throbber
VUMA Proposed Stan;
`
`.1
`
`Confldentiai
`
`
`
`MREQ#isd:ivenhwfromdoekedge1Cmelogienmplesitaefivednciockedge3.
`Arbitercangivethebtlszightawaxsocorelogicdrivesall-sftissignalshighfromthe
`' same cioek edge. Core logic In‘-states-all the
`(W3 and tie) and drives
`MGNT# activefromclockedge4.VUMAdeviee-min MGN'I‘# active atclock edge
`Sandslzirtsdzivingthebusfi-cmtbesameedge;
`
`‘I'hesharedDR.AM signalsared:ivenbyVUMAdevicewi1enitistheownerofthe
`physica1sysnemmemmybm.VUMAdeficemIinquishesthcphysicalsysmmmemonr
`busbyde-assertingMREQ#.BusArbitergives1heb11sbaektoeoreIogicbyde-asserting
`MGNT#.A1so, asmemionedabove,beforecoreiogicstansdrivingti1e'bus, VUMA
`device should drive the sftfs signals high for one CPUCLK clock and tri-state them.
`VU1'v[Adcviceshoulda1sotri-stateallthesharedtfs signals.-'1'he'£loa1eonditiononthe
`busshouldbeforone CPUCLKclock,bet'oreeerelegiestartsd:ivingthebus. These
`activitiesareoverlappedtoimproveperformanceasshowninFigure 5-6.
`
`Figure 5-6 Bus hand ofi from VUMA device to core logic
`
`
`
`VUMAdeviced1ivesalls!tlssignalshighfi‘nmclockedge3.VUMAdeviee1ri-statesall
`shared signals (sftfs and us) and de-asserts"-‘B/£REQ# fi'om'eldci:' edge 4. Core logic
`samples MRI-3Q# inactiveonclockedge5.Corelogicdrivesa1lsharedsignalsand
`deasserts MGNT# fmm clock edge 5.
`
`5.3.2 DRAM Precharge
`
`25
`
`Version. 1.01:. Rein 0.44:
`
`Page 211 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 211
`
`

`
`VUMA Pmpmd 5%”
`
`lard
`
`. VESA Confidential
`
`Whmthephysicalsystemmemonrh1Lsishmdedofi‘fiummaeiogicwVUMAdevioeor
`fieeavusa,&nDRAMneedstohepmehargedbefo:ed:enewmstummchivingit.
`PmofthisprechargecanhehiddenhyovuleppihgflfihthenhiuafiohprotocoL As
`showninFig:m_:_5-5 and5_-6.a1l_theD_R.AM .control.si5nals (includin§R.AS# lines)are
`driven high forand tri-stated
`lines are tri-
`nmed,mflll1psonfl1osefines-ptfllthemmabgiulhighlhmfihefiinewmamagets
`theconflDLtheRAS#finesareakeadybeen'pmehargedformuCPUCLKclocks.Ihe
`restofthepreehnrgeneedstohetalnencat-e'hythenewmaster.
`
`.
`
`.
`
`1.VUMAdevieegetsthebusfiomeore1ogic-AsshowninFigute5-5,whenVUMA
`deviceg:tsthephysicalsysaemmunoryb1mucloekedge5,1heDRAM-hasbeen
`prechargedfortv-woCPUCI.KciocIcs.VUMAdevieen.eeds,~totake-eareoftherestofthe
`DRAMpmchmgeTIfispre:hargecanbeovu'lappedbyVUMAgiflice.oversomeofiB
`fictivity e.g.VTJMAdevicemaybenmningatadifl‘e:entcloc.lcthan1heCPUCLKclock
`andtheprechargecanheoverlapped withthesynchronizafion;ofMGNT#signal.VUMA£
`deviceeancalculatethcnumberofcIocksitneedstoprechaIgetheDRAMwiththe
`followingformula:
`
`No. orvuma device clocks for DRAM plrecharge - {RAS# Preehatge Time (tR.P) - (2 *
`cpucuc Clock Time Period)}i VUMA device Clock Time Period
`
`Example: cpu running at 55.55 MHz, WMA device runniog at so-MHz. 70115 Fast Page
`DRAM used.
`
`No. of VUMA device clocks for DRAM precharge = {$0135 - (2‘ 15115)}! 20115
`= {Z0115}! 20ns
`- 1 clock
`
`2. Core logic gets the bus-from VUMA device - As shown in Figure 5-6, when core logic
`gets the physical system memory bus" at clock 5, the DRAM has beetrprecharged for
`two CPUCLK clocks. core logic needs to take care of the rest of the DRAM preeharge.
`This precharge can be overlapped by core logic over some of its a:ctivity.e.g. driving of
`new row address. Core logic can calculate the number 51* clocks it needs to precharge the
`DRAM with the following formula:
`_
`_
`__
`-
`
`No. of CPU clocks. for DRAM precharge - {RAS# Precharge Time (IR?) - (2 * CPUCLK
`Clock Time Periocl)}f CPUCLK Clock Time Period
`_
`
`Eltample: CPU ntnning gt 66.66 MHz. VUMA devlee
`DRAM used.
`
`ex 50 MHz. 701:5 -Fast Page
`
`No. of CPU clocks for DRAM preeharge - {5Dns - (2“ 1Sns.)}! 15:15
`- {2Dns}/ 151:3
`- 2 clock
`
`35
`
`Version. ‘Lop. Rev. 0.41:
`
`Page 212 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 212
`
`

