InternalMemo1XEROXToFromDragonCoreRickBarthPARC-CSLSubjectDateMBusCompatibilityNovember19,1984IntroductionLong,long,agoandfar,far,away,sayabout400miles,ChuckThacker,RonRider,SteveDashiellandGodknowswhoelsesatdownanddecidedwhatthemainbus,calledtheMbus,shouldlooklikeforanewmemorysystem.ForavarietyofreasonscommunicationsbetweentheDragondesigncrewandthefolksdownsouthbrokedownoverthelastcoupleofyearsandthedecisionsregardingthecommonbusfellbythewayside.SometimetowardstheendofAugustofthisyearRonRidercamecruisingbyandEdMcCreight,CarlBlackandIhadalittlemeetingwithhimwherewetalkedabouthowsimiliarourmainmemorybuseswereand,gosh,wouldn'titbeniceiftheywerethesame.Therehavebeenanumberofphonecalls,wemadeavisittoElSegundo,andanumberofmessageshavebeentransmittedsincethen.Thismemoisanattempttowritedownthestateofcurrentdesigns,futuredesignsthatarecontemplatedwhichdependoncurrentdesigns,differentoptionsfromwhichwecanchoose,theimpactoftheoptionsuponthedesigns,specificallyDragon,andtocometosomesensibleconclusionaboutwhatweshoulddo.CurrentDesignsESSThecurrentESSdesignhasonlyonecustompartthatdependsontheMbusprotocol.ItisacacheandwillbereferredtoastheESScache.ThespecificMbustowhichanESScachemaybeattachedisreferredtoastheESSMbus.TheESScacheisbeingusedtoconnecta8086stylebustotheMbus.Ithasasmallamountofcacheddata,about16quadwords,andhasfacilitiesformailboxesandstreamoperations.Ithas24bitreal,andvirtualaddressfieldsalthoughtheprotocolallowsspacefor31bitvirtualaddressesand28bitrealaddresses.Theprotocolhas3freecommands,oneofwhichisactuallyusedinternallybythecache,leaving2freecommandsforexpansion.TheESScachedoesnothaveanyflowcontrolledI/Ooperations.ThismakesconnectingthesynchronousMbustoanasynchronousbussuchastheVMEbusmoredifficult.ItalsopreventsthemapprocessorfromusingtheMbusduringmapoperationstherebyforcingmapmemoryaccessestotakeadifferentroute.ThepïfÔî]îõqïfÔîCNrïfýî{sïb î]ïb î2Pqï`dî]îžï`dî2Pî5«ï^½î2Pî6©î7§sïTÏî]ïTÏî2PqïS'î]î î ïS'î2Pî9`î;ÔtïI:î] uïEëîî(î®îWî!"î#µî&Iî*2î,¤î0™î3Bî7Lî;ÆîAŠïDDî]î…î(îzî!î$>î(pî+cî.î0î3Ìî6rî;‚î>ÎîAïBœî]îJî?vî�ïBœïBœîtuîeïBœïBœî@î#ºî&Òî){î+·î,ßî/Ëî57î:Šî=4î>[îBàï@õî]î6 î gî%Ñî(+î-%î1oî4°î7bî9¼î=î@Øï?Nî]îVî.