Microunity Systems Engineering, Inc v. Acer Inc. et al

Filing 1

COMPLAINT for Patent Infringement against AT&T Inc, AT&T Mobility LLC, Acer America Corporation, Acer Inc., Apple Inc, Cellco Partnership, Exedea Inc, Google Inc., HTC America Inc., HTC Corporation, LG Electronics Inc., LG Electronics MobileComm USA, Inc., Motorola Inc.,, Nokia Corporation, Nokia Inc., Palm Inc., Qualcomm Inc., Samsung Electronics Co., LTD.,, Samsung Semiconductor Inc., Samsung Telecommunications America LLC, Sprint Nextel Corporation, Texas Instruments Inc. ( Filing fee $ 350 receipt number 0540-2513044.), filed by Microunity Systems Engineering, Inc. (Attachments: # 1 Civil Cover Sheet, # 2 Exhibit C, # 3 Exhibit C-1, # 4 Exhibit U part 1, # 5 Exhibit U part 2, # 6 Exhibit U part 3, # 7 Exhibit U part 4, # 8 Exhibit U part 5)(Susman, Stephen)

Download PDF

Microunity Systems Engineering, Inc v. Acer Inc. et al Doc. 1 Att. 2 Exhibit C Dockets.Justia.com r ffi illilr il lffi �lll llil lil lfi| flln ro ffi ru il � us0057428404 � United States Patent Hansen et al. t54] GEI\�ERA.L PURPOSE, MUI.,il�PLE rrel tul Patent Number: t4st Date of Patent: 5,742r8l,0 APr. 21' 1�8 PRECIf|ION PARALLEL OPERATION, PROGR.AMMABLE MEDIA PR,OCESSOR Inventors: Creig Eensen Los Altos; John Moussouris. Palo �Ito. both of Calif' Assignee: lldcrountty Systems ['.nglnssdng' Itrc.. Sunnyvale. Calif. AppL No.:516'036 Hewlet-Packard Co.. 'PA-RXSC 1.1 Architecture and Instuction Set". Manual Part No. 09740-90039. (1990)' MIPS Cornput� Sys�ms. tnc.. "MIPS R4{�0 Uscr's Manual". Mfg. Part No. M8-O0�10. (1990)' Pr�mary Examiner-Npsh M. Shah Attomey Agent, or Fin�-McDc�mott. WilI & Emery f75l f�J f?ll t57l aBfiTRACT l22l Filed: ,{ug. 16' 1995 Gl�6F 7�t t51l rnL cL� 3e5�(X); 3&1736;364t232.8; t52l U.S. Cl. t581 Fidrl of Se�rctr 3�1258t 364/DIG' t .-...'.. 3&18�.775. 3&1325. 650. � r-�4. 736. 7 4t. 7 45. 7 48. 7 54. 7 61. 7 68. 67 0. 37 6, 280 A general purpose. progra�Dm�ble media processor for pro' cessing and transmining a media d�la s�eam of audio. video, radio. gapl�cs. encr�ption. authentication. a�d networking fufo�nation in real-time. The media Processor incorporates an execution unit that rnai�tains sub'stattially peak data throughout of media data strearns. The execution unit includes a dynamically prtionable multi-precision aritbmetic un�t. programmable switdl and progranunable extended matheroatical element. A h�Sh bandwidth external interface supplies media data suams at substutia[y peak t56l Rderences Clted U.S. PATENT DOCI.JMEI.ITS rates to a general purpose register �le and the multi- 4,8n267 Ul990 Al$�p 5,426,ffi 6/195 4,9?5,E6 lZl990 F�ec�*se� '." 3641748 52O1,056 4/1993 Danieletel. ........,'....'...'......" 395/800 526E,E55 t2l1993 Mason et al, ,.........-.."...".".'. Nakagawaetal.'.'....."........... OTHER PI.JBLICATTONS TF'FR Draft St�nd�d et al. .......'.'....".....'..'.-' ffi1745 3&n� 3&nU fc "Scalable Cohereot Interface- lnw-Voltage Difierential Signal Specifcations a�d Packet Encoding". rEFF Standards Depart�Dent, P15963/�0.15 (Mar. 1992). rF'.FF Dhaft Stand�rd for "lligh-Bandwidtb Memory In�rface Based on SCI SipalingTechnology (Ramlink)". rnnn precision erecution uniL A memory rnaDagemetrt unit. and instruction and data cache/bufiers are also providcd. High bandwidth memory cotrtrollers ue linked in serics to provide a memory ch�nnel to the general purpose. Program' mable mcd�a process�. The gencral purpose. programrnable media proce ssor is disposcd in a �etwmkfabric consisting of frber qptic cable. coaxial cable a�d twisted pair wircs to transmit. poccss and receive single or unifed media dst� s�e�ms. Paratlel general purpose mcdia processors are disposed througbout �e Betwork itr a distributed virtual rna�ner to allow for multi-procrssor operations and sharing of Sta�dards Dpar[nent, Draft 1.25 rnEe P1596.4-199X (May 1995). Gtrry Kane et al.. 'MIPS R�SCArcfiitecture", Prentice Hall (199s). resources through thc �etwcrk A me�bod for receiving. processing and trancmi�i�g medi� rlata sEearns over the co�n�nunications fabric is also providcd l1 Cl�lms, 25 Drewtng Sheets Micro�che Appcndix Included (4 Micro�chc. 387 Pagcs) IBM. 'The PowerPC Ardritecture: A Specifrcation For lr New Family of Risc Processors". 2�d Ed,. Morgan lftufma�� Publishers, Inc.. (1994). l1 a (n a F� FIG, I '.r l� 28 � F} (D trl fl F} 18'KU BAND DBS tc � �1 WIRELESS NETWORK CORDLESS PHONES PERSONAL COMMUNICATOHS REMOTE CONTROLS TELECOMMUTING ts \o \o � \� POTS,ISDN, ADSL NETWORK ct) (D tD CABLE OR FIBER 10/ PERSONAL COMPUTER SECOND TV WORKSTATION DIGITAL CABLE TELEVISION DIGITAL VCB VIDEO ON DEMAND CD.ROM ELECTRONIC YELLOW PAGES INTEGRATED MESSAGE SYSTEM VIDEOPHONE VIDEOGAMES ELECTRONIC PROGRAM GUIDE � bJ UI (,t \� � I b.) � � U.S. Patent |ryr.21,1998 -\n )'= Sheet 2 of 25 5,7421840 � =\rre q�H .<r \ H nr oll Y3 .^u lwr s\ Y'5ltr u \ �t C?I t1 a V) a FIG.3 (BTTSiSECOND) 1 4 1664256 K 1 4 16642s6 I 1 4 1664256 1 4 MG �, F} FY I rs t! 1664256 -� F TELECOM E MODEMS � E n l.J TV MOVI \o \� � (t) E (! (� (, l{! l\) t� 1 4 1664256 1 4 1664256 K 1 4 1664256 1 4 1664256 MG I (,t \h� , { � � � A e a V) a FIG. 4 H 'Y (Brrs/sEcoND) K 1 1 4 1664256 4 1664256 I 1 4 1664256 1 4 MG �l 1664256 F} (� I P (+ TELECOM E MODEMS � !r \J \o \o � V) (D o 5 �� bJ UI 60 '/ 1 4 1664256 1 4 1664256 KM 1 4 1664256 1 4 1664256 U (Jr GT b �J \� � � - � ta V) o H .Y FIG,5 �e FI o FI FI � E MtcRoPROCESSOR r - lr bJ \o \o � 1OO GBITS/SE ct) (! THE FUTURE OF THE OLD WAY (1999) � tt o� rr N) UI . S]NGLE CHIP . LOWER COST \. (r� b u .) -I 5 � � t4 a (a a FIG,6 MEDIAPROCESSOR 4OO F� �e FI (D FI -} F GBIT/SECOND � !l �\) \o \� � U) THE FUTURE OF THE UNIFIED WAY (1e95) o � o\ Hl l'� . SMALLER CHIP tt . MUCH LOWER COST u (,t { � � I b.) � � a (t) a FIG, 7 AUDIO VIDEO 134 RADIO NET 138 F� � F} (D E � Fl\ NYQUIST SAMPLED r/0 140 � � TD INSTR'N BUFFERi CACHE 118 REGS 110 � �r hJ z. \o \a � �APE DISK 142 a PCt STANDARD MEMORY AND I/O I � (f { = rn 1t 'T c) U) ETLB & TAGS BR&GATE114 L,S & SYNCH 116 (n 122 MULTIPREClSION DATA BUFFEFI/ CACHE 120 o � \� b.J tt rn SWITCH 104 12! t ut h � .J -'l � � � f.J.S, Patent Apr.21, 1998 Sheet I of 25 5r742,840 FlG,8(a) 274 FtG. g(�) 274 [.J.S. Patent Apr. 2L, 1998 Sheet 9 of 25 s,742,840 FlG, 8(c) GROUP DEAL: 128 BITS TO 128 BITS GHOUP SHUFFLE : 128 BITS TO 128 BITS FtG. 8(d) i t I (n FlG. 8(e) 2.MUX 4.MUX B-MUX F� �l (D -} F X BENES NETWORK : MULTISTAGE PERMUTATION 4-MUX, 3 �l !r b.J \o \o � STAGE 4-MUX, TWO D|MENS]oNAL (3-STAGE) XI D< t.za��. ffiffiffi + V) t� l! F� h� 16-MUX, TWO DTMENSTONAL (3-STAGE) (,I XI IX I ..> II I + + + + -> -> U �I + <. + + + <' t tt t + \� � v l.J � � LJ.S. Patent Apr.21, 1998 Sheet 11 of 25 5,7421840 FlG.9(a) 280 31 24 23 1B 17 280 31 24 23 18 17 12 11 280 31 12 24 23 18 17 12 11 65 IJ.S. Patent Apr.21, 1998 Sheet t2 of 25 5;742,840 � c (� -o c (� c) .E () .(9 n := q) ct) '= 'ut \\, o 'c 6 ct) -� o o E o) c! F(� c\� (\I q) = o o x) q *Q � c\� r (o r\\� < IJ.S. Patent Apr. 21, 1998 Sheet 13 of 25 5,7421840 y q) c) (l) o o) x *s N t4 FlG. r0�) LOCAL VIRTUAL ADDRESS a U) a F� ʡ F� tD FI i Flt ADDRESS LOCAL VIRTUAL TO GLOBAL VIRTUAL ADDRESS TRANSLATION ADDRESS CACHE DATA rc !r DATA LOCAL PROTECTION 168 � \) \o \o � GLOBAL VIRTUAL ADDRESS GLOBAL PROTECTION HIT v) (! (D 5 f{� GLOBAL VIRTUAL TO PHYSICAL ADDRESS TRANSLATION b� ut GLOBAL PROTECTION (JI I { � � PHYSICAL ADDRESS v b,� � 14 I (n a FtG. r0(�) 1148 VIRTUAL F� �. r+ (D ADDRESS x J F} 230 � T �.) \� \o � U) o (D (r� Frr h) tt U �r PHYSICAL ADDRESS 244 I \t � � b,.) � LJ.S. Patent Apr. 21, 1998 Sheet 16 of 25 5r7421840 \ \ r�S �\\ [I.S. Patent Apr.21, 1998 Sheet 17 of 25 5,7421840 � F z. :f r, I l( (J I F LlJ l{ I LL o tz--. t) (r = (J ff F lU) tU z. l! = s O (\I N $ l-1 a V) I t+i Y �, F} (D FIG, I3 FI Fr. - � F h) 64 KBYTE CACHE 64 KBY�E CACHE \0 � \� �g--/' FOUR 12 Bn' (t) o (! ADDRESS 72BIT DATA BUS +4BIT CONTROL BUSSES � rrl t\) (r� v (JI \� � b.) � � (1 a (t) t l+l Y p (D FI FF Fl. - FlG. 14 :r b.