Analytical Engine The JOURNAL OF THE COMPUTER HISTORY ASSOCIATION OF CALIFORNIA Volume 3.2 The 930 Comes to California February 1996 February 1996 Vohllne 3.2 Analytical Engine The JOURNAL OF THE COMPUTER HISTORY ASSOCIATION OF CALIFORNIA even paper can be saved by vigilance and care; but TIME! TIME! TIME! stories untold can die like a bolt from the blue, on Thanksgiving Day or any other. "Barney Oliver, a brilliant engineer and friend of If you read the ENGINE, please write for the the CHAC, died in Los Altos Hills on Thanks ENGINE - and remember that we want your giving Day." story most of all. Or, if you don't feel up to That's how much of this editorial had been written writing, certainly you can recommend someone when we received the news, from JAN Lee at Vir that the CHAC might interview. Do your part for ginia Tech, that Konrad Zuse had died in Hiinfeld, the history of computing and help us save a story Germany on December eighteenth. As Paul today! Ceruzzi points out, Zuse'sdeath means that the "first tier" of the great concept-makers in com TECHNICAL KNOCKOUT. ... puting - Vannevar Bush, Stibitz, Atanasoff, Turing, Aiken, Eckert, Mauchly, von Neumann, The cover of the November ENGINE proudly and Zuse himself - is no longer represented in proclaimed "Fall Double Issue." A few weeks later . mortal company. we turned around and realized that, if we included This turn of history will no doubt provoke highly all the proposed material, February's ENGINE particular feelings in each person who reads this. It would be a "Winter Double Issue" - something we makes us, personally and particularly, feel like all would have enjoyed, but a luxury that the running out to the sidewalk of El Camino and CHAC can't afford this year. yelling "They're dying! Help! Somebody do Computer history is roaring the world around. something!" Because at times it drives us to dis New museum projects abound, from California to traction - like a chant barely in earshot: will be Germany. Interest in legacy hardware and software lost .... slips away. ... has been lost. Like a drum roll has increased tremendously over the last year - at muffled but unending. least if the questions we're asked, day in and day Now more than ever, raise your voices, say with out, are any indication. And there's so much com us: We have to save the stories! puter history news that we're hard pressed to keep up with it .... which is a way of saying that we can't, With your help, with the power and will of us all, but we won't admit it. the CHAC strives to save some ofe very part of With a few more subscribers - another hundred or computing in California - computing, that land so - we could publish an ENGINE with fifty-six scape of dazzle, stair-carpet of hyperbole, heedless sower of jargon .and acronym, that cultural and or sixty pages every quarter. (Sixty-four would be economic and social jolt of ingrained flashing especially cool.) We could stop pushing material speed. It becomes us. It overtpok our land, and forward from one issue to the next. And the best other lands, reached out, and with a million sparks part is that, if more people subscribe to the ENGINE, everybody gets bigger and better issues. threaded a continent of consent - the net, the That includes you. Web - a world, where none was! in the ether. We cannot save it all, for it became us. But we must . ... doesn't it? You do subscribe, don't you? Good. save some of it all. And most of all we have to save That's what we thought. the stories. Silicon persists as long as the sand that gave it up; steel only grudgingly returns to earth; Page 2 The Analytical Engine February 1996 GET THE PICTURES! PURE GOLD. ... Beginning with the November issue, the ENGINE The golden years have begun. As you'll see in - in a way - was a casualty of its own success. Quick Takes, 1996 will comprise the fiftieth anni The documents in support of Chris Edler's fine versaries of the founding of Engineering Research HP 3000 history were the ENGINE's first internal· Associates (ERA,) the dedication of ENIAC, and illustrations. But they also meant that, for the first the establishment of the Association for Com time, the plaintext version of the ENGINE wasn't puting Machinery (ACM) and the IEEE Computer the entire magazine; a state of affairs that obviously Society. couldn't continue. We'd already considered dis tributing the electronic edition in more elaborate Clearly the ENGINE, from this day forward, must formats, such as ASCII PostScript or Microsoft commemorate - or at least take note of - signifi RTF, but every such choice or workaround cant anniversaries in the history of computing. seemed restrictive. There will be so many, especially after the turn of the century, that we'll need help keeping track of The perennial innovator in graphics, Adobe even the major dates. This is a job for our staunch Systems, now seems to have provided an ideal solu friends the spotters. tion with Acrobat cross-platform page