îUîÂî!Pî%Îî'”î+*î-ïî0\î6Eî<|î>éï=¦î]îôî]îYî¸ï;+îî›îÂî"'î$äî&£î+gî-%î/Íî2Öî5ëî9Òî=kîB�ï9„î]îîA î{î ‡î$Nî'î(î*½î+ãî/î4Pî7gî:<î>Qî@uï7Üî]îPîfî_îûî#Šî)î,Úî0<î3]î7î<Ïî>6î@8îC8ï65î]îoîÂî#î€î"„î%Ìî)+î*Vî/„î1?î5{î8äî; î>Ïî?ûîBêï4Žî]î,î5îìîî"Bî#ûî)×î-î0z î7¹î;,ï2îîhîÐîkîŠî$Æî&£î*Pî.Jî0Ùî49î6!î;î@uï0kî]îQî@î´ î$^î(Œî-—î/¾î4¨î9ôî?ºï.Äî]îÈîÚîî”î"vî$Úî)eî+"î-…î2hî6î8cî=’ ï-î]îµîlîîÄî sî$î)1 î0î3âî7@î9gî=ãtï&î]îvwï#/î]uïàîîî×îÜî#7î%·î(Üî+“î0]î3Lî6&î; î=²î@îBï9î]îÎîYîØîîôî"¿î%~î'‚î,×î.™î0Zvî2Íï9ï9î3Àî6uî9Oï9ï9î:Õî=ÁîB¿ï’î]îöî¦î¶î©î¥î"sî%yî'lî,éî.Wî3›î5Kî6üvî9]ï’ï’î:Pî<�î>kuî@\ï’ï’îAÁïêî]îdî=î¶îŽî ×î"’î'¶î(íî,@î/~î2!î3Ýî6Iî81î;½î=Bî?Äî@ûïCî]îzîDîèî Wî$Jî&Aî)Ñî.Gî1î3— î9 î;ZîAâïœî]î¼ î"îˆî ëî"¿î$Ëî'Èî*nî.¶î3�î72î<Ýî?+ïôî]îƒî1îgîDîZî#­î)¹î,iî.Gî0\î3 î9ìî<½îB$ïMî]î¨îŠîÞîŸî!nî%’î'î,@î/‘ î5Ñî7âî:Wî>‘îCdï¦î]î*îîL ï*îîøîòî¿î ðî#fî&­î)Oî,j î2íî5º î=Eî@`ïƒî] îeîÜ î"âî$Óî'�î)Gî+N î3ùî6§î9ðî;µî>,îBï Üî]î înîîøî% î(#î+`î1—î5!î8îî;qî=nî@(ï 4î]îh îîøî$›î'¦î-î2+î3Êî6´î7Îî=eîAÁÿ9TVm$¬InternalMemo2I/Oaddressspacedoesnothavelargecontiguoussectionsandislimitedtoatotalofnomorethan25bitsbecause4bitsareusedforthecommandfieldandatleast3bitshavebeenremovedtoease8086compatibility.ThisdoesnotallowaddressinglargeframebuffersintheI/Ospace.Provisionsformultipleaddressspacesarenotyetimplementedandtheproposedchangetoallowforthemdoesnotcleanlyimplementtherequiredoperations.EDiscurrentlyawaitingsecondsilicononthischip.Firstsiliconshowedgreatpromise.AnumberofI/Ocontrollers,includingdiskandEthernetcontrollers,havebeenorarebeingconstructedbasedonthecapabilitiesoftheESScache.Memorycontrollershavealsobeenbuilttothisspecification.DragonThereisadesignfortheDragoncachethathasexecutedeachtransactionontheMbusandallofthecommandsonthePbus.TheexecutionwasdrivenbyablackboxRosemarytestprogram.ThelevelofdescriptionintheRosemarycodeisslightlyabovethatoftransistors.Thelayoutforanentryofthecachehasbeendone,althoughthefirstpassatthecontrollogicforanentryrequirestoomuchspaceandsoitwillhavetoberedone.SomepreliminarylayoutoftheCAMandRAMdrivershasbeendonebutthedesignhaschangedsomuchsincethesewerelayedoutthatatbesttheyserveasguidelinestohowbigtheseblocksmightbe.Basicallyaveryfirmspecificationforthecircuitdesignandlayoutportionsofthedesignprocessisinhand.About6monthsofworkremainspriortosubmittingfirstmasksforprocessing.ModificationsthatwillbemadetotheDragoncacheBytewriteenablepinswillbeaddedtothecache.Abusmodepinthatcausestheparitybitofawordbeingwrittentobecalculatedinternallywillbeadded.ThesepinswillbeaddedtothedesignregardlessofwhetherornottheentireDragoncacheisredesigned.Theyallowforattaching68020and80186processorstotheDragoncache.AsmallpileofSSIandMSIcruftwillberequiredbetweenthecacheandsuchforeignprocessorsbutimplementationwillbestraightforward.Iplantoattach68020processorstoacompleteDragonmemorysysteminordertotestitindependentlyoftheDragonprocessor.Ifsomeoneweretositdownandplaywiththe68020and80186processorinterfacesenoughthatwecouldbereasonablysureofgettingtheinterfacelogicrightthereisplentyofsiliconareatoembedtheinterfacefortheDragon,80186and68020allononedieandusestrappinstoselectwhichprocessorisactuallyattached.Currentlythereisnowaytodetectwhenanillegalmemoryreferenceismade.Thecontrollersineverybusmasterwillbemodifiedtodetectwhenthecommandwiresonthebushavenotbeendriventoavaluedifferentthanthevaluewithwhichthebusmasterstartedatransaction.Abusmasterthatdetectsthisconditionmustraiseanerrorandreleasethebus.Thecachecanhandlethiscleanlybycausingitsprocessortotrap.I/Ocontrollersmightsetanerrorflag.ThemapprocessorcouldsetanerrorflagandreturnanerrorconditionastheresultoftheIOReadorIOWritethatstartedthemapprocessor.ThereareafewinconsistenciesbetweenthecurrentMbusspecificationandtheimplementationthatmustbecleanedup.ReasonstoredesignthecachepïfÔî]îõqïfÔîCNuïb2î]î4î)îéî $î"¤î%öî)h î0cî5�î8Qî9Èî>…î@?îAsï`‹î]îîî³îÚîÀî bî%…î&»î)]î+²î.ïî1-î3Žî9ðî=!î?ÚîAsï^äî]î|îî7î~î!î"°î%“î(à î1Åî4Éî7ãî:Bî=Õ ï]<î]îÄî»înîî!{î$HïZÁî î”î°î"î&Ñî*éî-î/oî1– î9Çî<]î>›ïYî]îî³î\îšî!#î$Sî&Èî+pî2Uî4´î:D ïVžîî¤î îãî#Vî'ãî,-î..î0Êî4¬î7íî<6îA&ïT÷î]îPîÇîßî‚î": î)Aî/@î2î4»î:K îARïSOî]î¹î{îÎî— î#ìî'Ìî)Ñî,0 î3Kî5î7bî:Zî>ÿïQ¨î] î!îgî4î!‘î$Øî&†î)( wïNXî]uïK îî4î²îïîRî!¢î$î)'î-î/èî2pî8?î;„ îB“ïIbî]îÂî£î?îüîòî±î!î(î*î,sî-Ïî1Mî4,î:gî=îAmîCoïGºî]îîÛî_î ùî'�î*gî-»î/ƒ î6£î8`î:ÍîARïFî]îÓî¦î²îŠî M î'¿î*¡î.Úî1!