首页 Integrated_Circuits(集成电路)电子信息类专业英语、计算机类专业英语

Integrated_Circuits(集成电路)电子信息类专业英语、计算机类专业英语

举报
开通vip

Integrated_Circuits(集成电路)电子信息类专业英语、计算机类专业英语Integrated Circuits(集成电路)TheIntegratedCircuit  Digitallogicand electroniccircuitsderive theirfunctionalityfromelectronicswitchescalledtransistor.Roughlyspeaking, thetransistorcanbelikenedto anelectronicallycontrolledvalvewherebyenergyappliedtoone connectionoft...

Integrated_Circuits(集成电路)电子信息类专业英语、计算机类专业英语
Integrated Circuits(集成电路)TheIntegratedCircuit  Digitallogicand electroniccircuitsderive theirfunctionalityfromelectronicswitchescalledtransistor.Roughlyspeaking, thetransistorcanbelikenedto anelectronicallycontrolledvalvewherebyenergyappliedtoone connectionofthe valveenablesenergytoflowbetweentwootherconnections.By combiningmultiple transistors,digitallogicbuildingblockssuchas ANDgatesandflip-flopsareformed.Transistors,inturn,aremadefromsemiconductors.Consulta periodictableofelementsinacollegechemistrytextbook, andyouwill locatesemiconductorsasagroupofelementsseparatingthemetalsand nonmetals.Theyare calledsemiconductorsbecause of theirabilitytobehaveas bothmetals and nonmetals. Asemiconductorcanbemade toconduct electricitylikea metal orto insulate asanonmetal does.Thesedifferingelectricalproperties canbeaccurately controlledby mixingthesemiconductorwithsmallamountsofotherelements.Thismixingiscalleddoping.A semiconductorcanbedopedtocontain moreelectrons (N-type)orfewerelectrons(P-type).Examplesofcommonlyused semiconductorsaresiliconandgermanium.Phosphorousand boronare twoelementsthatareusedtodopeN-type andP-type silicon,respectively.  Atransistorisconstructedby creatingasandwichofdifferentlydopedsemiconductorlayers. The twomost commontypes of transistors, thebipolar-junction transistor(BJT)andthefield-effecttransistor(FET)are schematicallyillustratedinFigure2.1.Thisfigureshowsboththesiliconstructuresoftheseelementsandtheirgraphicalsymbolicrepresentation aswouldbeseenin acircuit diagram.TheBJTshown isan NPNtransistor,becauseitiscomposed of asandwich ofN-P-Ndopedsilicon.Whenasmall currentisinjectedintothe baseterminal,alargercurrentis enabled toflowfrom thecollector totheemitter.The FETshownisanN-channel FET,whichiscomposed oftwoN-typeregionsseparated by aP-typesubstrate. Whenavoltageis applied totheinsulatedgateterminal, a currentisenabledto flowfromthe draintothesource.It iscalled N-channel, because thegatevoltageinducesan N-channel withinthe substrate, enablingcurrenttoflowbetween theN-regions. Another basicsemiconductorstructureisadiode,whichisformedsimplybya junctionofN-typeandP-type silicon.Diodesactlikeone-wayvalvesby conductingcurrentonly from PtoN.Specialdiodes canbecreatedthatemitlightwhena voltageisapplied.Appropriately enough,thesecomponentsare calledlightemittingdiodes,orLEDs. Thesesmall lightsare