� \c \C) � v) o o 154 154 154 154 154 154 54 154 \o l{� b.) UI 3264-BtT PCt (Jr \. u \� � b, � � [I.S. Patent Apr.21, 1998 Sheet 2O of25 5,7421840 FlG, l5 oo aa OO a It 8 BITS DATA CE o t� cn J a D7 . ,178 aa aa O E TU )� (n fJ.S. Patent Apr. 21, 1998 Sheet 2l of 25 5,7421840 Lll () () $ H �. U) L (U ) o) (� o- or ct E LU = LU trl o- (4 a V) a r6(�) MICROPROCESSOR WCACHE � 7t F} H Y (D T J F� T N) \� \o � V) (D tD bJ �.) r+ N t,I 8 MEMORY CHANNELS 4 HIBRAM/RING 216 \--Y-J gTH RING FOR MP 3yo CHANNELS (Jr I 210 212 214 { � � v b,J � LJ.S. Patent Apr. 21, 1998 Sheet 23 of25 5,7421840 urg HUHk <q Y=Ux'H� FHi�Xio+ Q� ==E t�(TJ\-r<2. U) LU �tr�s#�=s Or�JC�$tf)(Ol\ ��E��F z. oo (,t) LU �. U) ul � uJ ctr o LU ff, ctr LU ctr o E \ N IJ.S. Patent Apr.21, 1�8 Sheet 24 of 25 5r7421840 FIG. r8�) 12 12 12 220 12 218 OR 12 218 FtG. r0(b) '.� 218 12 222 12 FIG, r8(c) l4 FTG, I9 t (n a tYt .s �l F} (!I F F} - 120 12 100 �3 � b,) lbg 64 KBYTE CACHE DRAM INTERFACE �r \� \o � MEMORY PROCESSOR 1 GHz BYTE V) o tD l.) tt f{� bJ ur 262 266 32 KBYTE BUFFER 226 (Jt 260 268 u v \� A b.) � SWITCH � 5.742.840 I GENER�L PIJRPOSE, MTJIJTTPLE PRECISION PARALLEL OPER�TION' PROGRAMMABLE ME)IA PROCESSOR 2 mod"ms have developed atr-elaborate hier�chy of algo' ritbms implemen�di� DSP softwa�e. w�th automatic reduction of daia rates from 28'8 Kbits/sec to 19.6 I(bitVsec. 14'4 Kbitlsec. or muc� lower rates as needed to �cconmodate noise. echoes. and other impairments in the cop'per plant. To implement similar algorithms on an ASIC-based hoadband modem is fa� more complex to achieve in sof��'are. ciallv whcn viewcd i� the upstream directiou. Telephony ,{ Micro�che Appendix consistilg of 4 sheets (387 tot�l frames) of mic�ofiche is included in this application. Tfte Microfidre Appendix contai�s material which is subject to < - copyright protection. The copyrigbt owner has no objection Tbese problems of cost. conplexity. and rigidity are to the facsimile reproduction by any one of the Miclo�c�e �g compoundd �rther in mae cornplete tro�dba�d devices the Patent andTradema�t O�ce -Appendix. as it appears video such as digital set-toP boxes. multimedia PCs. patetrt files � recsds. but otherwise reserves all copyright conferencing equ�ment. all ofwhich go beyond the b�sic whatsovr. fights radio �equenq.C'nF1 mo�em functions to include a broad rangc of audio and video compression a�d decoding INVENTION FIELD OFTIIE t5 a��rithrns. along with remotc control and graphical user -This invention relates to tbe frdd of co�nrnunications inierfac�. Softwarc for thcse dcvices must control what pocessing. and more particularly. to a method and �ppa�atus amouots to a hterogeneors mutt�-processor. whre each ior real-time processing of mr�lti-mcdia digital comrnunicaspecialized Processor h�s a differenl and usually eccentric tions. or primitive. programming enviro�menl Even if thesc pr*' of pro' 2p gramming envi�onments are mastered. the degree BACKGROI.JND OF THE INVENTION grarnmabitity is timited- For example. Motion Picture Expert Optical �bcr and discs have made t[e transm�ssio! and Croupl ('MFEG-� ch�s m�nufac��red byAf&T C�postorage of digital infcrroation both cheaper and casier than ration will Dot impleme�t adva�ces such as fr�ctal- and older analog �drnologies' An irnproved syst�m for digit�l wavele�-bascd corn�ression algorithms. but these chips are pocessing of med�a data shearns is necessary in order to 25 not readily softwarc upgradcablc to the MPEG-tr st�rdardrselize 1�s full po�ntiat of these adva�ced media, Abroadband uetworkoperator who leases an MPEG ASICbased product is therefse at risk of having to continuusly For the past century. telqlhone service delivered over upgr"d" his system by purchasing significant amounts of copper twistcd pair has been the lingua franca of co�nmun�w hardwa�e just to track the evolution of MPEG stannications. Over the next century. b�oadband serrtices delivered ovcr optical frber and coax will �nore completcly �tlfrll � dards. The high cost of ASIC-based media processing results the human need for scnsory infor�nation by supplyilg voice. video. and d�ta at rates of about 1.000 times greater than from inefEciencies in both rn�mory and logic A typic�t narow band telephony, Current gencral-�xfpose micrcpro ASIC consists of a mult�licity of specialized logic blocks. cesso�s aud digital signal processo� ('DSPs") can ha�dte cach with a small memory dedica�d to holditrg the data digital voice. data. alrd funages at narrow band rates. but they 3s which cornprises the working set for that block- The silicon are rvay too slow for processing rnedia data at broadba�d arca of thesc multiple small memories is further increased by rates. thc overbead of mult�le dccoders. sense arnplifiers. write drivers, etc. required for eacir logic bloch Tbe logic blocks This shq�all i� d�ital pocessing of troadband media is are also cons�ained to qperate at frequencies deterru�ned by cr�rently being addres�d through the design of many difappl�cation-spec��c in�gra�d cl�cuits � the i�tern�l symbol r�tes of b�oadband algorithrns in �drr to ferent hinds of avoid additional bufer memories' These �equencies typi("ASICs"). F� exarnple. a prototypical broadband dwice opcrating point of cally differ from the suc*r as a cable modem modulates and demodulates �ligi��[ a given semiconductc�r technolos�. Interconnect and syndata at rates up to 45 Mbitlsec within a single 6 MIIZ c�ble chronization of the rnany logic and memory blocks �re also cha��el (as compred to rates of 28.8 KbitVsec q�ithi� a 6 KHz d�an�el for telephone modems) and transcodes it on�o +s maj� sources of overhead in �re ASIC aprproach. The disadvantages of the prior ASIC 4proac� can be over a 10/100 base'I co�nection to a prsonal coruputer ('?C") or come by I sirsle u�i�ed media prossor' The cost edvatrworkstation Current cable modcrns thus rcceive daa �om tages of such a unified lrooessor can be achieved by a coaxial cable cpnncctio� ttuough a chain of sPoci�lized ga�ering att Oe rnany ASC functions of a broadband rnedia ASIC devices i-n order to accomplish Quadra�re Amplitudc Modific�tion ('QAMT dcmodulation. Reed-Solomon errm 50 product into a single in�gnted circuit. Cost reduction is cor�ection. packet �ltering. Data Encryption Standa�d ft�rther i�crcased by reducing tl�e total memory a�e� of such a circuit by rqlacing thc multiplic� of small ASIC memoC'DES") decryption. a�d Bthernet protocol handling. The ries with a singlc memory hierarchy large enough to accomcable modems slso Ea�smit df,t� to the coa:rial cable link modate the sum tdal of all thc worti:rg scts. 8�d wide through a second cha�n of devices to ach�eve DES en��flion. Reed-Solomon block encoding. and Qua�rnary ss enogh to supply thc aggrcgatc ba�dwidh necds of all the il s - - optimum Phase Shift Keying C'QPSK") modulation. In these environmcnts. a general-purposc Processor is usually required as well in order to perfonn initieliz�tion. st�tistics circuitry to this memcry hierardry nr�y be streanlined by providing a generally programrnable swi�hing fahic. Many �f the logc blocks themselves c�D also r�laced with a collection. diagnosics. and network ��nagement functions. The ASIC aprproadr to media proccssing has three fun- co single multi-precision arithmetic utrit. which can bc internally partitioned under sofhpare control to perform addition. damental flaws: cost. cnmplexity. and rigidity' Thc commult�lication" divi�on. and other in�ger and floating point bined silicon area of all the specialized ASIC deviccs aritb�nctic qrerations on symbol streams of varying widths. required in the cable modern for exarnple. �esults i� a while sustaini�g �re full data throughpttt of thc memory incorDpatible with the per subscriber price coml�nent cost taryet ftr a cable service. The cable plant itself is a very 65 h�rfrdry. The residue of logic blocks that perform opcrations that are neither arith�nctic or permutation group orihostile service environmenl u,�ith noise ingress. reflections. etrted catr bc replaced with an extendcd rnath unit that nonlinea arnpli�ers. and other channel irnPairments. espe- logic blocks, Additionally. the logic block i�tcrco��ect 5.742,840 rinl. and tablc tookup. while also sustaiuing the tfE�ugbput of fte memory as machi�e instn�crions ��af ransrer trrer� oryrandr,. from a singlc widc multi-por�d rcgister filc. These tions ca� be further supplemen�ed with loadstore tions that hansfer register data to and