DISPLAY TERMINAL SYSTEM FOR THE COMPUTER SCIENCE DEPARTMENT


        This ␈α∧is ␈α∧a ␈α∧plan ␈α∧to ␈α∧equip ␈α∧the ␈αβComputer ␈αβScience ␈αβDepartment ␈αβwith ␈αβa ␈αβsystem ␈αβof ␈αβkeyboard ␈αβand ␈αβdisplay
terminals ␈απfor ␈αεuse ␈αεin ␈αεteaching ␈αεand ␈αεresearch. ␈αε ␈αεThe ␈αεterminal ␈αεsystem ␈αεwill ␈αεmake ␈αεaccessible ␈αεto ␈αεdepartment
faculty ␈α↓and ␈α↓graduate ␈α↓students ␈α↓from ␈α↓their ␈α↓offices ␈α↓the IBM 360/67 in the Computation Center, the PDP-10 in
the ␈α∧Artificial ␈αβIntelligence ␈αβLaboratory, ␈αβand ␈αβthrough ␈αβthe ␈αβPDP-10 ␈αβany ␈αβcomputers ␈αβon ␈αβthe ␈αβARPA ␈αβnetwork ␈αβthat
they ␈α¬may ␈α¬have ␈α¬made ␈α¬arrangements ␈α∧to ␈α∧use. ␈α∧ ␈α∧This ␈α∧includes, ␈α∧for ␈α∧example, ␈α∧the ␈α∧ILLIAC ␈α∧4 ␈α∧at ␈α∧Ames ␈α∧Research
Center.  Connections to other campus computers can also be arranged.

        In ␈α↓our opinion, such a terminal system together with suitable hard copy output in the most important
laboratory ␈α↓and ␈α↓teaching facility that a computer science department can have.  The interests of the faculty
and ␈ααstudents ␈ααare ␈ααtoo ␈ααvaried ␈ααto ␈ααbe ␈ααsatisfied ␈ααby ␈ααany ␈ααone ␈ααcomputer ␈α↓since ␈α↓no ␈α↓one ␈α↓computer ␈α↓will ␈α↓have ␈α↓all ␈α↓the
hardware and software facilities that are of research or teaching interest.

        The ␈α↓plan ␈α↓we ␈α↓are ␈α↓now ␈α↓proposing ␈α↓is ␈α↓based ␈α↓in ␈α↓part ␈α↓on ␈α↓a ␈α↓study ␈α↓made ␈α↓by John McCarthy at M.I.T.  in fall
1972 ␈αλin ␈αλsupport ␈αλof ␈αλa ␈αλplan ␈αλto ␈αλget ␈αλa ␈αλterminal ␈απsystem ␈απfor ␈απProject ␈απMAC ␈απand ␈απthe ␈απArtificial ␈απIntelligence
Laboratory. ␈αβ ␈αβIn ␈αβthis ␈αβstudy, ␈ααrequests ␈ααfor ␈ααproposals ␈ααwere ␈ααsent ␈ααto ␈ααleading ␈ααfirms ␈ααin ␈ααthe ␈ααdisplay ␈ααfield ␈ααand ␈ααthe
replies ␈αεwere ␈αεevaluated. ␈αε ␈α¬The ␈α¬best ␈α¬plan, ␈α¬however, ␈α¬was ␈α¬found ␈α¬to ␈α¬be ␈α¬scheme ␈α¬being ␈α¬developed ␈α¬by ␈α¬Peter
Wiener ␈ααof ␈ααYale ␈ααUniversity, ␈ααand ␈ααour ␈ααplan ␈ααis ␈ααbased ␈ααon ␈ααthe ␈ααYale ␈ααscheme ␈ααwhich ␈ααis ␈ααalso ␈α↓being ␈α↓followed ␈α↓by ␈α↓the
M.I.T. ␈απ ␈απArtificial ␈απIntelligence ␈απLaboratory. ␈απ ␈απThe ␈απStanford ␈απArtificial ␈αεIntelligence ␈αεLaboratory ␈αεalso ␈αεhopes ␈αεto
upgrade its Data Disk display system to the new standard.

        The goals that we want to achieve are the following:

        1. ␈αβ ␈αβA ␈αβterminal ␈αβin ␈αβeach ␈αβoffice ␈αβused ␈αβby ␈αβfaculty ␈αβand ␈αβgraduate ␈αβstudents. ␈αβ ␈αβThis ␈αβrequires ␈αβvery ␈αβlow ␈ααcost
terminals.

        2. ␈αβ ␈αβA ␈αβlarge ␈ααterminal ␈ααfacility ␈ααfor ␈ααthe ␈ααuse ␈ααof ␈ααstudents ␈ααin ␈ααcourses. ␈αα ␈ααAbout ␈ααhalf ␈ααthe ␈ααterminals ␈ααwould ␈ααbe
available for this purpose.