manufacturedbythemillionsandarefound indiverseapplications fromtelephones totraffic lights. The resulting smallchipofsemiconductormaterialonwhich atransistor ordiode isfabricatedcanbe encased in a small plastic packageforprotectionagainstdamageandcontaminationfromtheout­sideworld.Smallwiresareconnectedwithinthispackagebetween thesemiconductorsandwichandpinsthatprotrudefromthepackagetomakeelectricalcontactwith otherpartsoftheintendedcircuit.Onceyouhaveseveraldiscrete transistors,digital logiccan be builtbydirectlywiringthesecomponentstogether.Thecircuitwillfunction, butanysubstantial amountofdigitallogicwill beverybulky, because severaltransistorsare requiredtoimplementeachofthevarioustypesof logicgates.  Atthetime oftheinvention ofthetransistorin 1947 byJohnBardeen,WalterBrattain,andWilliamShockley,theonlywayto assemblemultipletransistorsintoasingle circuitwas to buyseparate discretetransistorsandwirethemtogether. In1959,JackKilby andRobert Noyce independently inventedameansoffabricatingmultipletransistors onasingleslabof semiconductormaterial.Theirinvention wouldcometobeknownastheintegratedcircuit, orIC, whichis thefoundation of ourmodern computerizedworld.AnICissocalledbecauseit integratesmultipletransistorsanddiodesonto thesamesmallsemiconductorchip.Instead ofhavingtosolderindividualwiresbetweendiscretecomponents,anIC containsmanysmallcomponents thatarealreadywiredtogetherin thedesired topologyto formacircuit.A typicalIC,withoutitsplastic or ceramicpackage,isasquare orrectangular silicondiemeasuringfrom2to15mmonanedge.Dependingonthelevel of technologyusedtomanufacture theIC,theremay beanywherefromadozentotensofmillions ofindividualtransistors onthissmallchip.Thisamazingdensity ofelectroniccomponentsindicatesthatthetransistorsandthewiresthatconnect themareextremely smallinsize. Dimensionsonan ICare measuredinunits ofmicrometers, with one micrometer(1mm)being onemillionthofameter.To serveas areferencepoint,ahumanhairis roughly100mm indiameter.SomemodernICs contain componentsandwiresthatare measuredinincrementsassmallas0.1mm!Eachyear,researchersandengineershavebeen findingnewways tosteadilyreducethesefeaturesizes topackmoretransistorsintothesamesiliconarea,asindicatedin Figure2.2.  WhenanICisdesignedand fabricated,it generallyfollows oneoftwo main transistor technologies:bipolarormetal-oxidesemiconductor(MOS).BipolarprocessescreateBJTs,whereasMOS processescreate FETs.Bipolar logicwas more commonbeforethe1980s,butMOStechnologieshavesinceaccountedthegreatmajority ofdigitallogic ICs.N-channel FETsare fabricatedinanNMOSprocess,andP-channelFETsarefabricated inaPMOS process.Inthe1980s,complementary-MOS,or CMOS,becamethedominant process technologyand remains sotothisday.CMOS ICsincorporateboth NMOS andPMOStransistors.ApplicationSpecificIntegratedCircuit