THE BIRTH OF A COMPUTER CONDUCTED BY JOHN C. NASH

An interview with James H. Wilkinson on the building of a computer designed by Alan Turing

The story of the construction of the first computers is both fascinating and instructive. Under• by 'furing himself. He was a man with standing the insights and decisions of computing's innovators may explain how the technology an original and inventive mind, His evolved to its present state and may illuminate the directions it might take in the future. design had practically nothing in com• Among computing's innovators were Alan 'lUring (see page 65 for a review of a 'lUring mon with the group of computers biography) and the men he assembled to help him build a computer based on his Universal which arose out of discussions at the machine. 'lUring's team included James H. Wilkinson, a mathematician who had studied at Moore School of Electrical Engineer• Cambridge and worked for the British government as a ballistics engineer doing numerical ing at the University of Pennsylvania, analysis of explosives problems during World War II. John W. Mauchly and J. Presper Eckert This interview was conducted for BYTE by Dr. John C. Nash and took place on July 13, had already successfully completed 1984, at the Ninth Householder Gatlinburg Conference held at the University of Waterloo, the construction of the first electronic Waterloo, Ontario, Canada. computer, the ENIAC (this was not a stored-program computer), and their BYTE: Dr. Wilkinson, how did you become and my friendship with Goodwin were influence was at its peak. When I went involved with Alan 'lUring and his computer? the decisive factors. So in May '46, six to NPL in May '46, 'furing was work• JHW Shortly after the war, I dis• and a half years after I joined the ing on what he called version 5 of Ihisl covered that a Mathematics Division government service, I moved to NPL computer, though I never saw any was being set up at the National (as I thought then, temporarily) in• documents relating to versions I to Physical Laboratory (NPL). I got in stead of going back to Cambridge 4, 'furing was not a great documenter, touch with E, T. Goodwin, who had University. and no doubt the earlier versions been a colleague of mine at Cam• 'furing had worked alone on the were buried in the rubble on his desk. bridge in the Maths Lab, He was one logical design of an electroniC com• Perhaps I should attempt to give of the first to join this new division. puter. When I arrived, he had pre• some idea of the flavor of version 5, He invited me to have a chat with him sented his plans to what you might a typical 'furingesque creation. It was at NPL in Bushy Park, 1eddington, and call a "review committee" at NPL. This (continued) there I met 'furing, who I knew already consisted of a small group of Fellows by reputation as something of an ec• from the Royal Society. The commit• Dr. John C. Nash (Nash Information services, centric. 'furing and I had a long discus• tee deCided that 'furing's ideas were 1975 Bel Air Dr.. Ottawa, Ontario, K2C sion, and I was very impressed with basically sound, and they gave him a OX I, Canada) is an associate professor with him. Presumably he must have been mandate to go ahead and recruit the the Faculty of Administration at the Univer• reasonably satisfied with me since he appropriate staff. sity of Ottawa, Canada. He is the author of said if I came to NPL he would like me Up to that time everything associ• two books on scientific computing and to work with him. I think that this offer ated with the project had been done numerous journal articles.

FEBRUARY 1985 • BY TE 177 INTERVIEW

a serial machine using mercury delay store. As you can well imagine. this lines for storage, with a pulse repeti• made for what one would call "dif• tion rate of what I still call a megacy• ficult" coding. I'm not sure that "dif• cle, being rather old-fashioned in such ficult" is the right word. I would say matters. such coding was tiresome or tedious. Also it made the design of automatic BYTE: Define a megacycle. programming languages more labori• IHW: The basic pulse frequency was ous. while at the same time it made provided by a master clock which had them more desirable. However. this a I-megacycle pulse rate. It worked in feature of the machine turned out to binary. of course. That decision was be rather important; it meant we taken early on and was regarded as could do up to 16 instructions per ma• irrevocable. The word length was 32 jor cycle. i.e.. about 64 microseconds binary digits. which is rather better per instruction. than 9 decimals. This practice later became known as BYTE: They were fixed point? "optimum coding" or "latency cod• IHW: Yes. They were fixed point. but Photo I: English mathematician James H. ing:' but Thring never used that term. Wilkinson. one of the builders and program• one of the earliest things that I did (at It was characteristic of him to see his mers of the early ACE computer. Thring's request) was to program a set machine as the basic one. all the of subroutines for doing floating-point others being out of step. arithmetic. These were later to be• Turing's machine by Mathematics Division BYTE: What was the ACE's total memory? come rather important in the history head I. R. WomersleY.1 IHW: Well. Thring envisioned a of NPL. Right from the start. Thring memory of 200 long delay lines, was impressed with the importance of BYTE: "Major cycle" meaning . .. ? which would have given 6400 words. speed. It is possibly not widely known IHW: "Major cycle" meaning the time that at that time most people weren't. of circulation of the main storage BYTE: About 24 K bytes? For instance, Maurice Wilkes at Cam• units, each of which held 32 words of JHW: Yes, and although that may bridge (who quite early became one 32 binary digits and hence had a cir• sound rather small now, it was really of our principal competitors) took the culation time of 1024 microseconds, very ambitious for that time. I am sure view then that electronic computers i.e., approximately a millisecond. A Thring would never have contem• were so fast that it was much more im• conventional design would have plated or supported the bUilding of portant to get one built than to make meant that the maximum speed of a smaller machine. special efforts to increase its speed. operation was one instruction per Shortly after I joined NPL. Thring and his views were generally shared. millisecond. moved on to version 6 and then rapid• Thring took the opposite view. and ly to 7 and 8. Those were four-address most of the special features of his BYTE: Because, unlike a dynamic RAM. code machines. IEditor's note: A four• machine were designed to make it as where you can get at any cell with one or two address machine had up to four address fast as possible. There was merit in clock cycles, this had to use a thousand clock operands after an instruction. one of which both views. but it was certainly true cycles. would be to give the memory location of the that the machines we were designing jHW: The other two early machines to next instruction. I then were not nearly so fast as they work-EDSAC at Cambridge (which The earlier machine, version 5, is appeared to be. However, Thring's Wilkes built) and SEAC at the National hard to describe in these terms. But obsession with speed certainly made Bureau of Standards (which Samuel its successors performed instructions for a very untidy machine. A great Alexander built)-did. in fact. store of the type A+B to C and selected the weakness of mercury delay lines is ac• consecutive instructions in consecu• position D of the next instruction. cess time. In order to make them rea• tive positions, so that by the time one which was necessary because they sonably economic, it is necessary to instruction had been executed the were not in consecutive positions. store a number of words in each delay next one had been "missed:' and one line. Clearly, if one stores consecutive had to wait a full cycle for it to BYTE: A complete instruction would occupy instructions in consecutive positions emerge. To avoid this, Thring stored one word? in a delay line, one could perform consecutive instructions in such IHW: Yes, but it was a more powerful only one instruction per major cycle, relative positions that the next instruc• instruction than that on a conven• and indeed the early machines (other tion emerged just when the previous tional one-address code machine. An• than ACE) that were based on mer• one was completed. Since different in• other striking difference in Thring's cury delays suffered from this weak• structions took different times for design was that he had a number of ness. IEditor's note: ACE-for "automatic their execution, consecutive instruc• one-word delay lines and the arith- computing engine':-'was the name given to tions were irregularly spaced in the (continued)

178 BY TE• FEBRUARY 1985 BUY A 1200 BPS MODEM . FOR UNDER 5200! ;d~ INFOMATE INFOMATE 212PC S149 1995A S199 fEATURES: fEATURES: • 1200/300/110 bps, • 1200/300/1io bps, full· duplex full duplex • Internal card for IBM • Bell 212A and 103 PC/XT or compatibles compatible • Bell 212A and 103 compatible • Stand alone modem • CROSSTALK software compatible for use with any computer or terminal with • MODEM-MATE software included and RS-232C port • Auto-dial. Auto-answer • CROSSTALK software compatible • Tone or rotary dialing, auto-selected • Auto-dial. Auto-answer • Call progress monitoring • Auto-speed, Auto-parity selection • 8 diagnostic test modes • Tone or rotary dialing • Asynchronous data format • Asynchronous or synchronous operation • 1 year limited warranty • 1 year limited warranty