î3î6ºî8}î:çî>½îA<ïDlî]îPî5î¾î ¾î#àî%¡î()î-î0î2äî4ýî8ºî>0î@ÂïBÄî]îîÜî¨îî¼î! î"Ãî$¼î*Fî.% î5—î9Îî;�î=ùîAâïAî]îQîìîjîÐî#Kî%Öî(>î,–î/î4�î6Zî::î=µîAHï?vî]î'î¾î­îfîkî"›î&Dî( î.¶î0…î3ªî6)î9Ôî>9îBQï=Îî]îîXîwî— î&‹î(Ùî+Jî/•î3õî6½î:üî@pîB9ï<'î]îœî_î»îYî!’î%Çî&ëî+Åî-nî0Ûî5öî9Xî:ö îAÁï:€î]îtî² vï70î] îÝî¯îDî!î$ªî&Oî(›î-†uï3áîîGî½îî" î$¯î&˜î*´î,[î.²î3Jî4Ïî7^î;,î=™î@_ï29î]îëîî^îFîžî!Vî%Pî*2î,î./ î4Ë î;#î=þî@ï0’î]îšîÎî´îàî"?î$(î&Âî+L î1çî3Ûî9dî;cî>î@­ï.ëî]îkîFîÀ î#Ÿî'6î*îî-:î31î76î9üî> ï-Cî]îî|î…î>î Óî$zî':î(ûî+›î.[î1‰î4äî7˜î9’î?*ï+œî]îÔî¸î‡îÐî"¥ î)Uî+î î5äî8§î:°ï)õî]îgîƒî3îCî; î%Ôî'„î(¯î.˜î3™î9î=…î?4îBêï(Mî]îî~ î î xî"÷î( î/Dî0åî6¿î:0î;ýî=÷îAâï&¦î]î¼î1îìî=î"Oî&Ÿî- î3’î8×î<î>‚îB¨ï$ÿî] îBîFî î!¦î$î)Áî-î0zî4î5…î9Éî;�î?çîBêï#Wî]îíîVîî Lî"µî(î,î.Öî2Öî4Ðî6ßî9”î;òî>²îA1ï!°î]îVîîÅîÓî#æî%Rî*hï4îî_îéîVî!]î$6î%åî)ôî-ªî/›î3¨î9î?î@�ï�î]î@ îîÇî †î#(î'£î*Yî,Tî2;î3ôî8î;Îî>8ïæî]îÜîÏîîžîÑî!2î${î(·î*Qî+gî.ýî4�î7¢î9îî=„î@‹ï>î]îÁî[îÐîPî ~ î(^î)ïî,Šî0þî3Ñî8hî;îA1ï—î]î•î‚îñî¦î!)î#…î&ñî)Çî-�î0î4œî7:î;ßî=×îB¾ïðî]îeîî¯îp î&(î*î,%î. î1pî4ëî7·î:Æî@ÍïHî]î]î1î‡î!î¼î!Ïî#£î&øî,ûî.�î0Ðî4~î6î8^î=yî? ï¡î]î+î¨îî! ï%îîCî¹îîÉî(%î-·î09î5%î7"î9Û îAâï~î]î¼ î›î iî#Ñî%Âî*Âvï /î]î¢îGîŒîØ ÌTVm$üInternalMemo3RegardlessofwhichMbusprotocolisusedtherearesomereasonstoredesignthecache.ThemostimportantreasonistomaketheRAMstatic.DynamicRAMisaholdoverfromThacker'sdesignthatIneverreallythoughtallthewaythrough.Theproblemwithitisthatwecannotcalculatetheerrorratethatwillbeintroducedbyalphaparticlehits.Wehavetobuildthecellandseewhattheerrorrateis.Waitinguntilwehavebuiltthecacheandembeddeditinthesystemtofindoutiftheerrorrateisacceptableseemslikearecipefordisaster.AsingleportstaticRAMwillhaveaboutthesameareaasthedualportdynamicRAMinthecurrentdesign.IcangothroughreasonablebandwidthargumentswhichshowthatadualportedRAMisnotneededfortheDragonanyway.Somethoughtshouldbegiventotheamountofbandwidtha68020or80186willrequireifattachingthoseprocessorsinanoptimalmannerisfelttobeveryimportant.Anotherreasontoredesignthecacheistoremovesomeofthecomplexityfromthebuscontroller.IamstilladaptingtothedelaycharacteristicsofduallayermetalandIputsomecomplexityintobuildingafastcontrollerthatcouldprobablybeeliminatedthroughjudicioususeofsecondlevelmetal.FinallyIhavesomeideasabouthowtodothedesignusingabettermethodology.Thisisnotaverycompellingargumentfromaprojectmanagementpointofviewexceptthatitkeepsmeinterestedindesigningthebeast.Designingitforthefourthtimewithchangestothespecificationsbutnottothemethodologyisboring.Changingbothcouldbeatleastsomewhatinteresting.Redesigningthecache,assumingthatstrictcompatibilitywiththeESSMbusisnotrequired,thatthedesignmethodologyisnotchanged,andthatIdotheredesign,willaddabout3monthstothedesignschedule,changingitfrom6monthsto9monthsuntilfirstmasks.FutureDesignsCWorkstationTheCworkstation,which,lestanyonebecomeconfused,hasnothingtodowiththeCprogramminglanguage,isplannedtobeintroducedbyXeroxinearly1987.TherequirementsthatIhaveseenforitspecifythatitbeamultiprocessorandallowforeignprocessorstobeattachedtoit.ThenotionisthatsomenumberofenhancedMesaprocessorswillbethemaincomputingenginesinthismachine.Theactualamountofmanpowerworkingonitrightnowisverysmall.Duringthesummerof1985itisexpectedthatthemanpowerwillrampup.SteveDashielltoldmethattheywouldliketomovetheI/OcontrollersfromtheESSdesigndirectlyintotheCworkstation.TheyexpecttodesignaprocessorcacheinCMOSsothatitwillperformadequatelywith16MHzprocessors.TheyrealizethattheESSMbusprotocolisinadequatefortheCworkstationbutbelievethattheI/Ocontrollerscanstillbesalvaged,atleasttemporarily.IhavetroublebelievingthattheI/Ocontrollersshouldbesalvagedforaproductwhichisslatedtocomeoutsometimeafter1Q1987.Morecosteffectivesolutionsarebecomingavailablefor2ofthekeycontrollers,diskandEthernet,andanewworkstationshouldbedesignedtotakeadvantageofthem.Itisnotclearhowrealthescheduleortheproposeddesignareatthispoint.ItisespeciallyinterestingtonotethatnoneoftheSBU'shavesignedupforthispïfÔî]îõqïfÔîCNuïb2î îî¶î Àî"—î%*î*”î+ýî/5î2»î5 î8›î=uî? ï`‹î]î¼ï^îîî�î!î%†î'î(Üî,®î/,î35î7Öî=ÜîAåîCoï\iî]îîbî ƒî$¾î'yî(nî,$î/Úî4éî6Åî9î;ÕîAÁïZÁî]îÞî÷îJî³îî!£î&î+Àî.î1Œî4Lî7î9Àî;® îB�ïYî]î îßî3î©î"Üî$vî(î*Lî,Âî/eî1¤î4íî78î:–î=Dî?bïWsî]îœî¼îûî;î“î#Wî&î,ºî.î/¯î2î6zî8!î;î=vî>ÏîA'ïUËî]î>îÉ î—î ¸î#‚î$Ëî)î+^î1mî3î7"î:;î=ãîAíïT$î]î­î˜îîˆî!Šî#Aî%ªî(Ïî+Òî1kî5_î7î9€î>SîC�ïR}î]îî;î¥ î"¥î)²î0’î4Ñî8‹î;‹î<æî@3ïPÕî]îJî·î-î÷î 6î"—î'—î-jî1Aî6eî:âî<Ôî@ŠîB9ïO.