`
`VUMA Proposed stam.
`
`.1
`
`;EsA cor-rfldantlal
`
`5.4 Synchronous DRAM
`
`Synchronous DRAM support is optional for both lcorelogic and VUMA device. Various
`SynehmnousDRAMsupportscenariosareasfol1ows:
`
`1. Core logic does not support Synchronous DRAM - Since core logic does not strpport,
`Synchronous DRAM, there would not -be u1y'S'y'nehronous DRAM is the physical
`systemoryandhencewhetherVUMAdevicesupponsSynchronousDRAM or
`not is inelevant.
`
`2. Core logic. supports Synehronom DRAM - When core logic supports Synchronous
`DRAM, VUMAdevicemayormaynotbe'supponingiLWheti:eroorelogicand
`VUMAdevicesupponSynchtonousDRAMornotshould'beuansparenttothe
`operating systemandapplieatiom programs. Toachievethe
`systemBIOS
`needs to findout ifbothcorelogicandVUMA=device support"this_feamreandsettl1e
`systemappmptiateiya1booL1he-foflowingalgofithmexplainshowitcanbe
`achieved. The algorithm is only included to explain the feature. Refer to the latest
`VUMAVESA BIOS Extensions forthen:1ostnpdatedBIOS calls:
`'
`
`a.Read<VUMABIOSsignan.tresI:'ing(tefertoVUMAVESABIOS
`E.xtensions)>. Check if VUMA device supports Synchronous DRAM.
`b. If VUMA device does not support Synchronous "DRAM, do not assign
`Synchronous DRAM banks for Main VUMA Memory. Assign Main VUMA
`Memory to Fast Page Mode or EDO bank. Also, ifmntiliary VUMA Memory is
`assigned by operating system to Synchronous DRAM banks, do not use it. Either
`repeat the request for Auxiliary VUMA Memory till it is assigned to Fast Page
`Mode or EDO bank or use some alternate method.
`
`c. If VUMA device supports Synchronous DRAM, read < VUMA BIOS signature
`string (refer to VUMA VESA BIOS Extensions)> to find out ifVUMA device
`supports multiple banks access.
`d. If only single bank access supported on VUMA device, exit, as the Main
`VUMA Memor'yandAtndiiaryVUMAMemo1'ybankisfixed.
`e. If multiple banks access is supported and if the CS# for Synchronous DRAM
`bank is supported on VUMA device. assign the Main VUMA Memory to obtain
`the best possible system performance and t.
`
`5.4.] Programmable Parameters
`
`Synchronous DRAM: have trarious prograrninable paiimnetm. Core logic programs
`Synchronous DRAM parameters to obtain_" " the best possible
`The most eficient
`wayforVUMA devicetoprogramits DRAMcontrolleristomnkeaBIOSealltofind
`
`27
`
`I
`
`Version." 1.0p. Rev. 0.4;»
`
`Page 213 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 213
`
`

`
`vum Proposed sax ml
`
`. VESA Confidential
`
`om1hepa:m%eorclogichasdecidedmdpmg1'a|nilsDRAMoonunflerwi1hthe.
`samepm-mnetus.Altunamly,VUMAdevieeoouldpmgmnisDRAMcomuHcrwith
`omorafldifluunpm-mneters. lfVUMAdevicepmgrunsitsDRAMconnol1erwi1l1any
`difi'er¢ntpmanmus,hisVUMAd¢vice’srcsponm‘bfli1ymmpmg:amSym1numus
`DRAMbackwith_fl1e-ofiginalpflfimfitfiaabefomdmephysinlsymmmemaiyhmis
`handedofitocorclogic. InoIherwords,VUMAdeVin&~isfi'eetochangeanyora1lof1he
`para:neters,btnthechangcshoI1ldbcu-anspazenttocmtlogic.
`
`How core logic progpams vuinns pamueezgand how VU1v‘£A-device could inquire them
`is as follows:
`-
`'
`-
`
`.l.BurstLengfl|- Bm'stLengthcanbeprog1'ammedas1,2ar4.VUMAdevioeneeds
`IaomakeaBIOS_call<RemmMemarySpeedType(refertoVUMA
`VESA BIOS.Exn:nsions)> to find outthe Bust Length.
`2.CASLatency- AsCASlatcncydep:ndsanth:speedofSynchrunansDRAMused'
`andt.heciockspeed,.1hisst.Indarddoesnotvmn1iofixthis
`pa:a:n:ter.Core1o_gic.-progralnsthispammetertoanappmpziate
`valuz.VUMAdeviceneaistomakeaBIOScall<Rm:rnMcmory
`Speed Type (1-cferto VUMA VESA BIOS Ext:nsions}> to find out
`theCASlatency.
`3. Burst Ordering-Most efliciant Burstordenhgdependsuponthetypeof CPUused.
`VUMAdsviceneedsto'makeaBIOScall<Rctu:nMemorySpeed
`Type(refcrtoVUMAVESABIOSExt:nsians)>tnfindoz.1tth
`Burstorder.
`'
`
`5.4.2 Protocol Description and Timing
`
`AlltheDRAMsigna1sare-sharedbyeorclogic.andVUMAdeviee.Theym'ed1ivenby
`currentbusmaster.WhencorelogicandVUM.Adevieeha:1doverthe.bustoeachother.
`theymustdriveallthesharedsftlssigna15highforonI:CPUCLKc1ackand~th:nui-slate
`thcm.A1so, they should tri-stateallthesharedt/s signals.
`
`SynchronousDRAMsuepmchmgedbypechuge¢ommmd.Wh:nfl::physica1systan
`muno1ybusishsndedofi'&omcomlogicmVUMAdeviceorvicea-vu3a,fl:eDRAM
`prechargehasnmoptions:
`I
`_
`.
`'
`
`1.-Pmchargeboththe.ii1tema1banks.befomhandeofi'=.I}fisisasimplecasewhueboth
`theh1terna1banksofthcacfivesynchmmusDRAMbankmep:edm-gedmdthenfime
`busishandedofi‘.
`.
`-
`
`-
`
`2.RequesfingMastermodpsthephysica1systunmunoxyh:sudsynchrumusDRAM
`imunalbanksneednmbeprecharged-hfl:is_caseth:ruquesfingmastumwpsthe
`DRAMaddmssandmmoIsigna1smu=ck.the.9pmpggesinthehmmaIbanksofthe
`acfives)mchmnousDRAMbank.Theintanalbanksof1heacfiVesyndnunousDRAM
`
`23
`
`Version. 1.0p, Rev. 0.4]:
`
`Page 214 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 214
`
`