The INfOMATE 212PC comes complete The INfOMATE 1995A is an intelligent with MODEM-MATE software and will run stand alone modem, compatible with CROSS• other popular programs such as CROSSTALK TALK software, and allows you to communicate and PC TALK III. The 212PC plugs into any with all commonly used 1200 or 300 bps of the computer's full size expansion slots and modems, communicates with all commonly used 1200 or 300 bps modems. Yes, it's true! Cermetek is selling your This is a deal I can't pass up! I would choice of two 1200 bps modems for under like to order: $200, There's no catch! We want you to take _ The INFOMATE 212PC for S149.00 advantage of this spectacular deal and find _ The INFOMATE 1995A for $199.00 out what communications can do for your _ CROSSTALK software for IBM PCs for S95.oo computer. It's as simple as that, Please send the above checked items to: CROSSTALK SOFI'WARE ONLY S9S! Now you can purchase CROSSTALK software. with your Name _ INFOMATE modem. for $100 under the retail value. This version Address _ of CROSSTALK will run on IBM PC's or compatibles with 96K of City State __Zip _ memory. 1 disk drive. and PC DOS 1.1 or 2.X. You may only pur• chase CROSSTALK software with one of the INFOMATE modems Daytime phone (_) _ and this offer is limited to one software package per modem. My check is enclosed for S_ Please charge my purchase on my: ORDER TODAY! _ VISA _MASTERCARD Offer expires 3/31/85. Offer is limited to quantities on hand. Acct. No. _ Signature _ INfOMATE MODEMS Exp. date _ by Cermetek Microelectronics Calif. residents add 6'h% sales tax. All orders must add $4.00 per 1308 Borregas Avenue modem for shipping and handling. Sunnyvale. CA 94088-3565 Please allow 4 weeks for delivery. (800) 862-6271 (In California call 408/752-5000)

FEBRUARY 1985 • BY TE 179 Inquiry 222

INTERVIEW

Let us earn your trust as we have that ofothers, such as: Goodyear • General Motors • US Navy • etc. Ask for: Sue. Marianne. Kathy, Tim. Rick or Roger metic and logical operations were dis• time they were very fast indeed, and tributed among them. On a conven• this enabled me to get really exten• PRINTERS tional one-address code machine the sive working experience with floating• =r:=:~~ 10X (120 CPS) 13" 12411 549 379 use of one accumulator leads to a point computation before it was prac• b o.n. 10 I o.na 15 (160 CPS) SCALLS b RadIx 10 I Radix 15 (200 CPS) SCALLS tremendous bottleneck. One is always tical elsewhere. I am sure this is why ~ RX eo (100 CPS) 3ft 22. EPsan AX 80FIT (100 CPS) 499 289 taking one number out of the ac• floating-point error analysis first made §!Ian f)( 80 (160 CPS) 689 389 EpIan JX 80 CoIar 799 SCAlLS cumulator to put in another. By hav• headway at NPL. ~lQ 1500(200 CPS) 1495 SCAlLS Olddeta 82P (80 Col, 160 CPS) 5lHl 388 ing a number of one-word stores llIring continued with the logical 469 =:r

180 BY TE• FEBRUARY 1985 Not long ago, PC Magazine called MOBS III "The most complete and flexible data base management system available for microcomputers." That's a powerful statement. But then, a MOBS III is an amazingly powerful software package. So powerful, in fact, that it lets you build mainframe-quality application systems on your micro or mini. MOBS III is not for beginners. It's for appli• cation developers with large data bases or complex data interrelationships who want to define data base structures in the most natural way-without resorting to redundancy or artificial constructs. It's for profes• sionals who can appreciate its extensive data security and integrity features, transaction logging, ad hoc query and report writing capability and its ability to serve multiple simultaneous users. And if you want the power and the glory that only the world's most advanced data management system can provide, MOBS III is for you. For information on MOBS III and our professional consulting services, write or call Micro Data Base Systems, Inc., MOBS/Application Development Products, 85 West Algonquin Road, Suite 400, Arlington Heights, IL 60005. (800) 323-3629, or (312) 981-9200. MDBS III. ABSOLUTE POWER. Inquiry 208 WE'LLGIVE YOU THE POWER. YOU TAKE THEoGLOR~

a

MDBS III IS ~ trademark of Micro D~ta [lase Systems. Inc INTERVIEW

(Formerly Warehouse Software) Call lor programs not listed Technical & Other Information (602) 246-2222 portant developments of the century. TOLL-FREE ORDER LINE 1-(800) 421-3135 However, from our point of view FREE COMPUTER ALARM CLOCK! Thomas's preferences were unfor• In appreciation of your patronage we are giving a program diskette for the IBM PC to all tunate. But worse was to come. our customers who make a purchase of $100 or more. This program features popup menus Thomas and 'furing had absolutely and permits continuous or intermittent time display, alarm for appointments, etc. A nothing in common and were scarce• Bellsoft program with a retail value of $20. ly capable of being civil to each other. DATA BASE MANAGEMENT SYSTEMS LANGUAGES So there we have the situation where Fox and Geller Quickcode $145 Lifeboat Lattice C Compiler $295 Knowledgeman $249 Microsoft C Compiler $315 the leaders of the two groups were Condor In $295 Microsoft Pascal Comp'iler $215 completely incompatible. NWA Statpak $265 Microsoft Basic Compiler $235 Tim IV 00 00' 00 00 00. 00 00 00' 00 00 00.00 .$245 Microsoft Basic Language $225 This, naturally. made 'furing even Infostar+ $175 CP/M-86 for IBM PC. 00 00 00.00 00 00 00 00$37 more unhappy, and he began to talk FRIDAY ...... •...... $159 Concurrent CP/M 86 00 00 00 00 00 00 00 00$160 seriously of leaVing. Finally. he left in Personal Pearl. 00.00 00 00 . 00 00 00 00 00 .$215 PFS File 00 00.00 00 00.00 00 00 00.00 00.00$79 FOR PC DOS 1948 and joined the group led by Electric Desk. 00.00 00.00.00 00 00 00 00 .$195 PC Paint. 00 00.00 .• 00 00 ... 00 ...... 00 .$85 Freddy Williams and Tom Kilburn at RBase 4000 ...... •...... $265 Norton Utilities ...... •...... $48 Manchester. They were making rapid CLOUT 2 $129 Copy II PC 00 00 00 00 . 00 . 00 00 00 00$24 WORD PROCESSING Prokey V3.0 $79 strides in the construction of a com• Wordstar, Pro Pack $245 Harvard Project Manager ...... •.....$225 puter based on what became known Samna III $335 Microsoft Flight Simulator .....•...... $32 as the "Williams-Kilburn store:' 'fur• Wordstar 2000 for IBM PC $269 HARDWARE Wordstar 2000 + CAll ing's decision was. in my opinion, an ABC Printer Switch 00.00. 00.00 00$85 unfortunate one. He should have Leading Edge Word Processor/Merge .. $B5 Micro Fazer Parallel $185 Mail Merge or Correct Star Call Hayes 1200 Modem. 00.00 .. 00 00 00$450 returned to Cambridge where he still Microsoft Word With Mouse $284 Hayes 1200b Modem for IBM PC $379 held a fellowship at Kings. Word Perfect $225 64K RAM Chips (9) 150 NSEC $34 Volkswriter for the IBM PC $110 Anchor Signalman 1200 baud Modem .. $235 I was left in charge of a team which Volkswriter DeLuxe $154 consisted of six people including Random House Spell Checker $36 10 MB Int. Hard Drive for the PC $750 30 MB External Hard Drive for IBM PC .$1695 myself. We had virtually no contact PFS Write $78 10 MB External Hard Drive for IBM PC $895 Multimate $235 Princeton RGB Monitor $459 with the electronics group. and at that Peachtext 5000 00 00 00 00 00 00 00 . 00 00 00 $185 Taxan RGB Vision 42500 00 00.00 00.00 .$399 stage Goodwin, who was in charge of SPREADSHEETS Calcstar 00 00 00 00.00 00 00.00 00 00$99 COMPUTERS the Desk Computing Section. had a Supercalc II $145 Leading Edge Computer 256K Call long discLission with me. He said, Supercalc III 00 00 00 00 00 00 00 .. 00 00 00 .$160 IBM Computer Call "You know this enterprise looks now Microsoft Multiplan $119 Corona Computer Call as though it's going to founder. Before !TK Solver $259 Televideo Portable & Software $1795 Lotus 123 $294 you can be held responsible for its Symphony $415 80ARDS FOR THE IBM PC failure, would you not prefer to ACCOUNTING OR LOOKALIKES become a member of the Desk Com• ~.·""',""'irt"""/o"""I-=P-' """""'''''''''''':-:-:::'''...--.0:-:7''''''' Hercules Color Board with Par. Port .. $169 iWa . AST Six Pack 00 ... 00.00.00 00.00 $245 puting Section?" gy 384K Board with 256K $275 Well. I just couldn't accept that. By Quadcolor I ... 00 00 00 .. 00 00 00 00$185 this time I was hooked on computers. ;~~···~~i;!~ STB Graphics II Board $335 Tecmar Graphics Master $495 so I said I would sweat it out and see New Quadram Multifunction Board Call what could be done. CYMA Call PRINTERS Then a miracle occurred. Thomas Dollars & $ence 00 00 00.00 00 00 00.00 00 .$95 MBSI Accounting (Real World) $350 Dot Matrix Printers include a Free left and went into industry where he $35 Print Set Program Call had always belonged. The person TRANSFER PROGRAMS Gemini 10X $245 Hayes Smartcom $85 Gemini 15X $345 who succeeded him, F. M. Colebrook, Move-it $70 Okidata 82A, 83A, 93P Call was an old radio engineer with very Microstuff Crosstalk $95 Okidata 92P $369 little knowledge of pulse techniques· ··J·····Jill.EST· P~ICE.·,H·IJ.S. Okidata 84P $675 ········~Oftl8MJPCOR~lO~ES Juki 6100 $389 but a great fund of common sense. Juki 6300 $699 When he'd been in the post about two etion 80aJ(l -lncludes ..•. ASyllc Call on all Epson Models ,/ParalleIAdapter, .Ch,ckWith Silver-Reed Call weeks, he came over to see me and i ...back-'upandSoft\'i'are,64K Toshiba 1351 $1245 he said, "You and I appear to be 'm~,"oryexpandable to 3841<. 1 nar \'I'~rran'y Abati LQ20 $325 holding a very unhealthy babY:' He " ...... 00 ...... ,...... 1239 Diablo Call went on to invite the four senior TERMS: Prices include 3% cash discount. Add 3% for charge orders. Shipping on most items $5.00. AZ orders +6% Sales Tax. Personal check, allow ten (10) days to clear. Prices subject to change. members of our group (Alway, Davies. TOLL-FREE ORDER LINE 1-(800)-421-3135 Woodger, and myself) to join him in WAREHOUSE DATA PRODUCTS the Electronics Section on a semiper• IZI 2701 West Glendale Ave., Suite 6 • manent basis and attempt to achieve Phoenix, AZ 85021 something together. This would be