from a data cac*�e static randomiccess memory ('sRAI�,r") �eld. i� the execution and transmission of multiple m�dia data d�t� sEe�ms. The system i�cludcs in one �pcct a general Pu�posc-. programmable media processor. and in another hicrarchy. fc receivlag' processi'og and The above multi-precision arithmetic, pcrmutat�on �it-ry,il]$,| ltthotl stre�I�' The general purpose' pr� be *grt;t� 5 9:T*9.:*l^dt" swi�ch. a�d cxtendcd math operations can then suppo�ts additiotrat arithmctic opfatio�s such as �aite full instrucinstrucbuf'/ *u ry.ii '. memory dynamic random access memories ('DRAMs"). and with bran� i�structions thal c�ntol the flow of instructions executed �om a� instruction bufier/cadrc SR-AM' execution tbreads gateways..so that video. radio. encr�ption. networking. etc. can efrcicntlV securely sharc .rnemory,and logic resources of a unifr-:d rnachine operating near the optim�m spoed-area point of th: targct scmiconduc�or process, The data puh fcr *"1-1 rrnified nedia proc�ssor c�� i��rface to a high input/outryut ("�O") subsys�m that movcs media sEe�ms acrrossdtra-hiBh bandwidth in�rfaces to extemal storage # ,"":�ffiltilii��#"i'�T: frtgltili�riil'":",*i iot"rf".". aa� ;;;t *- cau Ue-crnptoi"� in �paraUel multi-processor "S�f$t#f',ffiffi�i:T;*frf"r�i"ffr: cxccr�tio� of rn'lt�l� m��a �t tn, t�;"�i" ""�li"� �t�Uoo ;;;�it# ;"ld-o;;;i;Jri-"u" unit includcs a da� path. and a unir co'pled to the data path and iiffi".�#,iimultiplc ff ��"'**ilff �#'i"��i,'ifu *';$*#�"1g'f"-"J;'#;:n*mn of fs audi_o. patb 6at specd lh;a�t p;,h *d;rrg.rm;bb to implemetrt additional ��;��"p";-mJsubsantially'pcataaattuou�r- is ,ilt�"d.r; switch coupled ro rhe d�ta �1! ;;ffi"bi"�" -;;lf"f"ti d�ra received f��m the data ;;�ilf,[f,jd" �i� rr."*, to the data path. Atr e*tended * t '^d;;[� J..."t �i also providcd. wirich is coupled to �;;;�;fte-is ,"rU"-.f"��"*'"��r*ntoftheexecutionuo'latleast coupr"a to the dara path. sud, device would incorporarc ar-or the Foccs:Tc capabilities of the specialized mdti-ASIC combination a slnglc. unified processing device. Tfp uni�ed Pr�es_sg wor�ld be agile and capable of reprogramming tbroryh.$e � ;t,"",'W of oi.�i. data strea��s. The media processor �J"�"� frigf, t*�*��O �too.f i�hrface opcrable to Fansmission of new programs over the communication.";;;;;;u�-�iWof�"t"of varioussizesfrominexternal " ��edi"- This prog�amm�ble. general pmpose device is $ut '" ,�;;�;;;"nicate the received d�ta over the data parb less costly tha� the. spc�atized Frlcessol combination. .i ,"t" tlJ ,r�ru,�"s substantially peak operation of tbe casier 3o operuc and reProgram and can be tlstalled.m ;.d�"pr.*;etleastoneregisterfue�s included.which " applicd in many diEering devices and situations. Tbe may also be scalable to_communicatioos applications sufrport vast numbers of us�s thmugh rnassively para�el i�to " "#;'f-ffif'-ffi#:.i;"$i.�t;�t proc�ss a ilL-rt r'.fi�" ** -� " �*� path to d�itally device i, -*"�n*.Uf"io receive and d�t� from the data partr '�ore *ul ". ;ta ?�ro,ooJ"ut" rbe stored dara ro rhe dara pafh. A " unit is coupled to the data path distrib�tcd computing, and ij aynamically con�gurablc to partition data received _- ^rlU+*Ar�o� "x.cution It is therefore aa objcct of �'is invcntion to process med'ia f�om the-data path to accpunt for the elemental symbol size d�ta streams by exeorting op�ations at very high rates. bandwidth of the plurality of media It is also an object of this invention to unify the video. radio. graPhia. encrypioa. authentication. and wcrking protocols into a singlc i:rs�uction an object of this invention bandwidh rates i� a uni�ed processor ttrat is easy to 4j rrogam and more flex�ble than a hetrogeneous co�nbinaion of �ecial Frpos processors. It �s afurther object oftheinvention to supporthigh level mathematical processing i� a unified m�dia processor. It is also audio. netstrea�n to achieve h�gh stre�ms. and is progarnrnablc to opfrate on the d�ta to geoerate a uni�ed symbol ou�ut to t�e 40 data path. According to the peferred embodiment of the media processq. me�us are included registers and memory by performing load and store oprations. and for coordimting the sharing of data among a plurality of t�sks by pcrforming sync'l[onization operations based upon instructions and d�ta received by the fa moving data between exccution unit. Mea�s are also Pr,ovide{ for.seorrcly controllingtheseque�ceofexecutionbyperformingbrandtand -�ni�e �eld. �airc ting and table so gatew�y operations based upon instructions and dat� including �nit� g�oup. rcccived by thc exec�don unit A memory matragerDent unit look-up �peratiooi. aU at higb bandwidttr ra�s. ft is yet a fr�rtber object of tle invention to povide a opcrable to retrive data atrd �nstn�ctions for timely and instruction unifled media pocessor-that can be replica�ed i�to a multi- s�cr�re com�nunication over thc dat� Path atrq path respectively is also preferably itrclud�d in the media vast ar�y of users. Ixocessor system to support a includes a It is yct another object of f�is invention to allow for 55 pr�es!�. The preferred embodi�ncnt alsody�unically comallo bined ins�uction cache and bt�er that is parallel systems withi� tbc switctring fabric to rnassively ca�ed betweeu cadre space and b9fi1 space to ensure ,upport'i"ty large numbers of subscribers and sen�ices. of mul�le media instn�cion sEesms' It is also an object of rhe invention to povidc a general real-time ".Y*F�o cache and bufier that is dynamically and a co�nbi�ed data ; U" processor that prupose programrgble "*i� "t"ploy"d oo alloc�ted between cac�e space and buffer spacc to en�iure all points in a netwcrk real-time.response for mult�le �ncdia data stfe�ms' rt is a fu�ther objet of this invention to sustain very high of the i�vcntion' a high bandwidth ban��,idthratestoarbiEa�q�v�uv�sv�ryandi�puy"u�iprocessor futrface for receiving and tansmi�ing a mcdia i In another asPect u rilytargememor systems. srrMMARy oFrHE rNvENrroN ., Inviewoftheabove.thereisprovidedasystcmfumedia processing that maintains sutota�tiaUy peak-data nri�:ffi: ;ff*ffiff� fffiil#ffi.tr"#: of plurality width memmy proccssor in�rface includes a conEolle�s coupled in scries to commr�Eicate stued media throug�put 5.742,840 Other aspects of tl�c i�vcntio� indudc a mcthod for information to and ftom the data path, and a plurality of achieving high bandwidth communic�tions betweer � ge�memory elements coupled in prallel to each of thc plurality eral purpose med�a processor and cxtern�l devices by p'ro' of memory controllers for stcning and retrieving the �nedia vidi�g � high bandwidth interface d�sposed between the i�for�nation, In the preferred cmbodimcnt of the high bandplurality of memcry control' 5 medi� proccisor and tfte external dcvices. in which ne$S! widfh proccssor i��rface. the - bandwidrtr interface cornprises at least ooc uni-directional lers e�ch cornprise a paired link disposed between each channel pair having an input port and a� ou�ut port A menrory cotrholler. wherc ttre paired linls each rans�nil atrd plurality of media data s�ea�ns. comprisiag compotre[t re�eive plural bie of data and have differential data inputs heHs of va�ious sizes. arc transmi��d and rcceivcd between and outputs and a diffaential clock signal. processor a�d t�c exter�al dcvices at a rato that -^ Yet another aspect of the i�vention includes a system for tu the media sustains t,itatantiaUy peak data tluoughput al the media uni�cd media processing having a pluratity of general processcr, A me�od foi grocessing streams of E�di� 'l'tt is tr�rpose rnedia processcs. whe�t each media processor is also inch�ded that provides e bi-directional communications peak data rates and has a dylamtoperable at substantially fatric for trausmining ald receiving at least one s�esm of c:lly part�tiooed execution unit and a high bandwidth intermedi� data. whcre the at least one stream of media daa face for communicating to memory and input/output ele15 coryrises prescntation. trans�nissio� and storage informameDts to supply data to the media processor at substatrtially tio�. At least one gogramrnablc mcdia proccssor is povided peak ratcs. A bi-directional co�n�nunication fabrric ls withi� thc communications