Anapplication-specificintegratedcircuit(ASIC) isanintegratedcircuit(IC) customizedforaparticularuse,ratherthanintendedfor general-purposeuse.Forexample,achipdesigned solelytorunacellphoneisan ASIC.Incontrast,the7400 seriesand4000 seriesintegratedcircuitsarelogicbuildingblocksthatcanbewiredtogetherforuseinmanydifferent applications.Asfeaturesizeshaveshrunk anddesigntoolsimprovedover theyears,themaximumcomplexity (and hence functionality) possiblein anASIC has grown from5,000gates toover100million.ModernASICs ofteninclude entire32-bitprocessors,memoryblocksincludingROM, RAM,EEPROM,Flashandotherlarge buildingblocks.Such an ASIC isoftentermed aSoC(System-on-Chip).Designers ofdigitalASICs useahardwaredescription language(HDL),suchasVerilog orVHDL,todescribethe functionalityofASICs.Field-programmable gate arrays (FPGA) are themodernday equivalent of7400serieslogicandabreadboard,containingprogrammablelogicblocksandprogrammableinterconnectsthat allowthe same FPGAto beusedin manydifferentapplications.Forsmallerdesignsand/orlowerproductionvolumes,FPGAsmaybemore costeffectivethan anASIC design.Thenon-recurringengineeringcost(thecosttosetupthefactory toproduce aparticularASIC) canrunintohundredsofthousandsof dollars.Thegeneral termapplicationspecificintegrated circuitincludesFPGAs, butmostdesignersuseASIConly fornon-fieldprogrammabledevicesandmake a distinctionbetween ASICandFPGAs.HistoryThe initialASICsusedgatearraytechnology. Ferrantiproducedperhapsthefirstgate-array,theULA(UncommittedLogicArray),around1980.Customizationoccurredby varying themetalinterconnectmask.ULAshadcomplexitiesofuptoafewthousandgates. Laterversionsbecamemoregeneralized,with differentbasediescustomizedbyboth metalandpolysiliconlayers.Somebasedies includeRAM elements.StandardcelldesignInthemid1980sadesigner wouldchooseanASIC manufacturerandimplementtheirdesign usingthedesign tools availablefromthemanufacturer.While thirdparty designtools wereavailable,there was notaneffectivelinkfromthethird partydesign toolstothe layoutandactual semiconductorprocessperformancecharacteristicsofthevariousASICmanufacturers.Most designersendedupusingfactoryspecifictoolstocomplete theimplementationof theirdesigns.Asolution to thisproblem thatalsoyieldedamuchhigher densitydevice wastheimplementationof StandardCells. EveryASIC manufacturer couldcreatefunctionalblocks withknownelectricalcharacteristics,such aspropagation delay, capacitanceandinductance;thatcould alsobe representedin thirdpartytools.Standardcell designis theutilizationofthesefunctionalblocks toachievevery highgatedensity and goodelectricalperformance.StandardcelldesignfitsbetweenGate ArrayandFullCustomdesignintermsofboth its NRE(Non-RecurringEngineering)and recurringcomponentcost.Bythelate1980s,logicsynthesistools,suchasDesignCompiler,becameavailable. SuchtoolscouldcompileHDL