182 BY TE• FEBRUARY 1985 INTERVIEW

about Mayor June of 1948. Cole• brook was a remarkable tactician, and soon we were all working rather well together. There were one or two uneasy weeks, but soon the animosi• ty died down. E. Newman was in tech• nical charge of the electronics group; he had worked on the H2S airborne radar system during the war and already knew quite a lot about pulse techniques. He and I got on remark• ably well and that was a great help. In those days supplies 'were a prob• lem, but fortunately one member of the electronics group. W. Wilson, a giant of a man, knew everybody in the supply world and was able to solve this problem satisfactorily. After we had spent a month or two building bits and pieces and generally finding our feet. Colebrook said, "Why don't we get together now and try to build a pilot machine, the success of which will demonstrate to the authorities that we are competent and therefore ensure the continuation of the enter• prise." Then, in the light of success• we didn't hint at failure-we would go on and build the full-scale ACE. Now it so happened that we had done a little experimental work in 1947 in the the Mathematics Division About Bulls when Harry Huskey had spent a sab• batical year with us, At that time we had designed just such a miniature machine based on 11lring's version 5. & Bears& This enterprise had been stopped by Darwin when the Electronics Section was formed. Savings Bonds. 10 a large extent we resurrected this machine, incorporating, of course, a The stock market says that bulls are good and bears are bad, But if you substantial number of improvements. buy U.S. Savings Bonds through the Payroll Savings Plan, you can get It was to be called the Pilot ACE and, the most out of both markets. effectively, it would be the smallest machine based on the logic of version Rates are high during bull markets, so the variable interest rate you get 5, which would demonstrate the prac• on Bonds lets you share in those higher returns. ticality of it. But if the bear takes over and rates fall, don't panic; you're protected by a guaranteed minimum of 7.5%. ~tll s~ BYTE: How large a machine was the Pilot ~ ~~ ACE? Just hold your Bonds 5 years 1] kJ JHW: I suppose I was largely respon• or more, and you can ride a e * * sible for deciding on the size and the bull and beat the bear. 1 _r.t;. i scope of the machine, but any of the other three could by that time equal• • StOCK~GS Y,O ly well have done so, In order to have some specific objective, I decided that InAmerica. (continued)

FEBRUARY 1985 • BY TE 183 Other daisy wheel printers still make you choose.

T T interfaces for easy installation, and At 80 characters per second, the D· M 320 "& 830 " DaisyMax 830 is one of the fastest alSY ax friction, tractor and cut sheet feeders letter-quality, daisy wheel printers to handle all your office forms. All you can buy. these great features also are available And that means you no longer have in the DaisyMax 320, offering print to sacrifice image quality to increase speeds up to 48 cps. productivity! And of course both printers feature rugged reliability-a hallmark of Speed and superb quality are but Fujitsu products earned from over 30 two of a long list of benefits you get years as a technology leader and equip• with the DaisyMax 830. ment supplier to companies worldwide. Multiple users can share the Reliability backed by TRW service DaisyMax 830 since it is designed for nationwide. heavy volume word processing envi• Contact your nearest distributor for ronments. Plus, you get standard your local dealer. •• Inquiry 116 Authorized Fujitsu INTERVIEW Distributors