uetwork for rccciving. processsre provided. to which the plurality of media Processors ing and Eansmi�ing the at least one sEes�n of udfied media coupled. to transmit and reccive at least one media sEe�n �latr sver the bi-directional communications fabric. cornprising presentation. trans�nission. and storage medi;a communic�tion fabric pref- 29 The general pLEpGe. progranmable rnedia proocssor of i�forrnation. The bi-directional the i�vention combincs in a singlc devicc all ofthe necessary crably compises a f.ber optic netwGk and a subsel of the ha�dwarc i�cluded in the qpecializcd lrocessor combinapluratity of media processors corryrise oetwork servss. tions to �rocess and communicate digital media dala strcams Accordiag to yet snother aspect of the invention. a in real-time. The general pu�pose. prograrnmable media proccsstr system is povided having a parallel multi-media flexible than thc daa pa�r and a high bardwidth external interface coupled to 2s processor is therefore dreaper and mqe prrior approach to media processing' The gcneral purPose. fhe fat" path an�operable to receive a plurality of data of �rogranmaUte mcdia pocessa is thus mqe susceptible to various sizes from an external source and communicate the incorporation within a rnassively parallel proccssing netdata at a rste that maintains substantially Pe�k reccived work of general purPos media proctsscs thrt etrh�trce the operation of 6e paraltel multi-processor syst�n A plurality of regis�r files. each having at least one register cotpled to 30 ab�lity to provide re�l-time m��lti-media communications to thc m�sses. rhe data path and qerable to store dala are also included. At These fe�tures are accomplished by deploying server le�st one multi-precision execr�tion unit is coupled to the media processas and client media processors througholt the data path and is dyuunically configurable to partition data network Suc� a network provides � se�mless. global media recefved �om thc data parh to �ccount for the element�l symbol s�e of the plurality of media stre��s. and is 35 sufrer-comFtter which allows prograrnmers and netwcrk owners to virh�atize resourcos. Rarher than restrict�vely �rogf�rn�n�ble �o operate iu parallel on data sto�ed in thc access�ng only the memory space and processing time of a plnality ofregis�r fles to generate a unified symbol ou�ut local �esource. the system allonrs access to resources for cach register �le. throughow the network In small access points such as �ccordi.trg to the method of thc i�vent�on. unified s�eams of media data are processed by receiv�ng a stearn of uni�ed 46 wireless devices. where very litle rncruory and poctssing logic is avail�ble due to limited battry l�fe. the system is media d�ta including pesentation. tansmiesion atrd stfage able to d�aw upon the �esol�rces of a homogeneous multithe unificd stream of media d�t� is dynamii�formatio�. computer systert. cally partitionsd into component frelds of �t least onc bit The i�ventio�i also allows network owncrs the facility to based on the elemental symbol size of data received. the u�med shca�n of media .l�tr is then poccsscd at subst�n' 4i tack standards and to deploy uew services by broadcasting software across the netwo�krathcr �an by instituting costly t�a]ly peak oper;ation. hardware upgradcs across the u�hole network Broadcasting In one aspect of the i�vention. lhe unified strean of rnedia softwue across t�e network can be performcd at thc cnd of data is processed by storing ttte steam of unifed media data an advertisement or other program that is broadcas�d in a general regis�r file. Multi-precision arithmetic operations c�n thcn be perfcrmcd on thc stored stream of u�i�cd � nationally, Thus. scn�ices can be advcrtised a.od thcn transmi�ed to new subscribers at the end of the advstisemenl media data based on pogrunrued i�sEuctions. where tbe These and other featu�es and advantages of ttc invcotion multi-precision a�ithmefic operations includc Boolcan. intew�l be apparent upon considcrat�on of the following ger and floating point mathematical oPrations. The comdetailed descr�tion of the pesently preferred cmbodiments ponent fields of unified media dat� ca.o then be rnanipla�d based on prograrnrned iDstructioDs that implement copyirg. ss of the i�vention. t�ke� in conjunction with the appcnded drawings. shifting and ��-sizing operations. Mulfi-precision mathcrnat�cal qlerations can also be performed on the stored BRIEF' DESCRIPTION OF THE DRAWINGS stre�m of uni�ed media d�ta b�sed on programmcd FIG. I is a block diagram of a broad band media comPuter i�stn�ctions. wherc the mathematic�l operations including the general purPose. programrnable media pro' flni� group. �nite �eld. ��ite ring and table look-up opera- 5p employing cessq of�te �nvention; included to �ll tions. InsEuction and data Pre-fetch��g are FIG, 2 is a block diagram of a global media �xoccssor instruction and data pipelines. and memory �n��n�gcD�ent employing multipte general purpose media processors operations can be perfo�ned to re�rieve instructions and data according to the invention; �om external memory, The instructions and data are preferably stored in insruction and data cache/buffers. in which 6s FIG. 3 is an illustration of the digital bandwi�h spectrum for telecommunications. medi� and computing bufier storage i� the instruaion and data cache/bufers is dynarnicatly allocated to etrsure �e�l-ti�ne execr�tion. communications; t 6 5.742.840 FIG. 4 is the d�ital bandwidth spec�lun shown itr FIG- 3 taking into ucco,ro-t the bandwidth �verhead associ��d with com�resse� video 7 E puter 10 consi$s es�ntially of a gen�al -purposc media processor fl, As will be described in morc detail below. the generalpurpose mediaprocessor flreceives.processes and rti�gram of he current spccialized t�a�smitsmediadatastrea�ns�lsbidirectionalm�nnerfrom FIG. S is a block 5 upstfeam netwofk compooents to do\rtnstrcam devices' I� troaessor solution for mass medi� com�nunication. totut" rec�ived from upstrcam netFIG. 6 is a block diagram of two preseotly peferred g:^Yt:*-*: work corn�nne[ts c�D comfise any combination of audio' techniques; general purpose mcdia stn�ctufe of a general Pr�rpose. prograrnroable media pro- ro ap�reci-ate. howerrer. thc general purposc mcdia processor cessor according to the invention; t�is in uo way lim��ed �o recciving. pocessing aod transFIG. I is a drawing consisting of visual illustrations of the mining only 6ese t�rpes of media information. The general various group operations provided oo the media processor. pu�posc medi� Foc�ssor 12 of the i�vention is cryable of whe�e FIG. E{�) illustates the group expand operation FIG' processilg any form of digital media infcrmation withot�t 8(�) illustrates the group corryrEss or extract opration. FIG. rs dcparting from the spirit ald esscntial scope of the inven' E(c) illustr�es the group dcal and shufle o�rerations. FIG' �on. 8(d) illustrates the group swizzle operation and FIG. 8(e) Systcm Configuration In the preferred embodi�ncnt of the i�vention shown i� iliuin"�r the va�ious gr�up permu� o'per�ions; FIG. 1. rnedia d�r� sEc�Ds a�c corn�n��nicafed to thc mcdia 9 shorys thc preferred instruct�otr a�rd d�t� sizes for FIG. unified media the gen�alp,rrpose.�rogrammableraediapocessor.where 20 �trocessq 1�� &om several sot�rces. Ideally. FrG. z is a brock diasam or trre presently prererred Processors. ;'h:�S"ffiH#:"trffi."''if�"trii #i"tT; gClX"l�'aniUustratioiof thcvariousinitructiooform�ts 'dut" availab�� on the general purpose. prograrnrnable mcdia pu4rcsemcdiapr.ooessorl2overafiberopticcablenetwork p"o"".ror. FtG. 9�) iUustatei me iarious floating-point t�. es wiU bc describcd in rnore detail below. althoug� data s�eams are received �nd ta.nsmitt�d by the gcncral a presently existing available oo the gcneral purpose media processor. �ber optic cable nctwqk is prcfcrred. the illustrates t�" vrtious it:�-p"iot data sizes zs corn�uunic�tions netw�k in the Unitcd States consists of a and FIG. 9(c) combination of �ber optic cable.