`
`VUMA. Proposed stam .1
`
`__
`
`)Es_A Confidential
`
`uenotprechargedwhenthephysicalsystemmemnrybusishanded-ofi‘mthe'
`mquesfingmastenlfmededthemquesfingmenunkescaeofprechugeafiugetfing
`thephysicalsystemmernotybus.
`'
`
`Both core logic and VUMA ‘device have an option of either-implementing or not
`impiemenfingDRAMmooprfeanneWhetha-corelogicandVUMAdevicesuppon
`DRAMn1oopornotshouldbersnsparunmflreoperatingsysmmandappfica1ign
`programs. To echievetheu'nnspuency,systemBIOS-mdVUMABIOS'needto findout
`ifboth'com1ogicandVUMAdwiumpponfiisfianncmdsctfi1e§ysmmappropfiamiy
`atboot.ThefoHowingalgofithmarplainshowitcanbeaehieved.Thealgofithmisonly
`includedtoexplainfl1efeamre.Referto1helstestVUMAVESABIOSEnensionsforthe
`mostupda1edBIOS calls:
`
`1. System BIOS reads <VUMA BIOS signature string (referto VUMA VESA BIOS
`Exzensions)>, tofindo1nifVUMAdevicecansnoopthephysiealsystem:ne1norybus.
`2.Ifno, Systernfllospmgmmscoieiogictoprechargesynein-anousDRAMbeforebus
`hand-off.
`
`3.Ifyes, SystemBI0Spmg1'amscorelogicnottopreehargesynchrunmrsDRAMbefore
`bus hand-off.
`
`4. V'U'I\r1ABIOSmakaca1L<F_~¢Pon_VUMA-core logic (refertoVUMA
`VESA BIOS E.xtensions)>,
`to find out ifcore logic can snoop the physical system
`memory bus.
`’
`S. Ifno, VUMA BIOS p1fograms'VUMA device to precharge synchronous DRAM before
`bus hand-ofi'.
`
`6. If yes. VUMA BIOS programs VUMA device not to preeharge synchronous DRAM
`beforebushand-off.
`
`None, only one, or both of core logic and VUMAI-device can sujiport this feature. When
`only oneof1hemsupponsthisfearmemerno1yprech:rgewfllbeasymmeuicali.e. there
`will be precharge before hand-ofi‘ one ivay and no precharge the other way.
`
`5.4.2.1 Non-Snoop Cases
`
`The sharedDR.A.M signals are drivenby core logic when it istheownerofthephysical
`sys1emmemorjrbus.VUMAdevicereques13thephysicalsystunmunorybusby
`assenmgMREQ#.BusArbher3nnsthebusbyassadngMGNT#.Also,befmeVUMA
`device starts driving theib-us‘, core logic should drive all the shared sitls signals high for
`one CPUCLK clock and triestete them. Core logic should also tri-stateall the shared tis
`signals. The tri-state condition on the bus should be for one CPUCLK clock, before
`VUMAde\&cestamdfifingthelms.1heseacfivifiesmeoVerlappedtoimpmve
`- performance as shown in Figure 5-7. Since VUMA device does not
`DRAM
`snoop feann-e, DRAM is precharged before heading off the physical system‘ ory bus
`as shown in Figure 5-7.
`.
`"
`
`29
`
`Version. 1.01:. Rev. O.4p
`
`Page 215 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 215
`
`

`
`VUMA Proposed Sta ml
`
`.- VESA Confidential
`
`Figures-7 Bushandoflfrom _Core
`
`or-ucu<
`
`
`
`MR.EQ# is driven low fi-om clock edge 2. Core logic samples it active on deck edge 3.
`Arbitercangivediebusrightaway, so corelogicgivesprechargecoinmandto DRAM
`from the same clock edge. Core logic drivesall theahared sftfs signals high Eon: clock
`edge 4. Core logic tri-states all the shared signals (sftfs and tie) and drives MGNT# active
`fi'omcloekedgeS.VUMAdeviocsan1plesMGNT#acfivcatclockedge6andst£u'ts
`driving the bus fromthe same edge.
`'
`
`ThesharedDRAMsignalsaredrivenhyVUMAdevicewhenitistheownerofthe
`physical system memory bus. VUMA device relinquishes the physical system memory
`bus by de-asserting MREQ#. Bus Arbiter gives the bus baekto core logic by de-asserting
`MGNT#. Also, as mentioned above, before core logic starts driving the bus, VUMA
`device should drive all the shared sft/s signals high for one CPUCLK clock and tri-state
`them. VUMA device should“ also tri-state all the shared tls signals"; The float condition on
`the bus should be for one CPUCLK clock, before core logic starts driving the bus. These
`activities are overlapped to improve
`as shown in Figure 5-8. Since core logic
`does _not support DRAM snoop feature, DRAM is precbarged before handing ofi' the
`physi.cal'systern'n1en1ory bus as shown in Figure 5-8.
`_
`‘
`
`Figure 5-8 Bus hand offfrom VUMA device to Core Logic
`
`
`
`VUMAdetdccgivesprechargewmmmdfiomdockedge3.Itdrivesa11.sbm'edsMs
`signals high from clock edge 4. It tri-states all shared signals (sftls and tfs) and de-asserts
`
`30
`
`'
`
`Versiori. 1.op. Rev. 0.4;:
`
`Page 216 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 216
`
`

`
`Vl_JMA Proposed Shndrri
`
`"EBA Confidential
`
`MREQ# fromclockedge s. Core_logicsar_nples MREQ#ioactiveonciockedge 5. Core
`logicdrivesallsharedsigna!sanddeassertsMGNT#fi'omclocl:edge6.
`
`5.4.2.2 Snoop Cases
`
`'l'hesl:aredDR.AM'signalsaredrivenbycorelogicwhenitistheownercfthephysieal
`systen1memor'ybusVUMAdevicerequststhephysia|laystemcrybusby
`asserting MR‘:-3Q#. Bus Arbiter grants the bus byuserting MGNT#. Also, before VUMA
`devioestartsd1ivingthebus,oore_logicshouldd1'ivealltbeshared:/tfssiytalshighfor
`oneCPUCLKdoekandni-smtemuncomlogicshouldalsoui-smmaflthesharedfis
`signnls.'I‘heIri-stateconditiononthebusshouldbeforoneCPUCLKclock.before
`VU'MAdev1ce' sta1'ts'drivingd1ebns.'I'hesec1:tivifiesareoverlappedtoitnprove
`~ performance as shown in Figure 5-9. Since VUMA devicesuppcrts DRAM snoop
`feature, core logic does not precharge DRAM before handing of the physical system
`memory busasshcwninfigure 5-9.
`
`rigms-snug handoflfi-omeorelogictoVUMAdeviee
`
`J 2
`Z 1
`is
`14
`is
`27
`is
`:9
`arm '\_r‘\J"\_/‘N./‘\_/—\_/‘\.r\_r'\_/*\_
`
`WEGI‘ '' ' '
`
`
`
`
`cm
`-
`Signals
`' u.
`
`'“
`
`.
`.
`.
`.
`h
`_5,-9..., __
`
`.
`-
`
`. .
`-
`
`MR.EQ# is driven low from clock edge 2. Core logic samples it active on "clock edge 3.
`Arbiter can give the bus right away and since VUMA device supports DRAM snoop
`feature, core logic drives all the shared s_/tfs signals high from the e clock edge. Core
`logic tri-states all the shared signals (sftfs and tfs) and drives MGNT# active fi'om clock
`edge 4. VUM.Ad.evice samples MGN'l'#'activeatc_iock edge 5 andstartsdrivingthebus
`from the same edge.
`"
`
`1hesbaredDRAMsignalsare_dtivenbyVUMAdevicewhenitistheom:erofthe
`physical system memory bus. VUMA device
`the physical syste.m memory
`bus by de-asserting MREQ#. Bus Arbiter ‘gives
`to corelogic by de-asserting
`MGNT#. Also, as mentioned above, before core ‘logic starts driving the bus, VUMA
`device should drive all the shared sftls signals high for one CPUCLK clock and tri-state
`them. VUMA device should also tri-state all the shared t/s signals. The float condition on
`the bus should be for one CPUCLK clock. before core logic starts driving the bus. These
`activities are overlapped to improve performance as shown in Figure 5-10. Since core
`
`31
`
`Version. 1.0p, Rev. 0.4p
`
`Page 217 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 217
`
`

`
`VUMA Proposed St:
`
`‘ard
`
`.
`
`VESA confidential
`
`not
`logic supports DRAM snoop feanne,
`handing ofi' the physical system memory bus as shownin Figure 5-10.
`
`DRAM before
`
`.
`
`Figure 5-10 Bus hand ofl’ from VUMA device to core logic
`
`
`
`VUMA device drives all shared s/t/s signals high fi'om clock edge 3. It tri-states all shared
`signals (s/t/s and t/s) and de-asserts MR.EQ# from clocl; edge 4. Core logic samples
`MREQ# inactive on clock edge 5. Core logic“ drives‘ all
`signals and deasserts
`MGNT# from clock edge 5.
`
`5.5 Memory Parity support
`
`Memory Parity support is optional on both core logic and VUMA device. If core logic
`supports parity it should be able to disable parity check for Main VUMAiMemory and
`Auxiliary VUMA Memory areas while parity check on the rest or the physical system
`memory is enabled.
`A
`
`5.6 Memory Controller Pin Multiplexing
`
`but are
`The logical interfaces for Fast Page, EDO and BEDO DR.AMs are very
`significantly different than that of Synchronous
`If motl1er%.l;oard designers Want
`to mix difierent DRAM technologies on the samemother board,‘core logic
`have to
`multiplex DRAM control signals. The meaning of a multiplexed signal will
`on the
`type of DRAM core logic is accessingat a given time. If a
`supports
`multiple hanks access
`of
`DRAM technologies, it
`also have;to
`multiplex DRAM ‘ccntrolpsignals. Bourooro logic and
`devices have to have
`same multiplexing scheme. The appropriate JEDEC standarclshould beefollowed for
`multiplexing scheme.
`
`d.0 Boot Protocol
`
`32
`
`Version. 1.01:. Rev. O.4p
`
`Page 218 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 218
`
`