descriptionsintoa gate-levelnetlist.This enabledastyleof designcalled standard-celldesign. Standard-cell Integrated Circuits(ICs)are designedinthe followingconceptualstages,although these stagesoverlapsignificantlyinpractice.These steps,implemented withalevelof skill common intheindustry, almostalways produceafinaldevicethatcorrectlyimplements theoriginal design,unless flawsarelaterintroducedbythephysicalfabricationprocess.Ateamofdesignengineersstarts withanon-formalunderstandingofthe required functionsfora newASIC, usually derivedfrom requirementsanalysis.*The designteamconstructsa descriptionofanASICtoachievethesegoalsusing anHDL.Thisprocess isanalogousto writingacomputer programinahigh-levellanguage.ThisisusuallycalledtheRTL(registertransferlevel)design.*Suitability forpurposeisverified bysimulation.A virtualsystemcreatedinsoftware,using atoolsuchas Virtutech’sSimics, cansimulatethe performanceofASICsat speeds uptobillionsofsimulatedinstructionsper second. *A logic synthesis tool,suchas DesignCompiler,transformstheRTLdesignintoalargecollectionoflower-levelconstructscalledstandardcells.Theseconstructsaretakenfrom astandard-celllibraryconsisting ofpre-characterizedcollectionsof gatessuch as2 inputnor,2inputnand,inverters,etc.Thestandard cells are typicallyspecifictothe plannedmanufacturerof the ASIC. Theresulting collectionofstandardcells,plustheneededelectricalconnectionsbetween them,iscalledagate-levelnetlist.  *The gate-level netlistisnextprocessedbyaplacement toolwhichplaces thestandardcellsontoaregion representingthe final ASIC.Itattemptstofindaplacementof thestandardcells,subjecttoavarietyofspecifiedconstraints.Sometimesadvancedtechniquessuchassimulated annealingareused tooptimizeplacement. *The routingtooltakes thephysicalplacementof thestandardcellsanduses thenetlisttocreatetheelectricalconnectionsbetweenthem. Sincethesearchspaceislarge,thisprocess willproducea“sufficient” ratherthan“globally-optimal”solution.Theoutputisaset ofphotomasksenablingsemiconductorfabricationto producephysicalICs.*Closeestimatesof final delays, parasiticresistancesandcapacitances, and powerconsumptionscanthenbemade.Inthecaseofadigitalcircuit,thiswillbefurthermapped intodelayinformation.These estimatesareusedin afinalroundoftesting.This testingdemonstrates thatthe devicewillfunction correctlyover allextremesoftheprocess, voltageand temperature. Whenthistesting is completethephotomaskinformationisreleasedforchipfabrication. Thesedesignsteps(orflow) arealso common tostandard product design.Thesignificantdifferenceis thatStandardCelldesign usesthe manufacturer’s cell librariesthathavebeen used inhundredsofotherdesign implementations andthereforeare of muchlowerriskthanfull customdesign.Gatearray designGatearraydesignisamanufacturingmethod inwhichthediffusedlayers,i.e.transistorsandother