-coaxial cable and other available on ihe general purpose media the general pupose of a presently peferred memory trnr�-smission media Consequently. FIC. l0 is an illustrat�on 12 can also reccive and tratrsmit media data m�tragement unit itrcluded i! the general purpor" strea�ns over- coaxial cable 14 and traditional twisted pair shown in FIG. 7. where FIG. 10(�) is a translation l�' The speci�c cornmunications pfotocol a�d FIG. l0(�) illus�atcs tie functional blocks-;i � wire connections diagram employed over �re twistedpair l�' whetherPOTS' ISDN or the lookaside FIc. 1r is an ilustration or a sup�-string p�pethe .ir", processor; pr*..i� block Ti:-f��t� bufier; tech- nique; fih#."#i#;#,ti.*ffii�i;iffi;lili; further are senef,ally known t� those skilled in the art and no FIG. li� is a� illustration of tbe p'resently preferred super- rs discission is thereforc needed or provided herein. spring pipeline tedtnique; A�cther form of ups'cam network co��munic�tion is through a satellite link lE. Thc satcllite link lt is typlc�lly FIG. 13 is a btockdiagramof a single memory charnel for comrnunication to the general purpose media processor connecfedtoasatellitcreceivee20.Thesatcllitcreceiver20 comprises an ��16��. usrr�lly in the for�n of a s�elli� dish. shown i� FIG. 7; FIG. 14 is an illusEation of the presently peferred co��- 40 and-ampli�cation circuiry. The details of_ sudr satellitc -�e prefened com�nunlcations are also generally koj*1 q the art. �nd �ection of stand��d memory deiic"s � ' fu�ther detail is therefore not provided or i�cluded hcrein. memory interfacc; purpos media processor As described above. the general FIG. l5 is a block diagam of the inpulouput contfollf 12 communicates i� a bi-dircdional ma�ne� to receive' for usc with the memory channel shown in FIG, 13; FIc. 16 is a brock diagram shoc,ine murt�.re memory takcs placc in at re�st two rorns' First' a " :lff#*]�:ffiffH�Hi":*ffT"I�i: mrem"e*trt}n �'fFIG. rqalirui,G�,:ffii"�t'�ffii �:",:"i::j-"T.Y " n a ior a�d f;"i"r �:*( twave*�alli irnp frfltff� tffi�:� #r:�ffi�#::itJ�i.r: "'"#i",�",,.ates i� FIG. 13; thcpresentry pererredpacker conrmunications protocol for use over the memory chan�el shown �H t t"'"�:fi:?"J-i�ffLi'ffitoffi1ff�'i �v� EIG. lE shows a multi-processa confgrnation employing the general purpose nedia processc shown in FIG.7. where nC. f�e ioc� �enpork 22 is preferably enployed to eom��urnic�te betwecn thc uni�ed lrrocessor 12. a�d audiofuisual fnocessor 6onfig�ration. devices 24 or othe� digital deviccs 2�. Presently prderred l8(�) shows a processor ring configuration. and FIG. examples of audiolvisual deviccs 24 include digifal gbte shows a twodimensional �xoccssor configuration; �levi�ion. video-on-de�n�nd dcviccs. elec�onic yellow lg shows a presently prefened multi-chip FIG. p��porc. prograrnrnabie media 6o pages services. integrated mess�8e systems. vidco mentat�on of ne geieral ielephones. video garnes and gleclronig prograrn guides. As pr*"r.o, of the fg(a)strows a lines rr FlG. 18(c) and irnple- capable of comm�catiog 22 and tbe partiq�l� t'?c tation specifrc. any of thes� local networks of �twork selected is implemen- iniention. DE-'ArLED DEScRrprroN PRESENTLY PREFERRED EMBODIMENTS oFrHE n�il,Tffi,�"*'-T";11-'ilf�ii,iriif" ,.0p" of rhe invention. Presentli preferred embodiments of camera iff.i with the general Referring to the drawings. wh�e like-refercnce numcrals es otler digitat device s 26 fcr co�nmunic�tion band microcom- purpose media process� 12 include personal corDputers' refer to like elemeots thrdrghout. � broad pu,.rfOisp,rovidediDFIG:I.Thebroadbandmicrocom- ielcvision scts. work statiotrs. digital video 5.742.840 10 The general plrposc m�dia lrocessof 12 is.operable at tecrrders. atrd comPact disc reatlonly memqics. As dcvices sigaiflcantly high bandwidths itr ordcr to rceive' process upp�."L�.'�nttrer digital skilled i� the an rvill -the un�fred �nedia data sEc�Ds. Referi�g to FIG' general *l� 2� arc conternpla�d fo� comm��nication to 3. the reqpeaive-�equencies fcr vious t�pes of media data 12 without <lepart�ng ft�m tr�rpose m�dia procssor '"pi�i�"� ,""g"�f �e 5 streams ire set forth asahst a bandwidth qpcc'lrun 60' Thc those the "l* t*t-it �0 includes three component Second. th� general pqpose media �roccss� preferably bandwidth .sPectrum whidr downstreamde�iccs over�wi�eless spctrums. attiong the same range of &equencies. also com�nunicates with t.ot �fte various ftequenq' ra�s of.d�ital media comr�pt network 2g. In the p".rrouy pr.f.oJ cmbodiment of capabilities are --*i"at�oot. Cu�rent computing baudwi�h i�ventio�. wireless devices for communication over 62 shows re�note cornmuni- ro also displayed- The teleco�nmu�icafions spectrurn wireless network 28 can comprise either b*ds -used for telecoámunications cation devices 30 � remo�" ;;nputing dcvices 32. Presently the var�o�i tequency 6�amlle. t�letypc trminals and modcms pr"t�i� embodiments of �� re�otc communications transmissio�. opcrate in a range_betwcen aparoximatdy 64 bitVsecond to and personal devices 30 includc cordless telephones ISDN telecommunication protocol rnunicators. presently preferred cmbodirncnti of the remo� t'O tUoUittlsec"od- The the upper end of the devices 32 i�clude remote conbols and telecom- ls oprates at 64 .kilobitVsecood. At c.omputing invention- ttre the com- �l� anAlS.trunlsqerate municating devices. As those skilled in the art wilt �lecom�nunicationsspectrumG2.lf and 32 megabits per second' ,p,prr.lrt".-.At".iormsof remotec"ommunicationdevices3) at one megabit pcr second �cquency range ex�nds from and remote computing dcvices 32 are capable of comr�uni- resPctircly.. Tt�^SONEr pr""."- 12 without upfto*it-i.ty 128 rncgbits per second up to appoximately cation with the generat purpose�rd" in ader to cary such deprting �om thc spirit and scopc of thc invention. An agite � 3� gigabits.per seoond" Accordingly. br�d �an�conmunications. the general purpose rncd'ia A��ri�iotnotsh�wn)thatincuporatesagcn�alpurfrse of Eansffring i�for�nation at rates 12-is -�i, p-""�o. 12 rnay be used to communica� with 6cs" processcrgigabitsc4able per second range or h�be'r' into the presenred ir ;.d�"�;r" A ,p"�T "r-rr�pr."r purpose m.edia � the nidia spectrum �4 shown in FIG. 3. Voice a�d music Referring norw to FIG. 2. rhe general tra�smissions are ccntsred at frequcncies of approxinately Irocessor �;l is peferably-�irpor�-t-"gtroirt " aigi� 64 kilobits per second ��d one rnegabit pcr second. rcspecnetwcrk 3E, I� order to cnable *m.�oicorn�nunications cus- tively. �t n" ,rppe" e�d of thc mcdia spcctnrm 64' video cation anong large and ,r-U U.rrioarto. residential tranimission takes place P I fog" from l2^8-mcgabits pcr �omrs atrd mobile users. the �etwork 38 can consist of a i*�it�ii"*'., f�ry cornbination of rnaoy individurt ruUort**Cr cornprised three main forms of i�terconnectioo. The trunk and branc{res compression �chniqucs. howevcr. the video tralsmission tp.�tr- c�n-ue s�inc� dowo to between 32 kit�ia per cable 4ll as the preferred purpose s�cond to 128 r�gabits per second as a rcsult of the data optic cablc 40 is used to cotr[ect betwee� general As dlscribed betow. the prooessing required to media processors 12 disposed rs netw�k se�iers 4�-or frge a5 cornpression. achieve th" dtP compression results in a� increasc in businessi�stallations4Sthata�ecapableofcouplingdire�tiy of the network fg p."f.oUy fbcr optic md��s of ii�."�*��oo. r[uo erDptoy r�ai� of ro secondfor h�h Ocoiity �levision up to over 256 gigabits per' second for movie applicarions. when using common video to thc frbcr optic link 40. For �mmunications to busi�css and residential "ort�-ot direcily corpling to the frbcr optic cablc 40.- a gcneral ingspoc�n��n66ofFlG.3.Serialcpmmunicafionspresently to up p"rp.� -.aL fro""rr* r�c.n t" uscd as an interiace to lo take �lace in a range bctwegn two kilob�ts pcr second 'tilobits per Jecond' The FJhernet netwolt protocol stZ c�her forms of oetwork i�ftrcon�ection. pcr second' Cucent As shown iD FIG. 2, alternate forms of interconnection oprates at approximatety I rnegabits pair wiring 44. dynamic r�n��- access rncmcry a-n! othcr.digital inPut/ consist of coaxial catf" unes �i and nvisted ou�ut p��herals operate bet�r'een f] ncE3|its pcr second in place itrorgbori tt" Coaxiat c�ble Iines .r" ",l't"oUi ir rJrpi.alv eryr�i�� pr":"ld;.r�t i�*it�oo �s an�-sri m�gabits pr second- hcsently available micropro�.1 operatio-n il-�re.low g�gabits pcr ""d services to rcsidential tto �r. ecc."