        3. ␈α↓ Each terminal should be capable of displaying at least half a page of arbitrary character sets and
also ␈ααgraphics. ␈αα ␈ααThe ␈ααcharacter ␈ααsets ␈ααused ␈α↓in ␈α↓programming ␈α↓and ␈α↓in ␈α↓mathematics ␈α↓have ␈α↓proliferated ␈α↓to ␈α↓the ␈α↓point
where ␈ααallowing ␈ααeach ␈ααuser ␈ααto ␈ααhave ␈ααwhatever ␈ααcharacters ␈ααhe ␈ααwants ␈ααis ␈ααthe ␈ααonly ␈α↓solution ␈α↓compatible ␈α↓with ␈α↓the
goal of using any computers that are available to our faculty and students.

        4. ␈α∧ ␈α∧Printing ␈α∧facilities ␈α∧compatible ␈α∧with ␈α∧the ␈α∧above ␈α∧should ␈α∧also ␈αβbe ␈αβavailable ␈αβbut ␈αβnot ␈αβnecessarily ␈αβwith
each ␈α∧terminal. ␈α∧ ␈α∧In ␈α∧fact, ␈α∧there ␈α∧is ␈α∧no ␈α∧present ␈α∧way ␈α∧of ␈α∧making ␈α∧such ␈α∧facilities ␈α∧available ␈αβto ␈αβeach ␈αβterminal ␈αβbut
suitable ␈α↓public ␈α↓printers ␈α↓can ␈α↓be ␈α↓had, ␈α↓for ␈α↓example ␈α↓the Xerox XGP.  The cost of adding such a printer has not
yet ␈ααbeen ␈ααdetermined, ␈α↓but ␈α↓there ␈α↓is ␈α↓a ␈α↓strong ␈α↓demand ␈α↓for ␈α↓it, ␈α↓and ␈α↓its ␈α↓cost ␈α↓will ␈α↓be ␈α↓included ␈α↓in ␈α↓a ␈α↓future ␈α↓version
of the plan.

        The scheme proposed has the following characteristics:

        1. ␈α∧ ␈α∧The ␈α∧display ␈αβimage ␈αβfor ␈αβeach ␈αβactive ␈αβuser ␈αβis ␈αβstored ␈αβas ␈αβa ␈αβ512x512 ␈αβraster ␈αβin ␈αβan ␈αβintegrated ␈αβcircuit
read-write memory.

        2. ␈αβ ␈αβThere ␈αβare ␈αβ64 ␈αβterminals, ␈αβbut ␈αβonly ␈αβ32 ␈αβmemories ␈αβso ␈αβthat ␈ααonly ␈ααhalf ␈ααof ␈ααthe ␈ααterminals ␈ααcan ␈ααbe ␈ααactive
simultaneously.  However, a low duty cycle is to be expected if the terminals are in offices so this is ok.

        3. ␈αα ␈ααThe ␈ααcost ␈ααto ␈ααadd ␈ααa ␈ααterminal ␈ααto ␈ααthe ␈ααsystem ␈ααis ␈α↓about ␈α↓$500, ␈α↓and ␈α↓the ␈α↓cost ␈α↓to ␈α↓add ␈α↓a ␈α↓memory ␈α↓is ␈α↓about
$2200 ␈ααnow, ␈ααbut ␈ααa ␈α↓large ␈α↓reduction ␈α↓is ␈α↓expected ␈α↓in ␈α↓less ␈α↓than ␈α↓two ␈α↓years ␈α↓as ␈α↓4096 ␈α↓bit ␈α↓memory ␈α↓chips ␈α↓come ␈α↓into
production replacing the 1024 bit chips on which the present prices are based.

        4. ␈αβ ␈αβThe ␈αβsystem ␈αβis ␈αβcontrolled ␈ααby ␈ααa ␈ααmini-computer, ␈ααsay ␈ααa ␈ααPDP-11, ␈ααwhich ␈ααcommunicates ␈ααwith ␈ααthe ␈ααhost
computers ␈απ(initially ␈απthe ␈απIBM ␈αε360/67 ␈αεand ␈αεthe ␈αεPDP-10), ␈αεand ␈αεwrites ␈αεcharacters ␈αεand ␈αεpictures ␈αεby ␈αεdirectly
addressing ␈α↓the ␈α↓display ␈α↓memories. ␈α↓ ␈α↓The ␈α↓displays ␈α↓are ␈α↓maintained ␈α↓by ␈α↓a ␈α↓memory ␈α↓port ␈α↓that ␈α↓reads ␈α↓bits ␈α↓from ␈α↓the
memories ␈α¬into ␈α∧shift ␈α∧registers ␈α∧and ␈α∧transmits ␈α∧them ␈α∧together ␈α∧with ␈α∧synchronization ␈α∧signals ␈α∧through ␈α∧a ␈α∧video
switch to the displays over coaxial cable.