î]îkî&îî 0î$)î%ïî)èî,–î1aî2Äî8¯îî7éî:: îA2ïIdî]î¬î2 î,î%î hî"øî(Žî*,î,{î0î8Åî:mî=wî@ÎïG½î]î6î`îî· î"ßî%Ùî+tî,¿î/w î5Øî8Èî<µîB¨ïFî] î!îYî#6î%«î'dî+öî/<ïCšîî¦î…î¢î î#_î'î)ñî+vî-Rî/ˆî3­î7-î8-î;û ïAóî]îŠîî�îÌîî î$î*Nî-Èî/î3´ î;áî?’îA^ï@Lî]î�îZî îeî¨ î%ãî'�î-šî/éî4-î:�î;ãî>î@_ï>¤î]îÖîDîÃîÅî"x î+3î.î0Ðî2Ñî5„ î>î?Üï<ýî]îœîÙî¤î •î".î%Sî+¿ ï:‚î îîwî"«î(¹î+—î/ î7sî:žî=î@îBï8Ûî]îî®îÏîÕî!mî%ô î.sî0î2Äî8Ãî;³î>ºî?úîB9ï73î]îî³îWî &î!Gî&î'¹î*î.Bî4-î9÷î;:î>�î?¯ï5Œî]î î@î*îqîItï.ïî]î½wï+ î]îÔ uï(QîîãîR îÿî$Rî&¾î+sî0uî6Œî8îî=ðî?‹îA~ï&ªî]îÇîT îòî$î%’î*êî,¤î.  î5œî7£î;Ðî=Šî@ûï%î]î! î_îî î#@î&Fî(oî)±î.9î0óî25î4î5( î>Mî@ðï#[î]î î‹î&î î%kî'î)î+Öî0î1uî40î7°î<Éî>nï!´î]î îÌîŽî •î# î&Ÿî-‘î2œî4aî7î=§î@–ï î]î`îîÍî$î&î'hî*²î-®î/î2î6ƒî;Dî=›îBàïeî]î£îùîeî!îðî"Oî)î+¾î/]ïêîîâîSî :î"œî%yî(—î,Óî/�î1Jî5î7…î:` îA2ïBî]îÇîÊî#î #î#î%nî&ú î/Sî2æî7@î8ùî=Rî>…ï›î]î5îîî¬î|î Vî!·î$nî)Ò î0Ïî3õî5æî9† îAïôî]î°îyîÒîÆîšî"+î'”î(ù î/ûî22î4Œî6 î=kî?çïLî]îEî¾î¥ î ‚î#'î%âî'íî-âî/•î2Ô î;MîNïVî]îGî´îîî�î"U î) î,¾î/¤î5Òî8¹î:î=0 ï¯î]îÙîÊî…î4î"-î(¤î*]ï 4îî©î'î¯îî +î"ôî%gî+ î,÷î/iî5zî9Üî•î?äîAGï]<î]îîYî îî fî&)î':î)¬î.Ÿî0?î3Lî6y î>z tïVŸî]wïSPî]î îÅî#îHî -uïPîîtîî¡î#î$þî'$ î.V î5žî8ïî;„î>´î@ÄîC�ïNZî]îîÐîî’î!’î"Ìî%êî+*î-î2‡î4Dî6³ î=¾ïL³î]î|îâîî î"‹î'“î)(î,£î/Mî3W î9Íî;Wî=’ ïK î]îîuïH�îîÕîXîÃî"‚î%[î)sî-Âî.÷î2€î4ßî:…î;ûî>{î@4ïFéî]îî…îŒîéî¤î _î"Ìî(Âî+éî.¢î2"î3ôî6Ïî9<î>Uî@îBïEBî]îJîî!Vî#Üî(•î,[î-•î1’î7î:1î=[îAïCšî]îoîïîWîZî!ªî$î'ìî-î.Þî20î6>î7÷î;ÍîA~ïAóî]îCî` îøî°î tî#…î%Œî(î-ßî1Öî3Ýî6=î9»î=.î>ÜîA<ï@Lî]î‘îîéîvîÍî#8î'`î)�î,÷î1 î4>î6¸î9× îAï>¤î]îºî/îŸînî!ôî&½î)î,Kî0oî3‰î5�î;¹î> î@îBï<ýî]îîÿ îúî8î$6î&Ìî(Lî*Öî0ýî2Àî3þ îîNîî î!î({î.î/Äî2:î5Îî:ïî=Ôî@äï9®î]îŸîÄî‹îÙî$ïî+î-Üî1î3ªî5î7ˆî; î?¬îASï8î]îðîîOî Áî#~î%Hî'¹î+î.Œî1Uî3Ûî7€î9@î;›îBï6`î]îIîöî"î" î+@î.ïî1î3¶î7…î8Õîî3î45î9Fî;‡îB9ï*Ìî]îwî­îjîÎî!gî#î%~î*Þî-ý î4ƒî5±î;‡îBEï)%î]î îî!-ï&ªîî™î]î.îÏî$ðî&tî0¸î2Àî5dî7mî<Òî?(îAŸï%î]î îƒîÜî!öî#Àî' î,ãî/&î1Œî6‘î;Èî?Wï#[î] î]îîK î'Iî)î-î/äî4»î7ƒî:Î î@åï!´î]î¨î� î î!Éî$6î'=î)%î+Éî1~î3ëî82î9îîÓïî]îKîî¨ î#kî'™î*ßî,�î.~î4Zî6î7ø î?ƒîBïoî]îî{îÆ î ñî%qî(,î*Wî.'