`
`VUMA Proposed Stan
`
`:d
`
`VES_A Confidential
`
`6.1 Main VUMA Memory Access at Boot“
`
`In unified memory architecture, part of the physical system memory is assigned to Main '
`VUMA Memory. The
`operating
`are . not aware of unified memory
`architecture. Also, some of the existing operating systems size memory themselves. This
`poses a problem as the operating systems after sizing the total physical system memory,
`will assume that they could use all of the memory and might overwrite Main VUMA
`Memory. The solution to this probl is explained below:
`
`.
`
`As shown in Figure 6-1, the solution to this problem is to disable core logic access to
`Main VUMA Memory area at boot time. In that case even if operating system, sizes the
`memory, it will find only (total physical system memory - Main VUMA Memory) and
`will not be aware of the Main VUMA Memory existence. This will avoid operating
`~ system ever writing to the Main VUMA Memory area. If VUMA device supports.
`multiple banks access, it can access total physical system memory all the time. If VUMA
`device supports single bank access, it can access the bank of Main VUMA Memory all
`the time.
`
`If VUMA device is a graphics controller, it needs a special consideration. Video screen is
`required during boot and since core logic can not access the Main VUMA Memory, it can
`not write to it. The problem is solved by programming the graphics controller into a
`pseudo legacy mode. In this mode graphics controller treats Main VUMA Memory
`exactly the same way as in non. unified memory architecture situations i.e. as if it has its
`own separate frame bufier. So now, the total system looks just like a non unified memory
`architecture system and this mode is called as pseudo legacy mode. Core logic performs
`accesses to video through legacy video memory address space of A000:0 and B00020.
`These accesses go on the PCI bus. Graphics controller claims these cycles. Graphics
`controller still needs to arbitrate for the physical system memory bus. After getting the
`bus, graphics controller performs reads/writes to
`VUMA Memory (frame buffer).
`After the system boots, it is still in the pseudo legacy mode. When operating system calls
`display driver, the driver programs core logic to allowaccess to Main VUMA Memory
`and switches the system from pseudo legacy mode to unified memory architecture.
`
`In the case of other type of VUMA devices, device driver needs to program core logic to
`allow access to Main VUMA Memory.
`
`Figure 6-1 Pseudo Legacy Mode
`
`33
`
`Version. 1.0p, Rev. O.4p
`
`Page 219 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 219
`
`

`
`VUMA Proposed Sta
`
`ard
`
`. VESA confidential
`
`scram‘
`
`
`
`'1'hefol1owi.ngalgorithmsumsupthebootprocessii1thecaseofVUMAdevicebeing a
`graphicsccntroller:
`
`'
`1. System BIOS sizes the physical system memory.
`2. System BIOS reads the size of Main VUMA Memory at previous boot (where this
`valueisstoredisSystemBIOSdependeu1,butmedstobeinsomesortofnonvulatile
`memory).
`3. System BIOS programs its internal registersiao reflectthattotal memory available is
`[total physical system mory(fi'om step I) - Main VUMA
`at previous boo
`(fromst=p2)l-
`'
`4. System boots and operating system calls display driver.
`5. Display driver makes a System BIOS call; <EnablelDisable Main VUMA Mory
`(refer to VUMA VESA BIOS Extensions)>, to progrm core logic internal registers to
`reflect that it can access total physical system-memory.
`6. Display driver switches VUMA device to unified memory architecture mode.
`
`Even though core logic can not access Main VUMA Memory till the time display driver
`enables it, core logic is responsible for Main VUMA Memory refresh.
`
`VUMAdeviceshouldcIaimPCI-MasteraccessestoMsinVUMAMemoryfill display
`driver enables core logic access to that area. Core logic shouldclaim PCI-Master accesses
`to Main VUMA Moryafter display driverenables core logic access Iothatarea.
`
`5.2 Reset Stats
`
`0npoweronrueLbothcorelogicandVUMAdevicehavetheirImifiedmetnory
`architecture capabilities disabled. MR.EQ# is dc-asserted by VUMA device and MGN'l'#
`is dc-asserted by core logic. System BIOS can detect if VUMA device suppers unified
`memory architecture capabilities by reading <VUM.A BIOS signature string (refer to
`VUMA VESA BIOS Extensions)>.
`
`34
`
`Version. ‘Lop. Rev. 0.4;)
`
`Page 220 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 220
`
`