activedevices,arepredefinedandwaferscontainingsuch devices are heldin stockpriortometallization, inotherwords, unconnected.Thephysicaldesign processthendefines theinterconnectionsof the finaldevice.It isimportant to thedesignerthatminimalpropagationdelayscanbe achievedinASICsversustheFPGAsolutionsavailableinthemarketplace.Gate arrayASICisacompromiseasmappingagiven design ontowhatamanufacturer heldasa stockwafernever gives100% utilization.Pure,logic-onlygate arraydesignisrarelyimplementedbycircuit designers today, replacedalmost entirely byfieldprogrammabledevices suchasFPGAs,whichcanbeprogrammedbytheuserand thusofferminimaltoolingcharges, marginallyincreasedpiecepart cost and comparable performance.Todaygatearrays areevolvinginto structuredASICsthat consistofalargeIPcorelikeaprocessor, DSPunit,peripherals,standardinterfaces,integratedmemoriesSRAM,and ablock ofreconfigurableuncommittedlogic.Thisshift islargelybecauseASICdevicesare capableofintegratingsuchlargeblocks ofsystemfunctionalityand “systemonachip”requiresfarmore thanjustlogicblocks.Full-customdesignThebenefitsoffull-customdesignusuallyincludereduced area,performanceimprovements and also theabilitytointegrate analogcomponentsandother pre-designedcomponents suchasmicroprocessorcoresthatforma System-on-Chip.The disadvantagescan includeincreasedmanufacturinganddesign time,increased non-recurringengineeringcosts,morecomplexityinthe CADsystemanda muchhigherskillrequirement on thepartofthe designteam.Howeverfordigitalonlydesigns,“standard-cell”librariestogetherwithmodern CAD systemscanofferconsiderableperformance/costbenefits withlowrisk. Automatedlayouttoolsarequickand easyto useand also offerthepossibilityto manuallyoptimizeanyperformancelimitingaspectofthedesign.StructureddesignStructuredASICdesignisanambiguousexpression,withdifferentmeaningsin differentcontexts. This isa relativelynewtermintheindustry,whichiswhythereissomevariationinits definition.However, the basicpremiseofa structuredASICisthatbothmanufacturingcycletimeanddesigncycletime arereduced comparedtocell-based ASICby virtue of therebeingpre-definedmetal layersandpre-characterization ofwhatison thesilicon.One definitionstates that,inastructured ASIC design,thelogicmask-layersofadevicearepredefinedby theASICvendor(orinsomecasesby athirdparty).Structured ASICtechnologyisseenasbridgingthegapbetween field-programmablegate arraysand “standard-cell”ASIC designs.WhatmakesastructuredASIC differentfromagatearray isthatinagatearraythepredefinedmetal layersserve to makemanufacturingturnaroundfaster. In a structuredASICthepredefinedmetallizationis primarilytoreducecostofthe masksetsand isalsousedtomakethedesigncycletimesignificantlyshorter as well.Likewise,the designtoolsused for structuredASICcansubstantially lowercost,and areeasiertousethan cell-basedtools,because thetools do nothavetoperform all thefunctions that cell-basedtoolsdo.Oneotherimportantaspectaboutstructured ASICisthatitallowsIPthatiscommontocertainapplications to be“builtin”,ratherthan“designed in”.BybuildingtheIP directlyinto thearchitecture thedesigner can againsaveboth timeandmoneycomparedtodesigningIPintoacell-basedASIC.