��g to the peferred cessors are-capable of e��bodi�nent of the invent�o�" ;;;� �,** -"di" pr* second range. Fa example. the '386 Pe1lum micropocescessors 12 ca� be installcd ar these residential A"t t"V be incapable sm�ll of bandwi�h requirements. Cunent computing ta$yi{fp a�e shown in the corppu� sor manufacuued by In�l C_orporation of Santa Clra. C�lif' oprates in thc lower lulf of that range. and thc Alpha fo"oot"rtto6.specializedprocessorapproach.thegeneral bandti�h �o 50 micf,olxocessq rn�nufac��red by Digiul Equ�ment C�poI*rposc mcdia pr�cessor "i*gh ration a�ryroaches the 16 s�cubitt per second range' co�nmunications to and f�om allow for bi-directional locations'52 these When-video com�ncssion is emptoled'.as expressed residentiallocationss2. ovcrhead reduces the effecNetwork servcrs 4� conEolled by general prrpose media above. thc �ssociatcdprocesing tive bandwi�b of �� particutar processor. As a resull io l2are also .-pf"�ltd*igft*ttt"i"�**��. IFocessors ' r or video. these lrocessors must exarnple. the network�ervers 4� cil be used to in�rfaoe 5s ordr to hardlc conpresscd ralge' The bm�vidtl between the �ber optic tretwork 40 and twisted pair wiring opcrate in -F" -,*.t.-" frcquency thc cffect of 44. T�is�d pair wiring 44 is still enployed fs sna� ri,*qf- fO ^tL9*o in FIG. 4 represents nanOting �nedia dat� sgsarDs itrcluding cornpressed video' 52 that do not or busi�esses 50 a�rd residential locat�ons �� is skes�ed dos'n to propcrly cannot c��rently subscribe to coaxial c�ble or �be.r opt�c fte co�rputing slectrum u?"*�", reiyene{s with thc tcleDetwort services. Gcneral pu�pose media processors 12'ttt oo aligl thciornputing bandwidtt �4' at rhese #"u'l;ri"g;-ttJoos 50 and co�nu�c"tions spectrum 62 andthe-media spcctrum aso aispos� Acccdin$y. curent lroccssor- tcch-�olo8�/ is Dot $�frcient non-cable reside�tial locatio�s 52, Ge�eral F�rpose media processing associ�ted with nioUil. *;"r** ll! are also installed in rvirelcss or th�oug� to""- to handtC�e ra�sm�ssioo aud data. agite complex streams of multi-media 3t �ons 54. which are ooupled to the netwqk apProa.dl to media digital rad.ios (not shoivn). � shown iD FIG. ?. i"t"-*t os T�e cur�ent specialized Proc�lsol �n Ure block diagram shown in FIG' databases or otherperipherals 56 caa also coupled to general processing is ruustrat� 5. As shown in FIG' 5' special pufpose proccssors are pr&essas 12 in the networb3t. prlfpos media s.742.840 1t t2 in paallel by tbe general gigabits per second range. In a tnjical coo�guration. an audo pocessor 76. viJeo processu Ze. graphi�s poccssor 80 and 'network processor tiZ a�e all thi bact plane ?0. s memories couplod to a back plaae 70. whictr is capable of bansmitting mcdia data _can be pocessed instn�Aions and ��ta at f�e upper kilobits to lower PlLrpose media proccssor 12. "*p�"4'to Each of ihe audio. video. gnphics aod tretworkprocessors perforrns rtatahandling operations_on single-or-unifedmedia 7�-$:l t�pically emfloy �ret own private or-dedica�d data streams tratrsmitted overthe data pa�t lffi. Exa�nples of "o"r-.ibl" p�ocessor a�d not accessible over the bach plane 70. descr�bed abovc. however. unless video dats sEe�Ds �re ro cotrstant�y beingprocessed. forexamplc.thcvideoprocesscr iO. Fgr periods of time. Tire coryrtiag power ZS wi� "[ the dedica�cd vid� proccssor 78 is thus only available handle video d�ts sEea�ns and is not ava�lible to other media data srea�ns that are direc�d to other dedica�d pocessors. This. of course. is �n inefficient use of the �o."rr- 73 partiorlarly i� view of tbe ove�all processing �pability of �ris rnulti-irocessor 108. and also an elemcnt of the execution unit 1l�. is a programmlblg switch lM. Thc programrysble switdt 104 Couptedbthemult�-precisiooAlUl02viathedatap�tl� t4. which are only to the specific As Eramples of thc prescntly preferred daA manipulatiou operations performcd by lhe general p�pose mcdia p�cssor f-i! are shown i� FIG. t. � gr:oup expand qeration is 'tnc ginerat p�rpose nedia p�cessor 12. in contrast, visually illustated fu FIG' 8(�). Accqding to the group 270 can be handlcs a d�r� strea�u of audio. video. graphics and network � expand operation. a sequential n9q_ of !itsdividcd into const�tuent su�frelds nza-nU for insertion i�formation all at thc same time with t�e sarne processor. itrto a larger �eld aray 274' The reverse of the gr. oup expand sder to handle the evcr changing combination �f d�t� 12 is dynamically operation is a-8r_orlp cornpress or extract opcration' A visual rhe general purposc med,ia-pr&cssor illus�ation of the grory compress or ex�8ct operation is parti�onablc-to �toc"t" the appopia� amornt �f Lg for each combi�atio� of ine�i� i� � unified media data zs showu in FLG. 8(b)' As showoa separatc sub-�elds stla�n A block diagram of two preferred genaal purpose n2a-tlU from a larger bit ficld 274 c,an be-cornbined to form a contiguous or seque-ntial fleld of b�ts -?70. mcd�a proccssor system con�gur�tions is s�own h-FIG. Refelring io FIGS. E(c)-8(e). group dcal. shuflc. swizzle nif.".iig to FIG. �. generaipurpore media poccssor " Uack plane 90. The presently and pennute o�rerations performed by the programmablc is coupl�d �g s high-speed preferrra uacr pan� 90is capable of operation at 30 gigoit{s �o swit� 104 a�e also illustated The operations perf�med by tbese instr�ctions are readily understood from a review of secood- As those stiUed in the artu�ill appreci�e. thc draurings. The group man�ulation o.perations illustrat'ed of operation at 400 Sigabits planes 90 that are capable in FIGS- 8(�H(c) com�rise thepreseotly contemplated data kcood or grea�r Uan-Avi�ttr are cnvisioncd wittrio t e m�niFulation operations for the general.truPos media proanA scopc �f the invention, Mult�le rnemory deviccs 9i ptane 9�. whid� �ic acccssible by es cess� �1.�s �rose skilled i� the art will appreciate. eithr also co�fled to the back q the gencral purpose media processor 12. InpuVoutput a subset of thcsc operatioDs additional data man�ulation -ttrorgtr operations can be incorporatcd in gther albrna� cmbodi�evi|"s Ca are �oupled to tti" b""k plane 90 mints of the general Pu�po�� media processor 12 witho{�t 91�. The con�guration of 6e �uuf-port� de,parting frorn tfe qpirit and_scope- of the iwention. og�ui deviccs 94 on one end of the Ouat-p*teO memory of to ha�dle t� video ��dling oprations�nclude deals. st�uf�es. shiftS. expands. corDPfesses. sw�zzlcs. pcrmutes a�d rverss. atthough othcr dat� bandling opcrations arc contemplatedThese operations can beperformedonsinglehits orbit fields consisting of two or mse bits up to the entire width of the data path lOt. Thu9. single bits or bit fields ofva�iot�s sizes can be mar�ulated throrrgh prograrnmable operation of the such data switdt lll4. systenl In fypes. proc�i6. 12 � back per spirit arc a inprV lZ �f,i*t tt" sharing of these mer�ory devi'ces gl tnrougLout lo Referring again to FIG. 7. h�her level �nathe�natical operations than ttrgsc pqforncd by the multi-pecision ALU a network lE of leneral In�rpose mcdia pocessas l� 102 arc pcrfonned in the general �xrr�nse rnedia processor Altcrnatively. ɡC, e inow, a presently prefeced inr+ qx.t�nd�d math elemcnt 106. The extendcd gratod general purpos mcdia processor �, me iotegratcd 12 thrargh 8n on-board'memory and VO S�l .Ihe math elempnt 106 is coupled to the.qaP ia� 108 and also it*"rr� ind�de� -"-ory �o-UorrO l�1. The extende d rnath -"*ory is prefcrably of sufi�icnt size to optim�zc 45 compriscs part of the exec'ution unit throughput.andcancornpriseacacheand/orbuferruemory elemetrt 106 pcrfanns the ctrnplex ari�rmetic operations � the-tii�. tbe integrated media processa 12 also connects necessary for vidco data comprcssion and simildy in�osive to extern�l memor�rEE. which i�pcfcrably larger than the mathematical qerations. One presently.prcferred cxarnplc �eld on-board m"-ory't� a�d for�ns ti" ryrto� main mr-ory. of sB extended ruath operat�on comP�iscs_ a Galois so operation, Othcr exarnples of extended rnathematical funcExecr�tion Unit One presently prcferred cmbodiment of an integrated tions �rrformed by the extended math element 1116 include generagcneral �rposc meAia po"essor 12 is strow� in FIG. 7, The CRC generation and checking. Reed-Solomon code tion and checking. and spread-spectnrm cncoding and general purpose mcdia processor 12 �re of�niin�gra�d decoding. As those skilled in the rt qpreciare. additional comprises an ciecr�tio� unit lt�o.'Thrce �D�i.n-elemerts � sub�ctions a�e included iu tfie *ecution uqit 100. A mul- 55 mathematic�t operations re possible snd co�templatd' t�le precision arithmetidlogic utrit (.AL(r) 102 pcrforrns Accordiag to the peferred embodi�ncnt of the htgrated 110 is .if r�.