(%) Pascal C(%) Assign 45 38 13 13 14 15 Loop 5 3 42 32 33 26

Call 15 12 31 33 44 45 If 29 43 11 21 7 13 GoTo - 3 - - - - Other 6 1 3 1 2 1Page465賦值循環(huán)調(diào)用判斷直接轉(zhuǎn)移其他—Procedurecall-returnisverytimeconsuming2.OperandsMainlylocalscalar(標(biāo)量)

variablesOptimizationshouldconcentrateonaccessinglocalvariables Pascal(%)C(%)Average(%) Integerconstant 16 23 20

Scalarvariable 58 53 55 Array/structure 26 24 25Page466ProcedureCallsVerytimeconsuming

—Dependsonnumberofparameterspassed—DependsonlevelofnestingMostprogramsdonotdoalotofcallsfollowedbylotsofreturnsMostvariablesarelocal(c.f.localityofreference)Page466ImplicationsBestsupportisgivenbyoptimisingmostusedandmosttimeconsumingfeatures=>LargenumberofregistersOperandreferencingCarefuldesignofpipelinesBranchpredictionetc.=>Simplified(reduced)instructionset(RISC)Page467RISC--AlargenumberofRegistersRegistersforLocalVariables

—Storelocalscalarvariablesinregisters

—ReducesmemoryaccessOverlappingRegisterWindows

—CompleteprocedureCALL&Return—Transferparameters&savereturnaddress=>§13.2LargeRegisterFilePage467大寄存器方案※RISCCharacteristics*OneinstructionpercycleRegistertoregisteroperationssimpleaddressingmodessimpleinstructionformatsPage476RISCvCISCNow,notclearcut劃分不清楚Manydesignsborrowfrombothphilosophies

相互借鑒e.g.PowerPCandPentiumIIPage479TypicalofaclassicRISC(1)1.Asingleinstructionsize2.Thesizeistypically4bytes3.Asmallnumberofdataaddressingmode(lessthan5)4.Noindirectaddressing5.Nooperationsthatcombineload/storewitharithmaticPage480TypicalofaclassicRISC(2)6.Nomorethanonememory-addressedoperandperinstruction…….ptimizationofPipeliningDelayedbranch延時轉(zhuǎn)移DoesnottakeeffectuntilafterexecutionoffollowinginstructionThisfollowinginstructionisthedelayslotPage484NormalandDelayedBranchAddress Normal Delayed Optimized100 LOADX,A LOADX,A LOADX,A101 ADD1,A

ADD1,A

JUMP105102 JUMP105

JUMP106

ADD