中文翻译:集成电路数字逻辑和电子电路由称为晶体管的电子开关得到它们的(各种)功能。粗略地说,晶体管好似一种电子控制阀,由此加在阀一端的能量可以使能量在另外两个连接端之间流动。通过多个晶体管的组合就可以构成数字逻辑模块,如与门和触发电路等。而晶体管是由半导体构成的。查阅大学化学书中的元素周期表,你会查到半导体是介于金属与非金属之间的一类元素。它们之所以被叫做半导体是由于它们表现出来的性质类似于金属和非金属。可使半导体像金属那样导电,或者像非金属那样绝缘。通过半导体和少量其它元素的混合可以精确地控制这些不同的电特性,这种混合技术称之为“半导体掺杂”。半导体通过掺杂可以包含更多的电子(N型)或更少的电子(P型)。常用的半导体是硅和锗,N型硅半导体掺入磷元素,而P型硅半导体掺入硼元素。不同掺杂的半导体层形成的三明治状夹层结构可以构成一个晶体管,最常见的两类晶体管是双极型晶体管(BJT)和场效应晶体管(FET),图2.1给出了它们的图示。图中给出了这些晶体管的硅结构,以及它们用于电路图中的符号。BJT是NPN晶体管,因为由N—P—N掺杂硅三层构成。当小电流注入基极时,可使较大的电流从集电极流向发射极。图示的FET是N沟道的场效应型晶体管,它由两块被P型基底分离的N型组成。将电压加在绝缘的栅极上时,可使电流由漏极流向源极。它被叫做N沟道是因为栅极电压诱导基底上的N通道,使电流能在两个N区域之间流动。另一个基本的半导体结构是二极管,由N型和P型硅连接而成的结组成。二极管的作用就像一个单向阀门,由于电流只能从P流向N。可以构建一些特殊二极管,在加电压时可以发光,这些器件非常合适地被叫做发光二极管或LED。这种小灯泡数以百万计地被制造出来,有各种各样的应用,从电话机到交通灯。半导体材料上制作晶体管或二极管所形成的小芯片用塑料封装以防损伤和被外界污染。在这封装里一些短线连接半导体夹层和从封装内伸出的插脚以便与(使用该晶体管的)电路其余部分连接。一旦你有了一些分立的晶体管,直接用电线将这些器件连线在一起就可以构建数字逻辑(电路)。电路会工作,但任何实质性的数字逻辑(电路)都将十分庞大,因为要在各种逻辑门中每实现一种都需要多个晶体管。1947年,John Bardeen、WalterBrattain和andWilliamShockley发明晶体管的时候。将多个晶体管组装在一个电路上的唯一方法就是购买多个分离的晶体管,将它们连在一起。1959年,JackKilby 和Robert Noyce各自独立地发明了一种将多个晶体管做在同一片半导体材料上的方法。这个发明就是集成电路,或IC,是我们现代电脑化世界的基础。集成电路之所以被这样命名,是因为它将多个晶体管和二极管集成到同一块小的半导体芯片上。IC包含按照形成电路所要求的拓扑结构连在一起的许多小元件,而无需再将分立元件的导线焊接起来。去除了塑料或陶瓷封装后,一个典型的集成电路就是每一边2mm至15mm的方形或矩形硅片。根据制造集成电路的技术水平的不同,在这种小片上可能有几十个到几百万个晶体管,电子器件这种令人惊异的密度表明那些晶体管以及连接它们线是极其微小的。集成电路的尺寸是以微米为单位测量的,1微米是1米的百万分之一。作为参照,一根人的头发其直径大约为100微米。一些现代集成电路包含的元件和连线,是以小到0.1微米的增量来测量的。每年研究人员和工程师都在寻找新的方法来不断减小这些元件的大小,以便在同样面积的硅片上集成更多的晶体管,如图2.2所示。在集成电路的 设计 领导形象设计圆作业设计ao工艺污水处理厂设计附属工程施工组织设计清扫机器人结构设计 和制造过程中,常用两种主要晶体管技术是:双极和金属氧化物半导体(MOS)。双极工艺生产出来的是BJT(双极型晶体管),而MOS工艺生产出来的是FET(场效应晶体管)。在20世纪80年代以前更常用的集成电路是双极逻辑,但是此后MOS技术在数字逻辑集成电路中占据了大多数。N沟道FET是采用NMOS工艺生产的,而P沟道FET是采用PMOS工艺生产的。到了20世纪80年代,互补MOS即CMOS成为占主导地位的加工技术,并且延续至今。CMOS集成电路包含了NMOS和PMOS两种晶体管。专用集成电路(ASIC)专用集成电路(ASIC)是为了特殊应用而定制的集成电路,而不是通用的。比如,一片仅被设计用于运行蜂窝式电话的芯片是专用集成电路(ASIC)。相比之下,7400与4000系列集成电路是可以用导线连接的逻辑构建模块,适用于各种不同的应用。随着逐年来特征尺寸的缩小和设计工具的改进,ASIC中的最大复杂度从5000个门电路增长到了1亿个门电路,因而功能也有极大的提高。现代ASIC常包含32位处理器,包括ROM、RAM、EEPROM、Flash等存储器,以及其它大规模组件。这样的ASIC经常被称为SoC(片上系统)。数字ASIC的设计者们使用硬件描述语言(HDL),比如Verilog或VHDL语言来描述ASIC的功能。现场可编程门阵列(FPGA)是7400系列和面包板的现代版,它包括可编程逻辑块和可编程的模块之间的相互连接,使得相同的FPGA能够用于许多不同的场合。对于较小规模的设计或(与)小批量生产,FPGA可能比ASIC设计有更高的成本效率。不能循坏的工程费用(建立工厂生产特定ASIC的成本)可能会达到数十万美元。专用集成电路这一通用名词也包括FPGA,但是大多数设计者仅将ASIC用于非现场可编程的器件,将ASIC和FPGA两者区别开来。历史最初的ASIC使用门阵列技术。Ferranti在1980年左右制作了也许是第一片门阵列,ULA(自由逻辑阵列)。通过改变金属互相连接掩模产生了定制。ULA有多至几千个门电路的复杂度。之后的版本变得更通用,有适应用户的包含金属和多层硅的不同基底,有些基底包括RAM单元。标准单元设计在20世纪80年代中期,一个设计者要选择一家ASIC制造商,并用制造商提供的设计工具完成他们的设计工作。