î2èî63 îp ï Üî]îîÂîîjî î&î** î0¾î3Fî6<î7èî=(îAsï 4î]îîˆî%î˜îwî$eî&ö î/�î1oî5Tî8Áî>ûîAï �î]îhîŒîî°î"xî%òî*Rî+Çî-1 î4¯î9Ïî;ôî>¨î@¸ÿ ÕTVm$üInternalMemo5alloftheworkandthereisamuchlessclearpathfromresearchtoproduct.ConsideringtheDragonasanisolatedprojectIbelievethatthisroutewillgiveusworkingDragonsfasterthantryingtobecompatiblewiththeESSbusbutispotentiallyslowerthanthenextalternative.CooperativelydesignanewbusandthechipsthatgoonitThisoptiongivesustheopportunitytocleanupallthewartsintheESSMbusdesignthatwehavediscoveredsofar.Withthiscooperativeoptionwehavelesstimetonoticenewwartsbeforetheysolidify.TherestillremainstimebetweennowandwhentheCworkstationisfullystaffedtofinishthearchitecturedowntoquitedetailedsimulations.IthinkthismaybethefastestmethodofproducingbothworkingDragonsandworkingCworkstations.ConclusionsThemajormotivationsforbecomingcompatiblewithanEDMbusaretoeliminateneedlessreplicationofeffortinresearchanddevelopmentandenhancetheexchangeofideasandresultsbetweenresearchanddevelopment.Weshouldpursuethesegoals.WeshouldnotattempttodosobybecomingcompatiblewiththeESSdesignbecauseofthedeficienciesintheprotocoloutlinedearlier.ConvergenceoftheDragonandEDMbusesshouldoccurintheCworkstation.TheI/OcontrollerissueshouldberesolvedbyeitherredesigningtheI/Ocontrollersordesigningagatearraythattranslatesbetweenthenewandoldbuses.JointdesignofacacheforDragonandtheCworkstationshouldbeexplored.Technicalexchangeofportionsofthedesignsshouldoccurevenifindependentdesigneffortsarepursued.pïfÔî]îõqïfÔîCNuïb2î]î’î�î4îöîòî"Àî$qî%àî)ûî,Öî0lî3×î7ƒî=,î?ï`‹î] î îŒî —î"Rî$Oî)fî.î/%î3ãî6¾î9lî=î?ÏîBÉï^äî]îÿîÎî Ùî$Jî(—î*‘î,Í î4*î7�î:;î=î@aîC-ï]<î] î,î{î¡î"î% wïYîî] îûî2îgî"3î$²î'Sî)±î-9î0î2 î4uïVŸîî+îwîÛî œî"î î*kî, î/‘î1•î3wî5Éî9Yî:ùî=Kî@6îBïTøî]îÄî©îèîF î%-î' î*;î-Ðî0‰ î7ôî¥îBïSPî]îuîî!îî “î$Õî'×î-vî1kî3þî9î<4îA•ïQ©î]îîÈî'î© î#î$}î'¸î,3î-áî1·î4 î;—î?aîAïPî]îÇ îEî €î$Rî'&î*\î,î/î3sî8±î:œîA]ïNZî]î´î8îïî$Fî%È tïG½î] uïDnîî/î] î"þî%sî+ö î3@î6“î8»î;~î=�î@_îBêïBÇî]î~î î$î%áî)Ðî+¢î1*î4 î›ï5Œî]î_î îî·î"Óî'˜î+¢î/2î0Ü î9î=­îBEï3åî]GTVm$aÿ TIMESROMANþŸÿ TIMESROMANþŸÿ TIMESROMANþŸÿ TIMESROMANþYÿ HELVETICAþæÿLOGOý‡ÿ HELVETICAþŸÿ HELVETICAþ¶ � ¡ ˜)ú/j/20�Ûªnÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ[]<>MBusAnalysis.tioga$Monday, December 3, 1984 3:55 pm PST