`
`7.0 Electrical Spoci”-ation
`
`7.‘! Signal Levels
`
`levels fo;'tI1earbitra1ion signals only. DRAM
`This section
`signal1evelsdependonthetypeofDRAMusedmdheneeoannotbespecifiedbythe
`
`MREQ#
`
`MGN'I'#
`
`CPUCLK
`
`output
`input
`
`output
`input
`
`output
`input
`
`7.2 AC Timing
`
`5v 'I'I'L or 3.31: LVTI1.
`Sv TTL for 51: bufi'er, 5v tolerant LVTIL for 3.3? l:Iufi'er
`
`Sv 'I'I'L or 3.31: I..V’I'I'L
`Sv "TIL for Str bums‘, Sv tolerant LVTTL for 3.3V buffer
`
`5v 'I'I'L or 3.31: LV'I'I'L
`5v TIL for Sv bu:fl"er, Sv tolerant LVTII. for 3.31: buffer
`
`This section describes the AC timing parameters fiorthe ubitration signals only. DRAM
`Actirning paramete1sdependonthet}'peofDRAMusedandhencecannotbespecified
`bythestanda:d.BothMREQ#areMGNT#timingparamete1sa:ewithrespectto
`CPUCLK rising edge.
`
`MR.EQ#
`
`output
`
`_
`‘input
`
`MGNT#
`
`output
`
`input
`"
`
`- 10 ns
`tClk to Out (max)
`- 2 ns
`tClk to 011: (min)
`SetuptimetSU(m.in)-3ns
`-Hold time}!-I (min)
`- 0 ns
`
`- 10 ns
`tClIt to Out (max)
`- 2 as
`ICE: to Ont (min)
`SetuptimetSU(min)-3 us
`Holdtimet!-I(min)
`-Ons
`
`crucuc
`
`output
`
`clock frequency (max) - 66.66 MHz
`
`7.2.1 Timing Budget
`
`A1-nargin forsignal fl:'ghttimeandc!ockskewisaddedtothetimingpatan1eters.:!:2nsis
`allowed for the total of CPUCLK skew and signal flight time. Worst ease timing budget
`calculations for setup and hold time are as follows:
`'
`
`35
`
`Version. 1.05:. Rev. 0.4p
`
`Page 221 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 221
`
`

`
`VUMA Prop-md 5|:
`
`art!
`
`' V234 ‘confidential
`
`7.2.1.1 Wont use for Setup t'ime_
`
`Figure?-I showsmewo£s:case'£orsemp1sm;.:cikmou;fligmimeandc1ockskew
`haveconvexgedtoredueeavailablesemptime.
`
`Figure 7-1 Worst eneefor setup time '
`
`Drier Output
`
`Receiver Iran
`
`Sarroing CFLICLK
`
`[tCI.Kto0ut(max)+f1ighttime+tSU(min)+nega1iveCPUCLKskew]sCPUCL1(
`period i.e. 15n.s@66.66MI-Iz.
`_
`[10:15 + flig.httime+ 3ns +nega1:ive CPUCL.Kskew] s I5ns
`[flight _time + negative CPUCLK skew] 5 2:5
`
`7.2.1.2 Worst case for Hold time
`
`Figure 7.2 shows the worst case for hold time. zcik to out and clock skew have
`converged to reduce availahie hold time. Positive fligln time number helps in this case
`andhenceitisa.ssI.nnedtobezero.
`-
`
`Figure 7-2 Wont ease for hold time
`
`DliVlt_Ol.lDlI.
`Rnniver raw!
`
`Snuplag CPUCLK
`
`36
`
`Version. 1.0p. Rev. o.4p
`
`Page 222 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 222
`
`

`
`WI’-‘IA Proposed Sh:
`
`.111
`
`'vssA Confidential
`
`[posmve" crUcLIcst:ew+:H(min)]sucLxuoom(m:n)" ‘
`[positiveCPUCLKskew+0ns]S 2m
`positiveCPUCLKsk:ws2ns
`-
`
`7.3 Puflups
`
`Aflsfflssignusmedfilmtipswsmmintheinacfivssummfilhiomeragmtdrivesmem.
`Corelogichas:oprovidepullI:psfarallti1esv'tfssiy:als.VUMAdevicehasasoptionof --
`providing pullups on-some of the sftls signals. Alltfs signals need pulldnwns. Core logic
`hasmprofidepufldowmforali:h:fissigndsVUMAdsviéehuasopfionofprofiding
`pu1mowmmthefissignflsP1mupsandpufldowmmu1d'eifl1uhehncmalmthcchips
`orextcrnalonboard.
`
`Corelogicisresponsibie1_‘orpull_nps on_DRAMAcldress lines.
`DRAM Addr¢ss-
`DRAM control signals - Cor: logic is responsible for
`an DRAM control signals.
`VUMAdcvinehasm?optinnofprovidingpullupsonthem.
`Coniogicisrcspom:1'b1eforpull<i_ov.rnsonDRAMda1abus.
`VUMAdev-icehasasoption ofprnviding puiidownsonthem.
`
`DRAMDataBns-
`
`Pullupsandpulldownsareusedm
`signalsancihenceneedtobeweak. Rncommencledvaluefnrupulillpsmadpulidowns is
`betweenfllkohmandflokohm.
`'
`.
`
`7.4 Straps
`
`As some VUMA devices and con: lngic chips use DRAM data bus for straps, DRAM
`data bus-needs to be assigned for straps for difiemm controllers. The assignment of
`DRAM Data Bus for straps is as follows:
`‘
`
`VUMA device on Mothnrboani
`MD [0:19]
`MD [20:55] Reserved
`MD [55:53] Car: Logic
`
`Allthesnapsncedtobeplfllupsoflokohm.
`
`i'.5 DRAM Driver characteristics
`
`LoadingplaysauificslioleinDRAM'a§¢¢8sfimhg._hcaseofPCmothwboufisend
`‘ususcanecpmdfiieadsfingmanoryofasysmmby-addinggmaSIMMsHence,
`typicaIIythetota1DR.AM signal loadingisnotoonstantanglroouldvarysignificanfly.
`BothCoreLogicandVU1§/mdcvicemustbeablctodiivethemasdmiimloadthatthe
`
`37
`
`I
`
`Version. 1.0p, Rev. 0.4;:
`
`Page 223 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 223
`
`