        5. ␈α↓ ␈α↓Each ␈α↓display ␈α↓is ␈α↓connected ␈α↓to ␈α↓the ␈α↓central ␈α↓unit ␈α↓containing ␈α↓the ␈α↓memories ␈α↓by an individual cable, and
the ␈απkeyboards ␈απare ␈απconnected ␈απto ␈απthe ␈απkeyboard ␈απmultiplexer ␈απby ␈απtwisted ␈αεpair ␈αεunless ␈αεit ␈αεturns ␈αεout ␈αεto ␈αεbe
feasible to transmit the keyboard signal back on the same coaxial cable as is used for the video.

        6. ␈αβ ␈αβThe ␈αβvideo ␈αβswitch ␈αβis ␈ααa ␈αα32x64 ␈ααelectronic ␈ααcrossbar. ␈αα ␈ααIt ␈ααwill ␈ααbe ␈ααa ␈ααcopy ␈ααof ␈ααa ␈ααunit ␈ααnow ␈ααin ␈ααuse ␈ααin ␈ααthe
Artificial Intelligence Laboratory's display system.

        7. ␈αβ ␈αβIn ␈αβthe ␈ααinitial ␈ααversion ␈ααof ␈ααthe ␈ααsystem ␈ααall ␈ααcharacters ␈ααand ␈ααvectors ␈ααare ␈ααproduced ␈ααby ␈ααsoftware ␈ααin ␈ααthe
mini-computer. ␈αβ ␈αβA ␈αβPDP-11/45 ␈αβwill ␈αβtake ␈αβ150 ␈αβmicroseconds ␈αβin ␈αβthe ␈αβworst ␈αβcase ␈αβto ␈ααwrite ␈ααa ␈ααcharacter. ␈αα ␈ααIf ␈ααthe
host ␈α¬computers ␈α¬can ␈α¬support ␈α¬a ␈α¬higher ␈α¬rate ␈α¬of ␈α¬character ␈α¬writing ␈α∧then ␈α∧a ␈α∧special ␈α∧character ␈α∧writer ␈α∧can ␈α∧be
added.

        Here are the components of the proposed system and their estimated costs:

␈α↓␈α␈␈α↓1␈αα␈αε.␈απ␈α∪Integrated␈α∪circuit␈α∪raster␈α∪memory␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α2␈αβ$␈αβ␈α↓7␈αα␈α↓3␈ααK
complete with buffer registers and power supplies,based on prices paid by Yale and quotes to M.I.T.

␈α↓␈α␈␈α↓2␈αα␈αε.␈απ␈α∪Keyboards␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α&␈αβ$␈αβ␈α↓1␈αα␈α↓4␈ααK
(70 keyboards at $200, based on quote by Microswitch)

␈α↓␈α␈␈α↓3␈αα␈αε.␈απ␈α∪Monitors␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α>␈αβ$␈αβ␈α↓1␈αα␈α↓4␈ααK
(70 monitors at $200, based on quote by Ball Bros.)

␈α↓␈α␈␈α↓4␈αα␈αε.␈απ␈α∪PDP-␈α↓1␈αα␈α↓1␈αα/␈α↓4␈αα␈α↓0␈αα␈α∪with␈α∪␈α↓1␈αα␈α↓6␈ααK␈α∪words␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α∨␈αβ$␈αβ␈α↓2␈αα␈α↓0␈ααK
(D.E.C.  published prices)

5.  Shift registers and
␈α↓␈α␈control␈α∪electronics␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α¬␈αβ$␈αβ␈α↓5␈ααK

␈α↓␈α␈␈α↓6␈αα␈αε.␈απ␈α∪Video␈α∪switch␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α⊗␈αβ$␈αβ␈α↓9␈ααK
(cost of video switch built by AI Lab)

␈α↓␈α␈␈α↓7␈αα␈αε.␈απ␈α∪Keyboard␈α∪multiplexer␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α'␈αβ$␈αβ␈α↓8␈ααK

␈α↓␈α␈␈α↓8␈αα␈αε.␈απ␈α∪Connection␈α∪to␈α∪PDP-␈α↓1␈αα␈α↓0␈αα␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α;␈αβ$␈αβ␈α↓1␈αα␈α↓5␈ααK