Algoram Computer Products (415) 969• it should be capable of solving fully through '49; I'm afraid progress was 4533, (714) 535-3630, (206) 453-1136, (916) automatically a set of 8 to 10 linear not documented. It so happened that 481-3466; Allen Edwards Associates equations by Gauss elimination. This the first chassis to arrive had been Inc. (213) 328-9770; Four Comers Tech• it would do in a matter of a second designed by Alway and myself, two of nology (602) 998-4440, (505) 345-5651; or two, very impressive for that time. the mathematicians of the team, and Gentry Associates Inc. (305) 859-7450, naturally we put them into the main (305) 791-8405, (813) 886-0720, (404) 998• BYTE: So you needed to store at least 150 frame and got them working. 2828, (504) 367-3975, (205) 534-9771, numbers and the word width was 32 bits? Then when the next chassis arrived• (919) 227-3639, (803) 772-6786, (901) 683• IHW: In fact We decided to have 10 which Alway and I had not designed• 8072, (615) 584-0281; Inland Associates, long delay lines, that is, 320 words. We Inc. (913) 764-7977, (612) 343-3123, (314) we assisted in its installation because started to design the chassis in late 391-6901; Logon Inc. (201) 646-9222, we already knew about the earlier '48, some chassis being designed by (212) 594-8202, (516) 487-4949; Lowry chassis. Thus, without any conscious the "mathematicians" and some by Computer Products, Inc. (313) 229-7200, decision being made, Alway and I the "engineers:' In the event. the (216) 398-9200, (614) 451-7494, (513) 435• became the debuggers. 7684, (616) 363-9839, (412) 922-5110, mathematicians probably designed (502) 561-5629; MESA Technology Corp. slightly more than half the chassis. I BYTE: Weren't the chassis somewhat different (215) 644-3100, (301) 948-4350, (804) 872• must emphasize that I am now talk• from each group? Or were these different 0974; NACO Electronics Corp. (315) ing about the detailed electronic components? 699-2651, (518) 899-6246, (715) 223-4490; design, not just the logical design. We IHW: Of course, the various chassis Peak Distributors, Inc. (An affiliate put our newly won knowledge of elec• had entirely different functions. Thus of Dytec/Central) (312) 394-3380, (414) tronics to immediate use. several were associated with the line 784-9686, (317) 247-1316, (319) 363-9377; counter. several with the logical con• R2 Distributing, Inc. (801) 595-0631; We started to send our blueprints to R2 Distributing of Colorado, Inc. (303) the NPL workshop towards the end of trol. and then there was one chassis 455-5360; Robeo Distributors (215) 368• that year. As each chassis arrived for each delay line. (The latter were, 9300, (216) 757-0727, (703) 471-0995; from the workshop, we put it into the of course, all identical.) S&S Electronics (617) 458-4100, (802) main frame. 658-0000, (203) 878-6800, (800) 243-2776; BYTE: The line counter is ... ? IHW: This was the section which The Computer Center (907) 456-2281, BYTE: Literally a main frame? (907) 561-2134, (907) 789-5411; USDATA counted the basic 32 pulses in a word IHW: Yes, there really was a frame. We (214) 680-9700, (512) 454-3579, (713) 681• time. 0200, (918) 622-8740. In Canada, Micos decided to use a plug-in assembly Computer Systems, Inc. (416) 624-0320, and planned to have spares of key BYTE: All this is now on one chip? (613) 230-4290, (514) 332-1930, (204) 943• chassis. IHW: Yes, of course, and much more. 3813; SGV Marketing, Inc. (416) 673• By the standards of the time it was Our units were vast by today's stan• 2323, (1-800) 387-3860 (outside Ontario); an incredibly small machine physical• dards in spite of being small by the Systerm Inc. (514) 332-5581. ly. and yet it was in many regards standards of the contemporary more powerful than either EDSAC or design. Pilot ACE was also unique Additional Ribbon Distributors SEAC. Direct comparisons are not among the early computers in being really possible. but Pilot ACE was sub• extremely mobile. The main frame AItel Data (403) 259-7814; EKM Associ• ates, Inc. (416) 497-0605; Metropolitan stantially faster on most problems, was on wheels and when the com• Ribbon & Carbon (703) 451-9072, (800) and it ({)uld solve some problems the puter was finished, we wheeled it 368-4041; The Very Last Word (415) 552• other two couldn't. back to Mathematics Division without 0900, (800) 652-1532 CA, (800) 227-3993 affecting its performance. USA. In Canada, Tri-Media, Inc. (514) BYTE: I 731-6815. And the clock cycle was still mega• BYTE: Was it power-hungry? cycle? IHW: It consumed somewhat less than IHW: Yes, still I megacycle, a slightly 10 kilowatts, which was quite low. But tough decision. Wilkes had decided we didn't have any forced cooling, on 500 kilocycles. Certainly some of and perhaps the construction was a the problems we had would have little too compact for that. When we been a lot easier at 500 kilocycles. were assembling it. we were, of course, standing in front of it all day. BYTE: It is interesting that the Apple II is a I-megacycle or I-megahertz machine, by It was like working in front of a IQ-kilo• comparison. IEditor's note: This refers to the watt fire, a rather trying experience. Fujitsu Printers instruction rather than clock rate. I BYTE: Did you, have much component Maximum Quality. Maximum Value. IHW: Yes, that's right. trouble? The completed chassis would have IHW: Not really. Our main problem FUJITSU started to arrive, I imagine, well (continued)

Inquiry 1).7 FEBRUARY 1985 • BY TE 185 Inquiry 226

INTERVIEW • 180 cps dataprocessing printing .90 cps text quality printing .30 cps letter quality printing • Dot addressable graphics • Bit image or raster graphics • 10, 12, 13.3 and 17.1 cpi allows for up to 136 char. on 8 in. line • IBM compatibility • Serial and Parallel interfaces was with germanium diodes, which • Bi-directional tractors fortunately we didn't use on the same • Multipart forms handling scale as SEAC. We used them for some gating requirements. Because 3 speedsand graphics, too. our machine was so compact and Dataproducts 8010-under$550 at MY,. didn't have forced cooling, the diodes were working at a temperature which Whether you buy, lease or rent, you'll find MTI is the one source was much higher than specified by for all the computer and data communications equipment, applica• tions expertise and service you'll ever need. At great prices. Call us. the manufacturer. BYTE: So they would fail? IHW: Yes, there were some good diodes and some bad diodes. The bad diodes would fail after perhaps a week in the computer. A good A SUBSIDIARY OF DUCOMMUN INCORPORATED diode, on the other hand, would go Computer & DataCommunications Equipment on almost indefinitely, so bad diodes Sales / Leasing / Service / Systems Integration were eventually weeded out. Digital, Inte~ Texas Instruments, Hewlett-Packard, Dataproducts, Diablo, Then came a key stage in the assem• Lear Siegler, Esprit, Link, C.ltoh, Racal-Vadic, MlCOM, Yen-Tel, Develcon, PCl, U.S. Design, Digital Engrg., MicroPro, Microsoft, Polygon & Select. bly of the computer. This was the day the first delay-line chassis was inte• New York: Outside N.Y.: New Jersey: Ohio: Kentucky: 516/621-6200 800/645-6530 201/227-5552 216/464-6688 502/426-1497 grated. This was designed by 718/767-0677 Pennsylvania: 800/521-0167 California: Newman, and, as usual. he joined 518/449-5959 412/931-9351 513/891-7050 818/883-7633 Alway and myself while it was in• stalled, but from then on he stayed with us. The three of us worked well together and debugged the whole of the rest of the machine. BYTE: When did the first program run? IHW: On May the tenth, 1950. It is in• teresting that. unlike Wilkes, who had built everything he intended to have and then made it work, we added chassis by chassis as they were com• pleted, and as soon as it was possi• ble to do something (which was as soon as we had the control unit work• ing, the adder and the subtracter, the logical operations and one long delay line). we tried it.

BYTE: How would you feed the data in? IHW: Oh, at that point we fed the in• structions in (in binary) from a set of 32 keys. When it worked on May the tenth, it could perform only the sim• plest of programs. In fact. our first pro• gram achieved the following: it took the binary number set up on the 32 keys, and every major cycle it added that number into the accumulator un• til it overflowed. Now, in addition to the 32 input keys we had a set of 32 output lights. When an overflow took For distributorship information and product details, please contact: place, the program put on the next Ampere. Inc., Asahi Bldg.. 5·20, 7-chome Nishi-Shinjuku, Shinjuku-ku, Tokyo, Japan Phone: 03-365-0825, Telerax 03-365-0999, Telex J33101 AMPERE light. So successive lights would come IP Sharp Mail Box Code: AMP (Group Code APLWS) on at a speed which was directly Inquiry 366 186 BY TE• FEBRUARY 1985 INTERVIEW

related to the size of the number on the switches. Now this program, ad• mittedly rather small, had to be fed in one instruction at a time, in binary. from the 32 keys. At the time the design of the delay lines needed im• proving; the amplifiers were some• what unstable. So we kept feeding in the program, and it kept being forgot• ten before we could complete the in• put. So [ said to Alway, "Let's try it four or five more times, and if it doesn't work, we'll call it a day and go home:' Well, we put it in about four times, and suddenly all the lights came on. This could have happened in any ease, and it didn't guarantee the pro• gram was working. However, we made the input number smaller and the lights came on more slowly.