`
`VUMA Proposed Sh
`
`‘ard
`
`.
`
`VESA Confidential
`
`motherboard -designs DRAM
`system motherboard is designed to accommodate. In
`signal loading can be excessive (on the orda of lO00pF for some signals)‘ and hence care
`must be taken for DRAM driver selection. Some general guide for DRAM driver.
`design are as follows:
`W,
`p

`
`Slew-rate controlled drivers are recommended. Drivers with selectable current drive (such
`as 8/16 mA drivers) may be used. This can reduce overshoot and undershoot associated
`with over-driving lightly loaded signalsand can prevent excessive rise and fall time delay .
`due to not providing enough current drive on heavily loaded signals.
`
`As shown in Figure 7-3, buflers may be placed on the systemgmotherboardp to reduce the
`per signal loading and/or provide larger drive strength capabilities. DRAM Write Enable
`and DRAM Address signals are typically the most heavily loaded signals. Column
`Address Strobe signals may also become overloaded when more than two DRAM banks
`are designed into a system. 'I'l'L or CMOS bufi'ers,(typicalIy 244 type) may be used to
`isolate and duplicate heavily loaded signals on a per bank basis.
`244 type buifers
`typically have very good drive characteristics as well and can be used to drive all of the
`heavily loaded DRAM control signals if the Core Logic and/or VUMA device has
`relatively weak drive characteristics. If external bufiers are used, the bufier delays should
`be taken in to timing considerations.
`
`Figure 7-3 Optional Bufiers for DRAM Signals
`
`PCI Bus
`
`W
`Cora Logic
`
`‘
`
`.
`
`(._2:’_':_
`
`g
`
`VUMA
`
`Wider DRAM devices offer reduced system loading on some ofthe control signals. x4
`DRAMS require four times the physical connections on RAS, MA» (Address), and write
`enables as x16 DRAMS. The reduction in loading can be significant Ifthe
`has
`control over the DRAMs which will be used in the system, the DRAM width should be
`chosen to provide the least loading.
`
`33
`
`Version. 1.Dp, Rev.
`
`Page 224 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 224
`
`

`
`VESA®
`
`VUMA I Proposal
`
`(Draft)
`Video Electronics Standards Association
`
`1150 North First Street, Suite 440
`San Jose, CA 95131-2029
`
`.
`
`Phone: (408) 435-0333
`‘FAX: (408) 435-8225 '
`
`
`
`VESA Unified Memory Architecture
`VESA BIOS Extensions (‘VUMA-SBE)
`‘
`r
`r
`r Proposal _
`
` Version: 1.0
`
`
`
`
`
`Document Revision: 2.2p
`November 1, 1995
`
` Important Notice: This is a draft documt from the Vi_deopEIectronics Standards
`Association (VBSA) Unified Memory Architetitzne Committee (VUMA). 1: is only for
`
`discussion purposes within the committee and with any‘ other persons or _orga.nizafions
`that the committee has determined should be invited to review or otherwise contribute to
`it.
`It hm not been presented or ratified by the VESAigene1-al membership.
`’
`
`
`
` Purpose
`
`
`wilhoul specific knowledge or direct hardware access.
`
`To allow the video BIOS and other GUI specific software to control the VUMA hardware
`
`
`
`Summary
`
`
`This document contains a specification for a system and video BIOS interface, VUMA-
`SVBE. The VUMA-SVBE interface will allow the video BIOS and other GUI specific
`
`software to comrol the VUMA hardware without specific knowledge or direct hardware
`access. The hardware protocol is described in VESA document VUMA 1.0.
`I
`
`
`
`Page 225 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 225
`
`

`
`VUMA Proposed Sh. Jard
`
`VESA oonfiaontial
`
`scope
`
`Becauselhisisadrafi-dounnem.itmmmbeoonsiduodcomplmoracmnteinafl'
`mspectsalthoughevcryeiforthasbeenmademmixfinxizzeenm-s.
`
`Intellectual Property
`
`o Copyright 1995 -' Video Electronics Standards Assooiafion. Duplication of
`document within VESA mombercompaniesforretriewpmposes is pormittod. All other
`rights arc reserved.
`'
`
`' Trademarks
`
`All uademarks usedinthisdocumentarcthepropenyoftheirrespectiveowncrs. VESA
`a.ndVUM.Aaretradema.rhs ownedby‘ti:eVidooEiocl:'onicsSta.ndarrisAssocia1ion.
`
`Patents
`
`The proposals and standards developed and adopted by VESA are'in1c.nded to promote
`uniformity and economics of scale in the video elscn-onics industry. VESA strives for
`standardsthatwil] bcnefitbothltheindnisuyaiidedduscnsofvidoodocuonicsproducts.
`VESA carmotenmnethé1theadopfionofgs1_::idud;fl1euseofame$boddoom'bodasa
`standard: or the making. u;ing.'q1t.se1lJ_'ng ofa product in',_oompliance_ with the standm-d
`does not infringe upon the
`(including patents, utdomarks, and
`cop_\'1'igl-us) of others. VESA, thereforfimakes no
`or implied, that
`products conforrning to a VESA standard do not in:&inge on tho intoiloctual property
`rights of othcrs. and accepts no liability direct.
`or consequential. for any such
`infringemcnt.
`7
`
`3
`
`Version. -‘Lop, Rev. 2.29
`
`Page 226 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 226
`
`