� aod simple a�i�rietic opoutioos on incoming general purposc mdia pocessor lil. a register �leprocess mediadatasteams.Suchoperationsconsistofcalculateand provided in addition to the execution unit 100 � �ntrol operations suc� B*1"* functions. as well as m��a �tt"' Tlre register frle 110 store,s and-Eansmits d�ta "r ad.titio�. subtraction. mult�lication and division. These eo strearistoandf�omtheexecr�tionunitlo0viathedatapattt q"r"too. are pcrformed on single o� uni�ed media data 10E. R�ther than employing a complex set of specifrc or dedica�d registers. the general purpose media pocessa 12 strearns �a�smi�cd to and from t�e multiple precision ALU lg2overadatabusord�taparhl9S.Pref;abiythedatapath preferably includes 64 general purPosc registers in the register flle 110 along with one P_rogram countcr (not l0S is l2B bits wide. alth�ugh thosc sk�ltcd in the art will in the appreciate that the data patr-loE can t�kc on any width r e s shown). The 64 general PurPose registers. contained -from ttre spirit and scope of the registerfilellOareall�vailabletotheuser�rograrnmer.and i� wfthcx�t departing general Pu{pose invention. The wider the data p�1tl 10E the more unif,ed comprisc a portion of the user st�te of the 5.742.840 above.according to medi� lrocessor 12. The general purpose regis�rs re pref- operations T" tlru: arranged and listed for eacb i6fy^c�paUfe of stcrinf any fcim'of �ati facn regist� the presently preferred opcration code number instn�ction. As many as 255 sep�ale qcrations are conregistcr filc flo is conpteo to the data pa�ll fot Oi ",nuris accessiblc to the execr�tion unit lfi) in the same templated for the pefened embodiment 9{ the gencral a i d ii ��tuuer. Thus. the usr ca� enploy a general purpose s purposemediaprocessorl2'AsshowninTablelhowever' not all of thc operation codcs are presently implemented' As register accord.ing to the qpeci�; ner�s 1f a pa*iiulr -i� pfo"�n or uniqri applicari�n. As those skillcd thc art thosc skilled i� the art will appreci�te. altcmate schemes for �iliappreciate.'tle iegis�r �1e 110 can also comprise a olganizing the operation codes. as wclt as additional operapf*�iti' of register filei UO configured in parallel in order tion codes f� the general Purpose media proccssor 12. are to possible' io support parallel multi-tlreaded processing. The i�structions provided in the, i�stuction.set fs the Instn�ction Set and Uscr hogamming oiOur. p.�tr"O by the gen�al general purpose media Processg, tl control tbe transfer. Control � rnanipulacion the p,r� -.4i" ptoLrror 12 is �chi�ved by selected instruc- pn�essing ajld^mani�ulation of data .sryms^Thc prcscntly iiori, ptogr"--"d by the user. Ihose stiite� io thc art will iegister �le 110 a�d the cxecution unit 100. -betwccn L3 74 app"Aaf� that a gfeat trumbef of pogrens are possible rs t�r�ugh vrious sftuences of instru�t�s. Particu-lar pre gr�m;c�n be devel�ped for each unique irnplerne�tation of -�.t �t.� �i.ro* �e general purpose media prooessor � e preferred width of the instruction path llll is 32-bits wide. organized as for eight-bit by�s ('lyadlgb")' Thosc skilled in the a�t will ap�reciate. howevcr. that tl�e instruction path 112 c�n ake on any wi�h without departing from �e spirit sioiof .uctrsp.rifrcprogrannsisthereforebeyondthesc�e and scqle of tbe_�nvention' Preferably. each i�stnrction 20 withi� the hstruction set is stted or orgnnizsd i� mem�y o- this dcscr�tion. f peferred instuction set fcr the general on forn-byte boundaries. Tbe presently preferred fonnat for Oo" pr"r"itty FIG. 9(a). pnpor" �rocessor 12 is included i� the Microfrcl�e instructions is shown in ^"Ci" ipiendix. the �ontents of which are hereby incorporated As shown in FIG. 9(�). e�c� of thc prescntly prderrcd instruction fqrnats-for the general prnposemedia proccssor fraein ny reference. A list of tl� presently p�eferred number operatio� codes for rhe gen�al pupose nredia ptocessor 12 zs 12 includes a �eld 280 for the major opcration code shown in Table I. Based on tlre type of operation perfcnned. below in - maj� "ip.tr Tablel. TABLE mio� o��rstln codc I feld slu sAA564-Aj SATIS648AI &9�1?f 0 ERES CSIIUFFLB GSHIJFFI,E I4MTTN I ESHTiI�TLE EMI,]I.ADDT6 GMULA�DT LU16T.AI FMT]LADD32 GMT'LADD2 LUI6BAI EADDIUO BFNEUI6 E�DDIO BFEI6 14l,fux EMDEPI Gs�fPaT8 GMDEPI FMULADD64 GMIJL/qDDI FMI]I5TTEI6 GMI,]I.ADD16 LU16LI scAs64�iI SCAS�4BAI sr��s64L,Ar GMIJI!{DD8 LUI6BI LU32I.AI LU32BAI ESETI� ESETTL BFNU�T6 4 5 EMIIX EEMIIX ECIFMT'L�4 GMI'K GSMIIX GGFMI'L6 G.TR.ANSPOSB 6 ETR�NSFOS FMTIIST]B32 GMIJI,ADD32 FMI.JI.SI'B�+ GMT'T,ADID�T LIJ32II GE�ffRACT128 LI,I32BI LT6LAI SM�SIIBAI SMTIX�{,A� SMTD(�IBA] S16LAI DFE32 BFNUE32 ESETITJL BFNUGE32 ESTNNE ESENE BFNULI6 ESETru@ BFNI]L3z sMIIX 8 SMtD( 9 ESWULE EDEPI BUDEPI cswrzlE GSWEZECOP]I GSWUZLESWAP GDEPI OT'DEPI GWTITI t0 la II IJ L4 BWTHI EUWTHI F.T6 F.Tz F.64 GUMUI.ADDT6 I�2I.AI GI'MIILADD2 LI6DAI GI'MT]LADD4 LI6�.I OIJMI.]I.ADD8 L�6BI SI6BAI Sl�� r ST68I ESTTBIO �r�0+ ESTJBIUO �TT�UE�I BST'IB�L BFNUCE64 ESI]BIGE ESI''BINE s3u.AI 532SAI 532IJ S32BI ESUB�E BFNTJL�T BIE12I GI]MI'I.ADD32 L32BAI CTUMUI-ADD�4 l37r'�. ESI'B�T�L BFNIJEI2S BFNUGE 128 t5 16 11 GTIWTIII CT'EKTR.ACT L32BI @MIJ�,ADD16 GE(TRACTI L6{I.AI @MIILA,DD�I @ilRACT32 GFMI,]I.ADDT28 ESIJBIUGE BFNIJLT2E t2t s64.Ar S�{BAI S64LI S�4BI IE t9 � 2L 22 2l zs 26 I�/EJ CIJE�ffRAqNfi L64BI GFIINJI-SI'816 OEIffRACT LI2&.,AI LI28B� GFMTJISIJB32 T64 @MT'I,SUB&I GETTR�CT LI2E[.I LI28BI GMI.ILSUB!2S J.128 B.I G.2 a'4 L8I LUSI GFMIJI.ADD32 GE�(TRAC-IIT6 L&TBAI S128I,AI SI2SBAI ST2�I,J EADDI BA}�DE ETTOR� B�NDNE EORI gUBIz, EANDI EGE/BGEZ EST'BI BE ENAI{DI ENOR� BNE EIJTJBGZ sr28BI s8I BU@/ DLE BCATEI n 3l 2E n � E.t�iNoR GF.16 @32 GrF.64 G.�28 G.T28 G.8 G.16 G.32 G.64 ECOPYI L�{INOR S.}IINOR Bt BLTNKI EI{INOR B�{INOR As shown in Table I. the rnajor operation codes are grouped according to the function perf�med by the operations. The the remaining bi� cau povide additional operands according to the type of addressing ernployed with the operation. 5,742.840 tion ttuaug� a nlemory ryryTT!,"t:!]{t}ererins to of the *r"prir"-�to-"Oa�operand(.Tmm-).orqrerandsstorcd FIG, 1(�). one prescntly prefened erobodi:nentmemcry mmoly m�nagemnt unit 122 is show�' The in any of tte general registers ('ia-' 'ttJ 'tc.; and "rd). In global virtual add.ition. minor operatioo .od", 282 c�n also be included - manag�ment ulit tn' preferably tatrslates s ad�csses $o �Vsicat ad&es�s by sofhvare Programformats. among the operands of certai� 32-bit instruction looka-abtt routi�es augmentd by a hardwre Eanslation address The presently preferred eEbodiment of the genaal frr.side buffer C.TLB') 148' A facility fs local virtual processor 12 includes a limited iolt C�o��t pose media translation 164 is also prefcrably providcd' As those skillcd similar to those secn in Reduced rnsuuction set compri� in the an will apprecia�' the memory rnanagemcnt unit l�12 ifr (,RISC') systems. The peferred instruction set foi shown io r.br""'"I gcnerar �urposc media rrocessor iucludes operations which implemlt load. sto-re. t*Tli l4S. synchronize. brandr a�d gateway fr�octions. Thcse �ve eUta.