BYTE: So the amplifiers had settled down? /HW: Yes. Then we doubled the number, and the lights came up twice as fast. We made the number three times as large and they came up three times as fast. On a binary machine that was quite convincing, so we' said, "It must be working," and went home rejoicing. That program later became rather famous on the machine. It was known affectionately as "Successive Digits" or "Suck Digs:' Sometime before this, leddy Bullard (later Sir Edward) had succeeded Dar• win, and when he visited the Elec• tronics Section (in late April 1950) he asked me how it was going. I replied that we should have something going in a week or two. Bullard was a very forthright chap, and he said with some scorn, "Come on, you can't pull the wool over my eyes. I've heard it's going very badly." (He had heard this, r-:------~------1 Heath Company, Dept. 334-262 quite justifiably, via Harry Huskey.) [ Benton Harbor, Michigan 49022 said, "You may well have heard this, and indeed it was true, but it's com• Please send me the latest Heathkit Catalog Free. ing along nicely now, and in a week Name __-::-_...,.-_-::- "--__ or two I confidently expect it to be working." Address _ Naturally. when it did work, I tried to get in touch with him as I had City State _ promised to do. I tried to phone him. Zip _ He wasn't there. Now the machine L ~ wasn't really very good at that stage A subsidlsry of Zenith Electronics Corporation CL-777D (continued)

Inquiry 134 FEBRUARY 1985 • BY TE 187 INTERVIEW

Inquiry 104 and might stop working at any time. Here are 69 reasons Save 300/0 to 430/0 The director could not be traced, and to buy at Elek-Tek, off Manufacturer Suggested Ret. prices on I was pacing up and down, saying, not to mention the "The bloody director is never here EPSON • Okidata • Gemini • NEC when you want him," when he fastest delivery stepped into the room via the win• anywhere. • lOSHIBA • COMREX • TIX dow. His opening words were, "Here's EPSON

~LJ~K.~~K 6557 N uncaln Ave. ChIcago, Il 60645 our power supplies but that of the __ ~ .I __ 4-Inc _ (312) 631 7800 [312J 677 7660 Central Electricity Generating Board. For instance, in the evening when everyone arrived home and switched on electric fires, the voltage would drop suddenly, and that gave us problems. BYTE: Historically there was a coal shortage (continued)

188 BY TE• FEBRUARY 1985 CALL TOLL FREE PRINTERS 1·800·528·1054 Anadex 9625B 51129 WPGooo. .52039 *SPECIALS OF THE MONTH* DPG500 .52259 Brother DX-15 .5369 HR-25 5649 HR-35 .. 5875 COLUMBIACOMPUTERS C-Itoh A-IQ-30. .5479 All systems include fifteen software packages F-l0 Parallel or Serial .5909 55 CPS Serial or Parallel. 51049 8510 Parallel (Prowriter) . .5315 with a $3,200 value. B510SP 5389 B510SCP .5459 8510 BPI ... 5335 MPC4210 MPC4220 Comrex CR-2E .Call CR-4 ... Call MPC4610 MPC4620 Columbia VP 420. ..Call Datasouth DS180 .51149 DS220 .51469 PRICED TOO LOW Diablo 620. ...5684 630API .51499 TO PRINT 630ECS . .51669 630 ECS/IBM .51669 Call for Prices. Series 36. .51139 BOIF 52649 P12CQI. .5529 VIDEO TERMINALS COMPUTERS MONITORS P32CQI .5759 Amdek All Monitors S32CQI. .5839 ADDS Altos All Computer Models .. Call "Call .51639 A-2Green. ,5469 Princeton Graphic P38 A-3. .. 5469 Columbia . .Call S38 .. 51719 HX-12 ..5479 Cl50 5999 Viewpoint 60 ,,5485 Corona Viewpoint 90 ,5849 PC-22 Dual Drive, ,51919 Sanyo .Call CRT-30. Epson All Printer Models. Altos PC-HD2 Hard Disk. .52999 """,,599 PPC-2 Portable/Dual Drive...... S 1&89 CRT-36 .5149 Inforunner Smart It. .5699 ,5244 PPC-HD2 Portable/Hard Disk. . .52599 CRT-50 ",,,,,, .. Call Riteman w/Tractor . Qume CRT-70 .5549 Riteman 15. .5499 Leading Edge Personal Computer. Call OVT 102 Green 5399 Rlleman Blue w/Traclor . 5299 Taxan OVT 102 Amber. ",5419 NEC 121 Green . """" ,,5125 Juki OVT 103 Green . ,5816 PC-8201 Computer. ,.5315 122 Amber ",,,.5134 .. Call OVT 103 Amber 5500 . 5850 PC-8201A-90 Battery Pack , ,,515 420 RGB 5399 .5399 PC~8206A 6100 . OVT 108 Green .5449 32K Ram . "".5215 425 AGB/Green "'" ,,5410 6300 .5699 OVT 108 Amber. .. 5519 PC-8271A-Ql AC Adapter, ,,516 Zenith NEC Televideo PC-8271A-Q2 AC Adapter . .516 ,,5639 PC·8281A Aecorder ,.589 ZVM·122 Amber .595 20'0,20'5,2030 910 .5439 ZVM-123 Green . ,,595 2050, .5669 910+ 5559 Northstar ,51215 ZVM-124 .. "",5129 3510,35'5,3530, 914 5515 All Computer Models. .. Call ZVM-133 Color/RGB. .5410 3550 . """ ,,51359 924 , 5635 ZVM-135 Color/AGB W/Audio . 77'0,7715,7730 51649 925 . .5700 Sanyo MBC-775 Portable .. Call """,5459 MBC-S50 System .Call B850 , ,51779 950, ",5905 DISK DRIVES P2, P3, ,Call 970, .5985 MBC·555 System .. Call MBC~55O-2 System Call Alpha Omega Turbo 10 ",,5739 SOOA (User Station). .5979 Okidata All Printer Models, ,Call MBC·555-2 System. .Call Personal Terminal ,5385 Iomega MBC-885 .. Call Bernoulli Box for IBM Panasonic Wyse 1091 ",5275 Televideo 10 Megabyte 51950 "", .. 5439 50 .5489 20 Megabyte. 52660 '092, 75 , ,5565 802 H .54285 1093. , """",5709 803. .51765 Rana Silver Reed Zenith Z-29 . .5599 803H. '" ",52850 Elite I .5179 EXP400 . ", ,5235 806/20. .54640 Elite II ,,5339 MODEMS ..5869 EXP500 Parallel. "",,,,,5295 TPC-l Elite III "'" "".5405 Anchor Automation TPC~2 Dual Drive. EXP500 Serial. , ,5295 ,51749 Elite 10H/Apple .. "" ,51080 Anchor Express ..Call Tpc·2 Single Drive. .51509 EXP550 Parallel. ",5399 Controller CN/Drive Only) , ..569 Mark XII .5239 EXP550 Serial .5399 1605. .51909 1000W/DOS.forAtari ",,5305 770 Parallel . "",5705 Hayes Visual Commuter. ,From 51469 TECMAR .5705 770Serial. Smartmodem 300 Baud. .5185 Zenith Graphics Master . .",5449 Smartmodem 1200 Baud Star Micronics 5445 Z-150 Single Drive. .Save 25% 126K Dynamic Memory . ,5225 Smartmodem 1200B Baud (IBM) 5379 All Printer Models. ,Call Z-150 Dual Drive , ,Save 25% 256K Dynamic Memory .5299 Micromodem liE (Apple) . 5209 Z~150W/10 Captain 128K . .5299 Tally Megabyte. "Save25% Novation Smart Cat Plus. 5315 Z·l60 Single Drive .. .. Save25% Captain 256K . ,5399 Spirit8D .. 5245 Z·160 Dual Drive. .Save 25% Toshiba Racal-VadicAIl Models. ,.Call BOARDS P134Q Parallel or Serial. ,,5709 US Robotics Password 1200 . 5319 AST Six Pack Plus. ,5285 P1351 Parallel or Serial . .51215 Paradise PLOTTERS .. 5175 Modular Graphic 06-A1 ",.,5275 Enter ".5199 Five Pak .. 5159 Sweet-P600 .5780