`
`’VUMA Proposed Standai...
`
`T
`
`T VESA confidential
`
`Support For This Specification
`
`If you have a product that incorporates VUMATM, you should ask the company that
`manufactured your product for assistance.
`If you are a manufacturer of the product,
`VESA can assist you with any clarification thatyou may require. All questions must be
`sent in writing to VESA via:
`'
`
`(The following list is the preferred order for contacting VESA.)
`
`VESA World Wide Web Page: www.ve.ra. org
`
`Fax:
`
`Mail:
`
`Acknowledgments
`
`(408) 435-8225
`
`VESA
`
`2150 North First Street
`Suite 440
`
`San Jose, California 95131-2029
`
`This document would not have been possible without the efforts of the members of the
`1995 VESA Unified Memory Architecture Committee and the professional support of the
`VESA staff.
`
`Work Group Members
`
`Any industry standard requires information from many sources. The following list
`recognizes members of the VUMA Committee, which was responsible for combining all
`of the indust‘r_v input into this proposal.
`
`VUMA Chairperson .
`
`Rajesh Shakkarwar
`
`OPTi
`
`Software work group Members
`
`Tim Crawford
`Phil. Mummah
`Josh Grossman
`Christopher Rhodes
`
`,
`
`Work group leader. Cirrus Logic
`Phoenix Technologies
`53. Inc.
`Award Software
`
`3
`
`Version. 1.op, Rev. 2.2p
`
`Page 227 of 280
`
`Petitioners HTC & LG — Exhibit 1002,
`
`Petitioners HTC & LG - Exhibit 1002, p. 227
`
`

`
`VUMA Proposed Sta
`
`ltd
`

`
`.
`
`v5s_A cgmfidgnfiag
`
`Revision History
`
`Initial Revision
`Rev .7
`Rev .3
`
`1
`
`1
`
`Aug 14 '95
`Aug 24 '95
`Aug 28 95
`
`Rev .9
`
`Aug 28 '95
`'
`V
`AddednsmwomF.modificdkmesndmovedsornennsimipfimsranovere£aa:oes
`tomhndnuVUM.Ammory,r:aovedflne§onfixunVUMAnanay,modified
`speedltype function
`
`Rev 1.0
`AddodVU'MAdevioeDRAMnlPPOrLSDRAMpu':meie::infimc:ion6
`
`Revl.1
`
`1
`Add items to Boot sequence
`Remove ufunczion all mdeiexnedupothasv
`
`1
`Rev 1.2
`Add changes suggested in previous
`
`'
`

`
`Rev 1.3
`Added 32 bi1VF. Idd:dAnxflnIdifllIS.Ad&d I6 bixproteaedmode UF.
`Modified seveni ftmctions
`
`Rev 1.4
`Modified assumptions. issiis. updated some Emotions.
`’
`
`Rev 1.5
`Modified the able ofcomam
`
`Rev 1.6
`Minor modificauons to assumptions
`
`Rev 1.7
`Changes to assumptions. dc goals.
`Made cnzges in some functions per discusion I! last comrninee meeting.
`~
`.
`-
`.
`Made changes to how memory is reported. funes 1 & 6. Modified ROM sigmnire.
`
`Rev 1.8
`
`1
`Rev 1.9
`Modified funcuon 6 and assumptions. Other minor typos fixed.
`
`*
`Rev 2.0
`Modified Assumptions C. J. L. N. Added Issumptions O. P. Q. R. Removed issue 4.
`Added error code 5. 6. Changed unit oirnernory in ROM signsnoe fiom 1K to 64K
`Modified register definition in function 0.1.2. 6. and 7. Other minortypos fixed.’ '
`
`Sept 3 ‘95
`
`Sept11'95
`
`Sept 22 *95
`
`Sept 29 '95
`
`Oct 6 '95
`
`Oct 6 '95
`
`Oct 10 '95
`
`Oct 16 '95
`
`‘
`Oct 18 '95
`
`'
`

`
`Oct 18 '95
`
`Oct 23 ‘95
`
`Rev 2.

This document is available on Docket Alarm but you must sign up to view it.


Or .

Accessing this document will incur an additional charge of $.

After purchase, you can access this document again without charge.

Accept $ Charge
throbber

Still Working On It

This document is taking longer than usual to download. This can happen if we need to contact the court directly to obtain the document and their servers are running slowly.

Give it another minute or two to complete, and then try the refresh button.

throbber

A few More Minutes ... Still Working

It can take up to 5 minutes for us to download a document if the court servers are running slowly.

Thank you for your continued patience.

This document could not be displayed.

We could not find this document within its docket. Please go back to the docket page and check the link. If that does not work, go back to the docket and refresh it to pull the newest information.

Your account does not support viewing this document.

You need a Paid Account to view this document. Click here to change your account type.

Your account does not support viewing this document.

Set your membership status to view this document.

With a Docket Alarm membership, you'll get a whole lot more, including:

  • Up-to-date information for this case.
  • Email alerts whenever there is an update.
  • Full text search for other cases.
  • Get email alerts whenever a new case matches your search.

Become a Member

One Moment Please

The filing “” is large (MB) and is being downloaded.

Please refresh this page in a few minutes to see if the filing has been downloaded. The filing will also be emailed to you when the download completes.

Your document is on its way!

If you do not receive the document in five minutes, contact support at support@docketalarm.com.

Sealed Document

We are unable to display this document, it may be under a court ordered seal.

If you have proper credentials to access the file, you may proceed directly to the court's system using your government issued username and password.


Access Government Site

We are redirecting you
to a mobile optimized page.





Document Unreadable or Corrupt

Refresh this Document
Go to the Docket

We are unable to display this document.

Refresh this Document
Go to the Docket