尽管有第三方设计工具,但第三方设计工具和不同的ASIC制造商的布线以及实际半导体工艺过程的性能之间却缺乏有效的联系。大多数的设计者最终使用工厂特制的工具来完成他们的设计。解决这个问题的一个方法是实现标准元件,这一问题也带来了更高密度的器件。每个ASIC制造商都可创造他们自己的具有已知电性能的功能块,如传播延迟器、电容、电感,这些都可以用第三方工具来表示(实现)。标准单元设计就是利用这些功能块来实现很高的门密度以及良好的电性能。标准单元设计使门阵列和全定制设计之间在一次性投入的工程费用和循环元件成本方面相互适应。直到80年代后期,逻辑综合工具,比如设计编译器,开始向广大设计者提供。这些工具能够将HDL描述语言编译成门级的网表。这就使得称作标准单元设计的设计方法成为可能。标准单元集成电路的设计过程在概念上需经过以下几个过程,但事实上在实际生产中这些工序都有较大的重叠。以工业界普通的熟练水平实现的这些步骤几乎总是产生能正确实现原设计的最终器件,除非后来在物理制造过程中引入了缺陷。设计工程师团队开始工作于对新的ASIC所要求功能的非正式理解,这通常来自于需求分析。*设计团队构建对ASIC芯片的描述并使用HDL语言实现这些目标。这一过程可类比于用高级语言编写计算机程序。这一过程常被称为RTL(寄存器传送级)设计。*仿真验证目标的合适性。利用例如Virtutech’sSimics工具,用软件构建的虚拟系统能以高达每秒数十亿条模拟指令的速度来模拟ASIC的功能。*逻辑综合工具,比如设计编译器,将RTL设计转换成称为标准单元的较低层结构的集合。这些构成的元素是从一个标准单元库中得到的,这个库由事先规定好的门电路集合构成,例如2输入或非门,2输入与非门,非门等等。有计划的ASIC制造商有其特定的标准单元。所产生的所有标准单元,加上连接他们所需要的导线称为门级网表。*接着,门级网表由布局工具进行处理,将标准单元布局在代表最终ASIC的区域。努力寻找一种标准单元的布局服从各种规定的约束。有时,先进的技术比如模拟退火被用来优化布局。*路由工具获取标准单元的物理布局,并利用网表来创建它们之间的电连接。由于搜索空间很大,该过程将产生满足充分条件的解,而不是全局最优解。这个过程的输出是一套光掩模使半导体制造产生实物的IC。*接下来是对最终延时、寄生电阻和电容以及能量消耗的周全的评估。对于数字电路,这将被进一步对应为延迟信息,这些评估将用于最后一轮的测试。这一测试表明器件将在所有极端的过程、电压、温度下正常工作。当这项测试完成时,光掩模信息将被公布用于芯片制造。这些设计步骤(或流程)对于标准产品设计同样适用。重要的差别在于标准单元设计使用制造商的单元库,这些库已用于数以百计的其它设计实现,因而比起全定制设计来风险小得多。门阵列设计门阵列设计是一种制造方法,事先定义好扩散层(晶体管和其它有源器件),包含这些器件的晶片在金属化之前被库存,就是说先不进行联接。然后在物理设计过程中定义最终设计的连接。对设计者来说重要的是,ASIC相比在市场上可提供的FPGA解决 方案 气瓶 现场处置方案 .pdf气瓶 现场处置方案 .doc见习基地管理方案.doc关于群访事件的化解方案建筑工地扬尘治理专项方案下载 ,能达到最小的传播延时。门阵列ASIC是一种折中方案,因为将某一给定的设计与制造商库存的晶片相对应总是不可能达到100%利用率的。现在电路设计者已经很少采用纯粹的逻辑门阵列设计,而几乎都代之以FPGA之类的现场可编程器件了。这些器件可由用户编程,使工具作业费用最低,以略为提高的零件价格获得可比的性能。现在门阵列正在发展为结构化ASIC,其中包含很大的IP内核,如处理器、DSP单元、外围设备、标准接口、集成SRAM存储器、以及一组可重新设置的未确定功能的逻辑单元。这种转变很大程度上是因为ASIC器件能够集成大量的系统功能模块,以及片上系统所要求的(功能)比仅仅逻辑单元多得多。全定制设计全定制设计的优点通常包括减小的面积,性能的改进,以及能集成模拟元件和其它预先设计的元件比如构成片上系统的微处理器核。缺点包括增加的制造和设计时间,增加的不可循环工程成本,更复杂的CAD系统,和对设计团队熟练程度高得多的要求。但对于纯数字设计来说,“标准单元”库与现代CAD系统一起,可以低风险提供相当大的性能/价格优势。自动布局工具使用起来快速且简单,也提供了对设计的性能限制进行人工优化的可能性。结构化设计结构化ASIC设计是一个不明确的表达,在不同的上下文中有不同的意义。在工业界这是一个相对新的术语,这也是为什么在它的定义上有一些不同。不过结构化ASIC的基本前提是,由于有事先定义的金属层和事先规定了硅片上包含的内容,制造周期和设计周期相对于基于单元的ASIC都有所减少。一种定义是这样的:在结构化ASIC设计中,器件的逻辑掩模层是被ASCI供应商(有些情况下由第三方)预先定义的。结构化ASCI可以被看成是在现场可编程门阵列与“标准单元”ASCI设计之间建立联系。使得结构化ASCI与门阵列不同的是,在门阵列中,预先定义的金属层是为能更快地制造转向而服务的。而在结构化ASIC中预先定义的金属化主要是降低掩模的成本,并被用于使设计周期明显缩短。同样的,为结构化ASCI所使用的设计工具可以大大降低成本,并比基于单元的工具更容易使用,因为这些工具不必像基于单元的工具那样执行所有的功能。关于结构化ASIC的另一个重要方面是,它使对于某些应用共同的IP成为内在的,而不是设计在内的。通过直接将IP植入结构中,相比将IP设计在基于单元的ASIC中,设计者又能节省时间和花费。
本文档为【Integrated_Circuits(集成电路)电子信息类专业英语、计算机类专业英语】,请使用软件OFFICE或WPS软件打开。作品中的文字与图均可以修改和编辑, 图片更改请在作品中右键图片并更换,文字修改请直接点击文字进行修改,也可以新增和删除文档中的内容。
该文档来自用户分享,如有侵权行为请发邮件ishare@vip.sina.com联系网站客服,我们会及时删除。
[版权声明] 本站所有资料为用户分享产生,若发现您的权利被侵害,请联系客服邮件isharekefu@iask.cn,我们尽快处理。
本作品所展示的图片、画像、字体、音乐的版权可能需版权方额外授权,请谨慎使用。
网站提供的党政主题相关内容(国旗、国徽、党徽..)目的在于配合国家政策宣传,仅限个人学习分享使用,禁止用于任何广告和商用目的。
下载需要: ¥14.0 已有0 人下载
最新资料
资料动态
专题动态
个人认证用户
洛衣含
暂无简介~
格式:doc
大小:53KB
软件:Word
页数:0
分类:生产制造
上传时间:2021-03-02
浏览量:16