_GEMINI_ 2222 E. Indian School Rd. 10Xor10XPC Phoenix, Arizona 85016 with Cable & Paper Store Hours: Mon-Fri 10-5:30 Saturday 9-1 Su~rPrice Order Line Hours: Mon-Fri 8:30-5:30 Saturday 9-1 Prices reflect 3% to 5% cash discount. Product shipped in factory cartons with manufacturer's warranty, Please add 11/84 $8,00 per order for UPS shipping. Prices & availability subject to change without notice. Send cashier's check or CALL money order", all other checks will delay shipping two weeks, FEBRUARY 1985 • BY TE 189 INTERVIEW

at that time? it never once worked. You will find the prime, and if not. output a factor. JHW: Yes. Such things added to our early years abound with such bad- For the second program, they would difficulties. We knew. too. that when luck stories. ' give us any date from the year 0 up SEAC had had its first demonstra• [ must confess to having been pes• to the year 9999 and it would output tion-a little before us-it had been a simistic. We decided to have two what day of the week it was. It fiasco, even though SEAC had, in popular programs for the daily press. covered both the Julian and Gregorian general. been working reasonably For the first. they would give us a six• calendars and dealt with all leap well. During the whole of the time figure decimal number and the com• years, In all. quite an amusing little allotted to the press demonstration, puter would tell them if it were a program. Mike Woodger produced that program.

v BYTE: And where did he discover the tech• nique? IHW: He worked it out for himself. Such programs are good fun, of course, but they leave one merciless• ly exposed to the vulgar gaze. Some• v v one puts in the current date, which is Wednesday, say, and the machine' promptly says Thursday! So they're very much more dangerous. If you tell the press it's solving a partial differen• tial equation, you can swear blind it's solving a partial differential equation and they would be hard put to prove it is not. Finally, we were to have one Software Tools for Serial Devices serious program; this traced skew rays Are you developing a C or Pascal program that will through a set of lenses. v include functionality like this? Well. we decided on this last pro• gram and announced it. only to find • File transfer between computers that we couldn't get the program to • Realtime data acquisition work. 1Wo days before the press show • Printer/plotter control it had still never worked, and we didn't know whether the program had a bug Blaise Computing Inc. has a driven routines and Oexible buffer set oftools for you. ASYNCH sizes are featured. or whether it was due to computer MANAGER" is a library ofsoft• As with other packages in our malfunction. Then, just two days ware routines, including all source Programmer Productivity Series, code, which will dramatically re• before the show, Alway and [ acciden• there are no royalties for incorpo• duce the amount of time you need rating our routines into your tally found it was a minor machine to get your applications working. systems. Other high-level language fault which was not invoked at all in /' Using ASYNCH MANAGER:• interfaces are planned. Contact us V you can control any hardware de• directly to order or to obtain addi• our other programs. vice employing the RS-232 inter• tional specifications. We got all three programs working' face on an IBM PC or operationally ASYNCH MANAGER" then, just in time. The arrangement compatible micro computer. More• $175,00 includes the first language over, your PC will be able to oper• was that Bullard would entertain the interface, second language $75,00 ate at fast speeds. since interrupt- additional. popular press and [ would give the demonstrations. The whole thing was Other products to speed your devel• cation, program chaining, file and buffer v opment projects in C and Pascal include: handling from within your C or Pascal to cover three days; one day with the TOOLS" comprises a library of routines program-Sl00 popular press, one with the technical providing a$lvanced Siring handling. VIEW MANAGER'· is our screen de· forms utilities, screen handling and velopment system which dramatically press, and a third day for VIPs includ• more. Features a general BIOS gate• reduces programming required for $125 ing our competitors. Wilkes had his screen displays. Block mode data col· TOOLS 2'· is a second library of rou- leclion. verification, overlays are all machine running in Cambridge and tines focusing on DOS 2.0+ operating featured-$27S. Source Code for rou• systems capabilities. Perform memory allo- tine library-S1S0 was justifiably proud of it. Williams and Kilburn from Manchester were also coming. -BLAIS--E"'-COM-•PUrI--N-G-INC-. 2034 Blake Street Berkeley. CA 94704 BYTE: They had a machine too, didn't they? (415) 540-5....1 JHW: They had a little hookup at that v v (continued)

190 BY TE• FEBRUARY 1985 Inquiry 32 You've probably alreadyoutgrownyour personalcornputec

You'll know you've outgrown your The Personal Mini dramatically in• Your TeleVideo dealer now has the personal computer the first time you creases computing power. So it not only Personal Mini. Arrange to see it today realize it won't let people share resources runs PC software, but also hundreds by calling toll free, 800·521-4897. In or work on the same job simultaneously. ofpopular, fast minicomputer and multi• California, call 800·821·3774. Ask for The fact is, the isolated, standalone user software programs in your estab• operator 10. PC is only the beginning ofhow com• lished PC environment. This is the first PC compatible multi• puters can meet today's business needs. And, unlike less advanced networks, user, multitasking system. And the The fact is, your next step to growth the Personal Mini never sacrifices per• computer solution you'll never outgrow. is the new leleVideol!!> Personal Mini.™ formance or speed regardless ofhow Regional Sales Offices: Northwest (408) 745• Simply plug it in and grow. many workstations are on line. It also 77ffJ. Southwest (714) 476-0244. Southcentral offers multitasking. (214) 258-6776. Midwest (312) 397-5400, South• Runs PC. mini and east (404) 447-1231, Mid-Atlantic (703) 556• multiuser software. Enhance your original PC investment. 7764. Northeast (617) 890-3282, East (516) 496• With the leleVideo Personal Mini, Even system expansion costs are sub• 4777, Rocky Mountain (408) 745-77ffJ. users of IBMI!!> or leleVideo Pes, Xfs, stantially less than what you'd pay to add IBM is a registered trademark of International Business Machines. and portable computers can share data new IBM Pes. And your original invest• and expensive peripherals like printers ment in hardware, software and personal and plotters. computer education is never lost.

The TeleVideo Personal Mini. The first PC compatible multiuser system.