mapping software. Files can be printed to the Acrobat If there's an imminent anniversary worthy of engine, which creates bitmaps of the pages, natu notice, please, don't hesitate to remind us! Send e rally including any graphics. On the subscriber's mail to [email protected] with the exact date, the end of the wire, these bitmap files can be read with particulars of the event, and - if you happen to - and printed by - the Acrobat Reader program, have one - a source for background. Thanks! which is widely available and free to individual users. ADS, ANYONE? Thanks to the Acrobat Distiller and Reader, the print and electronic editions of the ENGINE will We've been wondering lately if the ENGINE be in step once again. As soon as we can convert could be improved by accepting advertising. Cer the files to Acrobat format, we'll make 3.1, 3.2, tainly there are potential advertisers who would and later issues available bye-mail to subscribers appreciate the chance to reach the ENGINE's and through our Web site to visitors. The latest highly qualified audience. We, in turn, might as version of the Reader will be available through a well concede that a classified section could easily be link from the CHAC's Web site. We're grateful to supported by the number of enthusiasts who ask us Adobe Systems for its part in making pictures, as about buying and selling historic micros .... and oc well as words, an indispensable part of the Net's casional non-micros. Finally, the revenue from ads worldwide language. could postpone price increases. (Looked at the cost of paper lately?) Our current intention is to include a small section of classified advertising in the May issue, and pro ceed judiciously from there. The ENGINE would appreciate hearing from anyone who feels strongly about advertising in its pages - and, of course, from anyone who'd like to take out an ad. But we recognize this as a significant change in policy and we'd be reassured by the consent of our readership. February 1996 The Analytical Engine Page 3 NOT A COMMERCIAL, THE 930 COMES HOME just some sage advice. We've noticed that At 8:30 in the morning of October eleventh a ENGINE subscribers very often order back issues, moving van, pulling onto a blacktopped lot in the and in full sets at that. Naturally, an order for the South Bay, finished its fourteen-hundred-mile trip complete ENGINE is one of the most flattering from Table Mountain Observatory in Colorado. testimonials we can receive. Inside Mike Byrd's formidable truck (and not really filling much of it) were a dozen racks, a con All good things come to an end, though, and our sole, peripherals, tapes, docs and cabling - in other back issues are no exception. At this moment words, the SDS 930 permanently loaned to the there are probably twenty or thirty full sets left in CHAC by the Space Environment Center in storage. We don't intend to reprint them, at least Boulder, Colorado. After a year and more of not in their current form. planning, outreach, publicity, and (at last) frantic If you'd like a set - nine issues, 1.1 to 3.1 - send a fundraising, the X-PROJECT came to a successful check for US$51 to the Palo Alto address. This conclusion with a unique and important small amount includes first class (domestic) return mainframe safely in storage. postage. But do it today! This 930 was the fifteenth of (probably) 168 con FOUR MORE yEARS. ... structed. It was delivered from Scientific Data Systems in Santa Monica to the National Bureau of The turn of the century is imminent, and with it, Standards in Boulder, in December 1964, and in the dreaded 1999 Anomaly in full force. The stalled in a low cinderblock building called I-l0C, CHAC, as it happens, is interested primarily in on Table Mountain, a windswept mesa in the significant hardware that will be scrapped because shadow of the Rockies. It became the main data it misinterprets the dates of the new millennium. acquisition computer of the early-warning program But plenty of DP professionals are still working called HANDS, for High Altitude Nuclear Detec with live, running computers - significant in quite tion System. This was not trivial work, and in its another way - that have· this nasty little kink day, the 930 was about the only commercially hidden deep in their opcodes. available computer capable of it. In 1993, when we published the first CHAC FAQ SEC and the 930 settled into what would be a and the earliest issues of the ENGINE, we called startling thirty-year relationship. The computer attention to this fact and felt like a voice crying in was too good to let go, and the Center's techni the wilderness. Well, no more! The Year 2000 has cians and programmers labored to make it better. become such a hot topic, in this context, that some SDS factory diagnostics required that running tasks people are referring to it - with more urgency be shut down; SEC wrote diagnostics that could than precision - as "Y2K" for short. Programmers run while