␈α↓␈α␈␈α↓9␈αα␈αε.␈απ␈α∪Connection␈α∪to␈α∪the␈α∪IBM␈α∪␈α↓3␈αα␈α↓6␈αα␈α↓0␈αα/␈α↓6␈αα␈α↓7␈αα␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α7␈αβ$␈αβ␈α↓1␈αα␈α↓0␈ααK
(in addition $300 per month rental for IBM hardware is required)

␈α↓␈α␈␈α↓1␈αα␈α↓0␈αα␈αε.␈απ␈α∪Contingencies␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α*␈αβ$␈αβ␈α↓1␈αα␈α↓0␈ααK

␈α↓␈α␈␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α∂TOTAL␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α?␈α1␈αβ$␈αβ␈α↓1␈αα␈α↓7␈αα␈α↓8␈ααK


        Not ␈αβincluded ␈αβin ␈αβthese ␈αβcosts ␈αβare ␈αβengineering ␈αβwhich ␈αβwe ␈αβhope ␈αβto ␈αβdo ␈αβwith ␈αβvolunteer ␈ααlabor ␈ααand ␈ααwiring
costs. ␈α ␈αThe ␈αnecessary ␈αprogramming ␈αwill ␈αalso ␈αbe ␈αdone ␈αwithin ␈αthe ␈αComputer ␈αScience ␈αDepartment.
Maintenance of the system should cost about $5000 per year counting student labor and parts costs.

        Perhaps ␈α∧$50K ␈α∧of ␈α∧the ␈α∧costs ␈α∧can ␈α∧be ␈α∧obtained ␈α∧from ␈α∧projects. ␈α∧ ␈α∧The ␈αβAI ␈αβLaboratory ␈αβshould ␈αβpay ␈αβfor ␈αβits
connection ␈αεand ␈αεthe ␈αεcost ␈αεof ␈αεthe ␈αεterminals ␈αεused ␈αεby ␈αεproject ␈αεmembers ␈αεon ␈αεcampus. ␈αε ␈αεWith ␈α¬NSF ␈α¬not ␈α¬giving
facilities ␈α¬grants ␈α¬any ␈α¬more, ␈α¬the ␈α¬outlook ␈α¬from ␈α¬that ␈α¬source ␈α¬is ␈α¬not ␈α¬bright. ␈α¬ ␈α∧We ␈α∧believe ␈α∧that ␈α∧the ␈α∧University
should pay the main cost as the major laboratory instructional and research facility of the Department.

        Regrettably, ␈ααthere ␈α↓are ␈α↓a ␈α↓few ␈α↓costs ␈α↓not ␈α↓yet ␈α↓listed ␈α↓above. ␈α↓ ␈α↓The ␈α↓major ␈α↓one ␈α↓is ␈α↓stringing ␈α↓the ␈α↓cables, ␈α↓and
this ␈α↓has ␈α↓not ␈α↓yet been explored.  The cost of the cable itself is not large - 4 cents per foot, but the cost of
stringing ␈αβit ␈αβmay ␈αβbe ␈αβsignificant. ␈αβ ␈αβIf ␈αβwe ␈αβfigure ␈αβ500 ␈αβfoot ␈ααaverage ␈ααcable ␈ααlength, ␈ααwe ␈ααget ␈αα$20 ␈ααper ␈ααterminal ␈ααfor
the cable or $1280 altogether for the bare cost of the coax.  The cost of the keyboard cable will be less.

        The ␈α∧AI ␈α∧Lab ␈α∧is ␈α∧hiring ␈α∧an ␈α∧expert ␈α∧programmer ␈α∧and ␈α∧digital ␈α∧engineer ␈α∧who ␈α∧is ␈α∧presently ␈α∧building ␈αβa ␈αβYale
type system at M.I.T.  He can help with this work, but he is not free.

        A ␈απrecent ␈απsurvey ␈αεof ␈αεoffices ␈αεin ␈αεthe ␈αεComputer ␈αεScience ␈αεDepartment ␈αεturned ␈αεup ␈αε57 ␈αεlocations ␈αεwhere
terminals ␈α↓would ␈α↓be ␈α↓wanted. ␈α↓ ␈α↓I ␈α↓am ␈α↓not ␈α↓sure ␈α↓whether ␈α↓secretaries ␈α↓were ␈α↓included as they should be according
to ␈α↓AI ␈α↓Lab ␈α↓experience. ␈α↓ ␈α↓This suggests that a somewhat larger system would be wanted if we are to meet all
CSD ␈αβdemand ␈αβand ␈αβprovide ␈ααa ␈ααsubstantial ␈ααterminal ␈ααroom ␈αα(at ␈ααleast ␈αα16 ␈ααterminals) ␈ααfor ␈ααstudents. ␈αα ␈ααTherefore, ␈ααwe
recommend $200K as the total cost of the project not counting any overheads.