FEBRUARY 1985 • BY TE 191 eSANVO ~lrNldata commodore OSANVO Systems commodore All Zeniths fully software & hardware com· WC6420 Auto Modem (also .v.il.ble '0' Sanyo 550 & 555 PC's, Buill'in software patible with the PC and XT .superior Alari. & Apple .t slightly higher price). SlI, includes Easywriter, Easy Filer, Word Star, keyboard: Tech Sketch Light Pen & Micro Calcstar and more. Great Prices. Compuler Aim Drive Porls Price Illustrator 44. ZF·151·2' 128K 360 IPLl2 SER CALL MID ..,...rt 1 CALL Computer Rem Drive Price IF·'51·52 320K 720 (2 drovesllPU2 SER CALL 550·1 128K'180K(ldrNtI CAU CBC 4/12 Analog to Digital ZW·'5'·52 320K 106 MB +360 lPU2 SER CALL 4 Chanl12 8rt...... 550·2 128K' 360K(2 dnvesl CAU ZF·'61·21· '28K 360 'PU2 SER CALL ,...... c n •. 555·1 128K' 360K (2 dr....1 CAU ZF·'61·52· 320K 720 (2 droves) ,PU2 SER CALL Typing Tutor III w/l.ner Invaders. ... 35. 555·2 128K' 720K (2 drNts) CAU Panasonic KXP 1090, IBM cable. ·Port.bIe (Also fo< AiJIJI. & 18MI 'EXPANDABLE TO 256 Friction and tractor. Paper Clip Word Processor Personal Pearl database manager only Reg. list $430.•0.1210 CBM/C64...... ID, $100 with any Zenith System purchase. bO Column Display Card OTHER COMPUTER SPECIALS SOFTWARE HIT PARADE by "Batteries Included" . 14•. Math Blaster! (Davidson) ...... II. Symphony (Lotus) ... 411. Oracle (Consultant) Data Base Typing Tutor III (S&S) .. '. .. II. PFS:Report (Software Pbshg.) 7•• by "Batteries Included" ...... I'. IBM PC and PCXT in stock, •••C...... available at special prices. . CALL Alphabet ZOO (Spinnaker). . 22. PFS:Write (Software Pbshg.) ... I'. by "Batteries Included" ...... 14•. Success with Math (CBS) 1.. dBase III (Ashton-Tate). ... 410. Cable from BusCard 10 parallel Printer ... 25. ALI'A ... CPII C••,.t.,.. 64K Typing Tutor II (Microsoft). . 1.. dBase II (Ashton·Tate) 321. memory workhorse at super special All other "Batteries Included" items prices. Algebra I (peachtree) ..... n. Framework (Ashton·Tate). . 410. in stock...... CAll 1/55 .. SIGO. Story Machine (Spinnaker) .. 22. Multimate (Multimate). . 320. FOIITII'er PIlle 14 Word Attack! (Davidson) ... II. Chart (Microsoft) ...... (Full Fog. Model) by C giIe/R~ey 5D, 2/55 S7IO. 2/05 ... St••• n •. Get Organized! (Electronic Arts) 131. Multiplan (Microsoft) .... 121. DItt. Dlsil ..(e," IIIIIC " .... ee. ..-fuliline. . .. CALL I. ClIIY ,mKtltl) , 31. Spotlight (Software Arts) 12.. Wordstar (Micropro) 311. Stat for PET/CBM/C64 '5. Microsoft "Word" (w/Mouse). .. 321. J.K. Lasser's Your Income Tax Comprehensive Statistic.1 Analysis Routines DISKS & ACCESSORIES Unix Operating System for PC caLL Book and 2 diskettes • II'. CI4 '...... lit: 'ncludes BusCard Copy PC 21. (Apple & IBM only)...... 41. II. IEEE cable and MSO SlJlIOfdrive 500. Pric. per 5'4· Olslc Flight Simulator" (SUb Logic) 3.. Your New Baby. . . . . 2•• KMMM PASCAL IV.l (C64/PET) '5. NEW +4 PASCAL.. . CALL • III' AT 1-2-3 (Lotus). . .. 211. Kermits Story Maker. .23. ..,. 11111 11111 • TPI • TPI • II.' PFS:File (Software Pbshg.) .. ... ·",..••I'etrI IIIftI-Data• FW·FILf II-U••r ." ••". I.t· III 1...... ••IIIt.'. III'" lIlI... '.clll4•• r ert 10 ,1.1. 1.14 2.11 I.. CAU base filing/feporting system '" 1211. 100 I." I.. 1.11 I.. CAU .,'t.r I ••11,.11., relltl••..... 5•• 200 U7 I.. 1.11 1.7. CAU ------_.We carry lull software lines by Electronic Arts. Scholastic, Scarborough, Petscan-Connect up to 37 CBM/PET PFS. Spinnaker. Batteries Included. Others. If vou don't see it here. CALL. computers as networked cluster to shared ....". .. I' AT drives & printers. . .. Clll ..,. ..111 11111 • TPI • TPI • II.' C·Scan-Connect up to Bcomputers (C 64, 10 ,1.1. 1.11 1.11 '.n 4.17 OUTPUT DEVICES VIC 20, or similar buss) to shared drives 100 1.14 Ut I.n U' 4.• 200 Ut 1.17 1.71 •.• 4.• Prl...,. q It.'. III.... C. 1t.1I. and printers. .... 125. MONITORS ....11, ,...... c. 0111...... 111•• IIUIU II II AT Int"": . ..,. ISIII IIIH • TPI • TPI ••1•• UII-2D .lIz I_II•• U.""II. Star Micronics Gemini lOX IETC. 10 '1 .• I." 1.11 1.13 1.11 meld•.lII.11y uplI"•••eU,· 100 1.11 1.14 1.7. I.. 1.11 ICIer .1 '. - 10- c roage. FIll20 CPS .... . SHI. ...t. T.cll.I..,T•••••t. Drllll• 200 I.. 1.71 I.. 1.14 ·120D11 31·12" , 11 11 ' III' 1 1. 24'111. supe, new system. PuIS 3.3 M8 on 5% lIoppy Fo,· t. ... CAll mats to 28 FOf PC Xl ownecs, an ahernative to hard IPlC.Al.... . -...., .... 111_ D·IO-T,ICI., F". disk. 4dISk back·"" '0 megabytes CALL 30 Mec:lnlos/l 31't· disk,n,s in Am.ray - n.tTI 250. .lecH.t., Orl".. (10 MB) 7.11. dlskbsnk .. S135. ZVM-124 12" amber-22 MHz TTL ...... ,..+ . Okidata 92 . ... CALL ",••1...... 1.' &,...lIle• for IBM SI50. Amdek Printers C...... 310. PI.AII.IC 5025·25 CPS O.isywheel. 2K Buller CAll ",••1•• 5 lIull. ... 113. 12" Green 20 MHz (Sound). . .. S137. 5040·40 CPS O.isywheel. 2K Buller CALL ....,.. O•••II.....-p.,.II., port. s.,i.1 12" Amber 20 MHz (Sound) . 141. 5055·55 CPS o.isywheel. 2K 8ufter CALL port. clock/c.lenda'. IllelIlNG.1 Brother HR·3S Daisywheel· No RAM.S!". w/64K-$371. w/384K-$4I1, ECM 1226-12" Green Sli. 35 CPS Bi-Directional ...... A.OII .AI 1,.,lIle. c '.r ECM 1302-1-13" Color RGB 111. Brother HR·2S Daisywheel- II. PC (Includes Halo softwa'.I 200, ECM 1302-2-13" Color (Hi Res) 33D. 23 CPS Bi·Directional ...... 14•• Hercules graphics board 3511, U·....IIIIEmI Brother HR·1S Daisywheel· ...4I'llt ...., ret...... Icll ....11 Votrax speech synthesizers- Color 300 .. . CAll 13 CPS Bi·Directional . ... Personal Speech System 24•. 50 top quality ds/dd diskettes packaged in Star Radix 10 ... 500 caLL AST-full line of IBM cards & Boards. CAll an Amaray Media Mate 5. Only sn. 600 call Star Radix 15 . ...HI.... Keytronics 5150 keyboard. . 17•. Itoh AB carrys all major braRds...3M, Verbatim, 700...... call C. Koala Pad (w/software) ...... II. Maxell, Wabash, Sentinel, Dysan...in all 710A...... call Prowriter I (Parallel) ...... call 4164 Chip· Memory expansion for • 1. X-TRON Monitor...... 121. Starwriter (40 CPS) caLL IBM, 64K...... 3'. popular sizes and configurations. CALL for Diablo 620 (Daisywheel) call super prices. Interface Cables-6, 10, 12 ft. all IBM Parallel Printer Cable. . . 1•• popular connectors CAll POWER DEVICES (Dealer inquiries invited) IIIIITIUII Datash,eld back-up power source COMMUNICATIONS Mini Flip 'N File (50 5" _I S17.41 200 PC-200 wall. ... SU5. Mark X Auto Dial/Auto Answer Sl1 •• Rolltop 100 (100 disks. 300 XT-300 watt. . 310. 10 dividers) .. H .• Anchor Mark XII Smart Modem !til, ORDER LINE, 9 AM·6 PM EST Mini Kas-ettel1 0 (fo< 5" disks) "BITS" Power back-up-2S0W .. •• Hayes SmartModem 1200B 411. True uninterruptable tl2.21 tOI2•• II. (inctudes Smartcom II sohw...e) Brooks 6 Outler-Surge Supressor/ It.• Hayes SmartModem 1200/300.... 4". 800-&11·1111 ·A.,., ...... 1. Noise Filter ...... 14. "Crosstalk" software. .. 131. (IN PA., 215-822·7727) Innovative Concepts-fold out style for Computer Power Inc.·SOO VA 132D. • Voad keybolrd-phone IOftwar. for 5" disks: Tripp Lite 425 VA 310. IBM PC 8nd compatibles. F••tur.. CUSTOMER SERVICE Flip 'N File/2S. . tUI Flip 'N File/50...... ft.• =I,.":r~I.~i~~.':~o.t~All Smith & Bellows Wooden Storage 0N0ri0g 1oIonooIion: ()rde< by check. MasterCard or VISA r------, 115-811·7717 Boxes for 5" disks. Natural or dark finish. Personal checks ,ale 15 days to clear no .i.lL1g OIl C!!tliled FREE CATALOGI I For 50 disks ...... 1•• checlts JI """"'Y ordeededucol""'" IL-items. For a free copy call or write. -.l 252 BETHLEHEM PIKE IBM drive analyzer (Verbatim) .. ft.• rnst_ut""" ReQuesIs tor bod OIl _ 'eq""omen~ "'lied COLMAR. PA 18915