data acquisition continued. SDS support who need deep wizardry to keep their mainframes was costly and involved the delays of travel; SEC running can now find a lot of useful knowledge, brought hardware support on-site by learning to most of it on the Net. perform it themselves, for everything but the two Peter de Jager, with the sponsorship of the Tenagra vast Bryant drum memories. Corporation, has created a Web site - The Year 2000 Information Center - "to provide a forum Thanks to SDS' rugged design and construction, for making information available about the year and SEC's razor-sharp maintenance, this computer 2000 problem and for the discussion of possible is pristine today; it may even be bootable. It is, ab solutions." There's an associated Internet mailing solutely, imposing and beautiful. What follows is list, too. Browse the Information Center at not the whole story of the 930, which would take a http://www.year2000.com/and be grateful for book, but a few details and pictures that will let what you may find, because this clock is really you - advisers, donors, and members of the ticking! CHAC - know what you accomplished. - KC Page 4 The Analytical Engine February 1996 Fourteen hundred miles .... All photos by Janette Kiehn out of Boulder, van driver Mike Byrd pulled his five-axle truck - with deceptive ease - into the parking lot where we waited. The October morning was bright with yellow leaves and banners of sunlight that made Northern California, perpetual chameleon of the seasons, look like New England at its best. About a dozen CHAC stalwarts, distinguished affiliates, and guests were pacing impatiently, talking in huddles, or stopping by a tailgate for doughnuts and desperately needed coffee. As Mike put the truck where he wanted it, and the crew dragged out the long aluminum ramp, I thought about the history and purpose of what we were getting .... PURPOSE From the time its ARPA contract was written in 1963, this computer had a single aim in life: data acquisition. The Space Environment Center's dishes pointed every which way, and strip chart flowed ceaselessly from recorders, as the 930 took in data from satellite teleme try, magnetometers, radiometers, radiotelescopes, and about every kind of radio antenna from VLF to HF. All of this was intended not only to monitor, but to forecast, conditions in the upper atmosphere and in space, however they might influence the Govern ment's state of readiness. The system's reliability was paramount, since downtime would create gaps in recordings that were meant to be continuous. A near-perfect uptime record was achieved with the help of redundancy, clever pro gramming, and as much on-site maintenance as pos sible. Native ruggedness helped too - not only of the SDS hardware itself, as illustrated by these tape drive motors, but of the numerous special-purpose peripher als built by the Center's engineers. Today's answer to these demands might involve a duplicate computer, but the SDS' full system cost of nearly $1 million made that unthinkable in the mid-sixties! Tape drive motors. February 1996 The Analytical Engine Page 5 HARDWARE The 930's hardware was and is solid enough to drive nails. Mil-spec TTL on thick boards is all soldered, without a trace of wirewrap. Finned heat sinks are everywhere. Precious metals are abundant - the silver mostly in conductors, the gold mostly in connectors. Each module has an individual power supply with a voltage meter. As a former SDS engineer remarked when he learned that this 930 had had a thirty-year useful life, "We built better than we knew" - and one look inside the racks will prove that they knew they built damn well. This lack of fragility, however, didn't slow the 930 down. A cycle time of 1.75 J.lsec, with most instructions requiring one to ten cycles, made this about the fastest small scientific mainframe available in its day. To keep up with the hardware, all of SEC's programming was done in raw machine language. System memory, in well warmed mag core, is 16K of 24-bit words - the same size as an Apple II plus. Installation of a second 16K rack was contemplated but never done. - Naturally, with 16 Kwords available to handle • .. ..... torrents of data, nothing could linger in core a split 01; • - •••• second longer than necessary. As much as possible ., was swapped out to the two capacious "RADs" - fifty-megabyte Bryant stainless drums, tunable only with a masterful hand and an oscilloscope, but marvels of engineering at a mere $50,000 each. Further away from the core stood seven seven-track tape drives, IBM-compatible at 556 bpi. Eventually these were supplemented by the interesting Magpak (at right,) a dual-cassette drive. Unfortunately less re liable than the reel-to-reel drives, the Magpak was still handy for archiving programs and data. If the 930 as installed had a single greatest strength, The Magpak. of course it was I/O. Sixty-four analog inputs, thirty-two hardware interrupts and thirty-two