��ugu-of one�referredembodimentof beTLB groups of cpcrations can bc visually rqtrescntd �s two 14t is shown � FIG. 10(t). The TLB l�E reccives a virt�al general classes of rela�ed �erations. The branch ald g1e' address 230 as its input For cadt cntry in the TLB 148. the way operations perforrn related �ctions on medi� data 15 virt�al add�ess �O is togcatty AbfDed with a rnask 232' -thus visually repesented as block 114 in The ou�ut of each reryectivcAl'ID gatc 23{ is cornpred via str�alos and are FIG. 7. Similarly. the load. Sae and synchronize oper;ations a comp�rator 23� wit6 eadr entry in lhe TLB l�lt. If a rn�rch are grouped toiether in block 116 and perform simitar is dc�ctc�. $ ou�ut from the,comparator236 is used to 23t' Asthosc skilled i� d�ts 30 rluo�gh a �r�too', oo trt�-��u dar� streams. (Blo�ks ll4 and 116 g�tart will appreciate, Eansoeiver a match indicates the entry of the o'nly represent the above classi�c�tion of thcse operatiotrs � ihe corresponding physic�l address withi� the contents of the function in the procesiing -dd A,ii not indicatc any speci�� of med.ia d�ta'streams. TLBl4tandnoexternaln�emtryorlO�gce�isrequired. under�ing clcctronic a� do (FIG' 10(a)) is then cotrtrections.) A more �ciaile� discusiio-� of these Thc data Ao tor the daa c�che l�6 bits of the op;d";r. *� tt* functionality of the general purpose combin{--with the rernaining lowergatc 2{2. The vl�tual resultpro""rr." 12. appears in tie Micro�in" AId:D,i�x. x add�ess 230 through an exclusive{R For exarnple . the rcr��inder of the 32-bit instruction freld 15 L6 c�l n ' '�'i"ff::rff;#�1ff�i��Hl*T�t�'rii ,irdi" The forn-bytc structur of i�stuctions for the general antcornbi�ationisthcphysicaladdress244ou�utfrom�e puriposcmediigoccssor12isp,referablyindependeniofthe Tf.B l�lt. If a rnatch is not detected between thc logical -Uyt address and tho contents of the tag cachc l�t. the mcmry or�criog uird for *y �at rtru"u��s. Nevertheless. �e g;"w"y �n#uctions ar� speci�calty de�ned as- l6by� m�mgement unit 122 an extcrnal nrEmory or VO acc�s is �."t*", containing a code add.".. �sed to securety i�v�ke �o nec�ss�ry to retrieve the relevant port�on of memory and accordingly' a procedure at a hi[her privilege tevel. Ga�ways re pref- update the conte�ts of the TLB 148 -po�'ction Using generatly loown memory rna�genent techniques' infonnation rp"an.A io tr" ��tiy m"tt"C Uy thaf i�structions tanslation lookasidc buffer l lB in �re memory msDagement the menory ms[tgement unit tl! ensures memory (or utrir f�2. Gareways are thus prcferably afignea on IOU�e (anl data) are properly rctrieved from .ex�rnt rs other sources) over ao exter�al �nput/ot��ut bus 12� (see bounda�iesinthe externalme�ory.fn�d�iti�ntotheg"n�al 'countcr. privileg� level FIG. �, As describcd in more dcait betow. a hig� ban�vidth a p.,rpor" registrs and program piovided within ihe register fite 11-9 6�� le�t"ins interface 1Z is coupled to th: ex.trnll i�lfloutPut blts Ul� i"girto is n�piuif.'g" level of ne current�V exeatting instn�dion. to communicate i�rstuctions (a�d media data s�eams) to �c fie i�str�ctioo sct preferably i�cludes Ioad and store general�xrposemediapoccss�l2.lhepresentlyprcferred general pu{Posc media pro i�structions that move data tletw�n memory and the register � �hysical address widfh for the of cess� Ul is e�ht bytes (�a-bit�. I�� addition. the mem�y frte ll0. brandr itrstn�ctions to comparc ttre con�t nran�gement \\it l:�2 peferably prwides �n�tch bits (not registers and tratsfe� control. and a�itinetic operatiotrs to pst�rtuc* ncomputations on the cortetrts of regiitees. Swap shown) rh�t allow large m�mory regions to be assigncd a " oio s 'i� provide mrfti-rt�re�d and multi-pocessor syn; single TLB eotry �llov/ilg f* �ry grain rnemory m��geltra��age�n�nt Otoo�""tioo. fnese o��rations are peferably indivisiblc �nd 45 me�t of large meeory sections. The mcmory (trot showtr) such i�structions as aAA-'anO-swaf. compare-and- unit lZZ also preferably i�cludes s Pricdty bit i�clude i�stuctionsl The ni.C-poiot thatallows forprdercntial queuiug of memory reas accordswap. and mult�lex-anA-swap piority' O�r� memory �nflngeco�npare-an�-br�dr instucti�ns within the i�struction set ing to respective levels of m�nt operations generally l�own in �9 4-are also per' Tabtc I provide the necessry �ith�uetic tests f6 ,t oio io equatity and inequatity of s�ncd and unsigped fxed-point so formed by the memory nurn�gement Ttt 12: Referring again to FIC. ?. instructions received by the values. The branch througb i.t"*"y instru-aio� prov�Ls a general purpose media process�_l2 re. stord in a cpmsecure mca��s to acc�ss codJat a higtrer privileged level i� bufer/ a for�n simil�r to a high levet laiguage proceAur" catl 6ined i�itru-aion buffer/c�che 11E. The instuction c�cbe llt is dynunically suMivided to _store the largest generally known i� thc �rt, - �he general p��pose media �xocess� 12 also prcfcrably 55 sequetrc-e of instructions capable of exeortioo by �re exeo�external support� noatiig-�tut *toplo-O-Utaoch i�itructio�i. Uoi unit lfi) rvithot�t the necessity of accessing of the invention' The arit�meticoperatioor.*lietaesupportedinbardwae. memory. In a preferred embodi�nent tofie smallest a�d nost includc �oating-�iot oaait�oo. subtrai�on. mult�lication iustruction buff� spacc is alloc�tcd general purpose rredia pro- frequentty executed blocks of medi� i�structions' The division and square root. The helps rnaintain-the high ba�lvidh cessor l2prefcrably supportsot�erfloating-pointopera�ons co instruct�on bufier thus pqpos� media processor 12 by defrned b! t�c nNbf-�ib n"ating-po�Disir"d*,i ttrougb c�palrty of the general sustainiagthenumbcrofi�structionsexecutedpersccondat point vahe �n the use of softwarc libraries. n �o'ati�g -fxamples of �rc or near peak operation. That port�on of the instn�dion ppefc"ati"y b 16. 32. 64 or 128-bits wid� - r r.r r r bl i �vailable preferred floating-point data sizes arJilustatcd bufer/cachc llE not uscd as a br�er is. the�efore. g(�). 65 to be used as cache memory. The i�stn�ction buffer/cacrhe in FIG. path 112 and is preferably The general pufposc media proccssor 12 preferably sup !!a i1 goupt{ to. the lnstruction addressing and virtual rnachine o�r�a- 32 kilobytes in s�e. port, "�tuut."^o� 5.742.840 l7 A l8 data As indicated in Table I above. �re general purpose media 100 ptocessor 12 does not include delayed branch i�struaions. also and so relies upon branch � fetch prediction �cduiqucs to the keep the pipeline �rll in pograrn flows around unconditional s �nd coodit�otral kanch i�stn�ctions. Many such techniques unified are generally hown in the arl Eranples of some prescntly of preferred techniques include thc use of group compare and of sct a�d mult�lex cperations �o clim�mte unpredictable bra.nches; the use data buffer/cache lZ) is also provided to store tansmitted and received to and fro�nthe execution unit and register file 110. The d�ta bufier/c�che ti� is dynarni-catly subdivided in a manne� s�mil�f to that of The buffer portion of �re data instruction-buffer/cache bufer/cachc 120 is opimized to stac a set size of media data capable �f executio� without the necessity accessing external nremory. In a prcferred cmbodi�ncnt the i�veition. dat� br�er ipace iJattocated to the sm�llcst working sets of media data ro and most frequently ""c"sred Like the inshuction bufier. the date buffer thus �n�intains peak bandwidth of the general p��pose media procassor The data buffer/cache fZO is coupted to the data path and is preferably also 32 kilobyb� iD ffre prefer� embodi�nent of the general puryose media ts ll8 p�linc nq�traliz�tion; protected virtual Eemory spa.ce. The gateway instn�ctions. thcrefac. allow an efrcient rtreatrs to tratrsfer between various levels of privilege. pocessor 12 �ncludes apipelinedinstructionpr�fetch As dcsoibed abovc. two basic forms of media data are Uue, Although pipelined oper'ation is suppcrteO. Uro po�reosie or r pp ."t o media-pocess� i2 rlso alowJf� non-pipelined �nocessed by the general Pu�pos nedia pocessor 12. as shown in FlG. 7. Thcsc data strea�ns gencrally cornPrisc to execu� u�ithout any operational penalty. peoeeesor pipeline structure for the general Etrpose media zo Nyquist samPlcd VO fjlt. and standard memory and VO f crs O -pr 130. As shown in FIG. 7. audio 132. video 134. radio 13�. 12 cornp

Disclaimer: Justia Dockets & Filings provides public litigation records from the federal appellate and district courts. These filings and docket sheets should not be considered findings of fact or liability, nor do they necessarily reflect the view of Justia.

Why Is My Information Online?