192 BY TE• FEBRUARY 1985 Inquiry 7 INTERVIEW

time, but it could scarcely be called unfortunately we knew that the The computer a computer. They hadn't built the amplifier, as it warmed up, would Mark I by that time. Their little hook• become unstable; the amplifier would up was the first anywhere ever to run then need to be retuned and in 10 factorized numbers a stored program. It worked in \947 minutes all would be fine from then and found the highest common fac• on. like a charm. tor of two numbers. This was, of So we were expecting to run into course, a great deal smaller even than trouble almost as soon as the dem• Further press representatives came in the Pilot ACE. However, it was an im• onstration started. Well, the press ar• the afternoon; still a faultless per• pressive "first" and I well remember rived. They threw numbers at us and formance. being very heartened when I saw it the computer factorized them like a The next day we had the technical working. charm. It was indefatigable! press, and it was the same story. My point then, is that Open Day was We moved on to the "dates" pro• Never before had it worked for any• doomed to be a failure. The plan for gram. It worked as it had never thing approaching this time period the first day was that Bullard was to worked before: the day of Trafalgar. without a fault. The third day the ViPs entertain the press upstairs, while Waterloo, King George V's birthday. came. Surely it would let us down downstairs we made sure the com• We moved on to the ray tracing. It now? Not a bit of it. Wilkes was there. puter was working. We were to receive traced rays like a fiend; nothing could I have always found him a very fair a signal when Bullard was almost stop it. It continued in this vein from man, but naturally he was not pre• through. We did, and immediately the 10 till I o'clock. Then the press went pared to give anything away. He didn't machine stopped working. We found away to lunch. We immediately looked get a chance; it was perfect. It had out. almost at once, that it was a at the output from the delay line, that already been decided that there chassis associated with one of the is, the shape of the pulse coming out. would be a fourth day when it would delay lines. We plugged in a spare, but It was the best output we'd ever seen! (continued)

2 YEAR WARRANTY ON CARD AND DRIVE

Yes, send me more Information on: o 10 Mbyte Disk Drive !!J!byle o 20 Mbyte Disk Drive Unde Technology, Inc. GUARANTEED 8820 S. Sepulveda Blvd., Suite 204 los Angeles. CA 90045

NAME _ SUggested rdaII price 51095 TO OIIDBI saND CHECK OR MONIIY OIIDIIR 101 ADDRESS, _ "4.~PRICE UndeT...... " Inc. ClTY STATE __ ZIP __ 8820 S. 5epu1l.teda BM:f., 100% refundable within 30 days. SuIte 204 Los Angeles, CA 90045 For IBM pc. & c:orrptibIes. 2 Years warranty on disk c:IrI\ie and OR CALI. TOLL .... COl moiler card. I f8ClOt 227-2400 ext. 974 • HaIH1l!ight drive wiIh In QIIIIbmII C4IlI COI'ttJ'oIIer card I f800J 772-2666 ext. 974 • Low error rate VIsa and MasterCartl accepted. • Lowpower QIbr1la Il!SldMls add 6~" SIIes lU. • High performance and reliabiiity ...... - • Easy installation ...... ,••,'...,.11 -..... 'IIM Is a I'l!glSlMd trDmarlc d IrolernalklnaI 20 ..,..far ·1,.... IIuisln5 MachInes Col'p.

Inquiry 182 FEBRUARY 1985 • BY TE 193 INTERVIEW

timacy gained with the computing process was fully comparable with that which one gets on a hand desk machine. where perforce you see every number. But on ACE that familiarity was gained quickly and painlessly. This experience was in• valuable. Is there any way you can get it now? Of course there is. but one needs to know what is worth having and to have the incentive to output it. BYTE: I would say my own experience is that we are transferring large-machine."faceless" programs down to the personal computers. where in fact one can go back to the ACE ideas. jHW: Yes. I agree. The potential is Photo 2: A hand-cranked desk calculator. like the one used Olj Wilkinson to perform there. and it's much greater. really. numerical analljsis during World War II. This machine was manufactured Olj Brunsviga. than it was on ACE. But in my experi• ence. many people who do comput• be put on show for the staff of NPL. realize what it is one should be do• ing are reluctant to look at numbers. This was a Saturday. The computer ing and why. For iterative methods we At Stanford the general level of our had a small fault before our audience used acceleration techniques which students has been pretty high. but I arrived. but we soon put this right. were actually under the direct control would say their main weakness is in and once again it performed flawless• of the operator. (For instance. when their inability to look at outputs and ly. The chances of such a performance we were using the power method for extract the meaningful information in must have been a million to one the determination of the dominant them. In fact. somewhat to my sur• against. eigenvector of a matrix. we could prise. they are generally less efficient On Monday we came in feeling follow the progress of the vector on at this than the assistants I used to rather jubilant. The computer was a cathode-ray tube screen. We had a have at NPL in the ACE days. in spite down. and it took us about a week to cathode-ray display which showed the of having far superior mathematical get it working again! contents of any long delay line. We qualifications. Most of those assis• would look at the screen (which tants had experience with desk com• BYTE: Today a lot of people are coming into showed 32 components of the current puters and had learned to "look at computing with no background in calculation. vector). and we would see how fast it numbers:' The Pilot ACE forced them Many of the machines they're using don't have was converging. We would put a piece to continue with this habit. the properties that ACE did. with double• of paper over it. and we could say. for I certainly do not want to suggest precision accumulation of inner products. Peo• example. "It's gaining a binary digit that the way to acquire this habit is ple have very little knowledge of this. How every three iterations. so the ratio of to serve an apprenticeship on hand can these ideas be got across to them? the dominant to the subdominant desk computers. but we have yet to jHW: It's a really difficult question. eigenvalue must be about 2 to the learn how to instill the relevant knowl• and I wouldn't claim to know the com• power Y):' We could then set up a shift edge. _ plete answer to it. Our experience of origin on the input keys that would with the Pilot ACE was really rather give much faster convergence.) This FURTHER READING special. In order to get the most out work was commonly done by assis• This interview examines lames H. Wilkin• of a machine with such a small store. tants who were in no sense qualified son's role in building the computer de• user cooperation was essential on a mathematicians. but they became signed by lUring. For additional informa• scale which in many ways is not very expert indeed. It is surprising tion on this subject. see Wilkinson's "lUr• achieved even now. This gave one an how well they understood the battery ing's Work at the National Physical labora• intimacy with the machine; we were of acceleration techniques available tory:' in A Historlj of Computing in the 'TWen• tieth Centurlj. N. Metropolis. j. Howlett. and forced to look at the numbers and and how efficiently they used them. G. C. Rota. eds. (New York: Academic thereby achieved a deep understand• When later we went over to more Press. 1980) and his articles on this topic ing of what was going on. One can. automatic techniques. they com• in The Radio and Electronic Engineer (July 1975). of course. do this with modern com• plained we were "taking the guts out plus a transcript of an oral history in puters; indeed the potential for doing of their work:' They really loved these Pioneers of Computing. C. Evans. ed. (London: it is actually greater. but one has to early programs. The familiarity and in- Science Museum. 1975).

194 BY TE• FEBRUARY 1985