sense inputs meant that this computer was rarely at a loss, however much it was asked to juggle. Although up to eight hardware 110 channels could have been installed, SEC made do with a single channel and relied on a gorgeous array of multiplexers to decode and drive thirty to forty channels of strip chart. Console 110 was handled originally by a KSR35 Teletype; later some Model 33's were added along with a pair of "homebrewed" VDT's, one of which is a Bendix (at right). The Teletypes, like the seven track drives, grumbled at constant duty and required scrupulous maintenance. All in all, the 930 is a curator's prize for any number of reasons - its exceptional standard of design and construction, its complex configuration, its pristine condition, and its sheer physical beauty. Someday, we hope soon, this exemplary device will be set up again for the whole world to see. Console Teletype and a VDT. Page 6 The Analytical Engine February 1996 WRAPPING IT UP Getting the 930 into its locker was hard work made easier by racks on casters, but the whole process is still listed under "Don't try this at home." As soon as the moving crew had shifted everything from the van to the locker - with the kind of determination prevalent in action thrillers - it was the turn of the CHAC faithful to swarm over the pieces, pull off yards of ductape, remove some non-historical labels, and check the inventory. Generally, the only part that showed deterioration was the foam insulation in the cabinet doors, which we removed rather than let it crumble allover the hardware. We also spent a while just gawking! Some of us were tempted to slit open boxes of docs and parts, but they were so nicely taped that (sigh) we just put them on top of the racks for the moment. As I write this, the 930 is in stable condition, with a steel roof overits head. The container is ventilated and we're hoping for relatively dry weather in February and March, after which the South Bay's rainy season is generally over. Some parts still need a good cleaning, some or all of the docs will be taken to archival storage, and the racks should Reading the console message (see cover). ideally be wrapped in plastic. We hope to report L to R: Len Shustek, Edwin EI-Kareh, that this work is finished in the next issue of the Bob Crosby, Kip Crosby. ENGINE. TIGER TEAM PARTY While we're on that subject! The preparation of the 930 for storage should be finished between now and May, if at all possible. An enthusiastic tiger team could probably finish the necessary cleaning, lubrication and packing in one whole Saturday or Sunday. Experience in electronic assembly and maintenance will be preferred but not required. We haven't picked a date yet, but if you'd like to volunteer, send e-mail to [email protected] or leave a note on our Web page at http://www.chac.org/chaclindex.html. TIA! February 1996 The Analytical Engine Page 7 HEROES' LIST The 930 was saved by generosity and hard work. Some of our heroes have been supporters of the CHAC for months or years, some stepped forward to introduce themselves when this rescue became an all-or-nothing proposition. The greatest reward of this project is the safety of the 930; second only to that comes the pleasure of saluting the rescuers. Many thanks to these principled men and women who gave the cash, the logistical help, the advice and the physical work that did the job. Anonymous Everybody at Space Environment Center, National Oceanic and Atmospheric Administration Allan Alcorn, Silicon Gaming Mark Lentczner, Glyphic Technology Gwen Bell, Computer Museum Frank McConnell, Attachmate James Birdsall Steven McGeady, Intel Corporation, Intel Architecture Labs Mike and Paula Byrd and crew, Adas Van Lines Nancy Mulvany Robert Cousins, AIM Technology John Novitsky, MicroModule Systems Bob and Enid Crosby Max Palevsky Ted Drake Martin Reynolds George Durfey, Perham Foundation Arthur Rock Edwin V. El-Kareh, AB Networks, Inc. Bill Sarill Ralph Gorin T. J. Seaney, Golden Van Lines Alan Hawk Len Shustek Dewayne Hendricks, Warp Speed Imagineering Roger Sinasohn, Roger Louis Sinasohn & Associates Jerry Jensen W. Bennett Smith Janette Kiehn Marie Sullivan, IBM Ed Klein Tim K. Swan, ZuniTech Jerry A. Klein, The Chatham Group Larry Tesler Don Koijane, Perham Foundation Chuck Thacker Ted Laliotis Bill von Hagen Mark Laubach Pierluigi Zappacosta, Logitech Corporation Page 8 The Analytical Engine February 1996 So I figured, let me just call on people who want to COMPUTERS FOR THEIR talk, but restrict them to the kind of things they OWN SAKE: would say for introductions. And then when people talk, I'll ask them to stand up so that they From the Dompier Music can be identified. If there had been some way to to the 1980 Computer Faire make them wear a number, I would have done that, but I didn't know how to do that. This was a An Interview with Lee Felsenstein kind of secondary information transfer and I got it (Part 2) from Community Memory, which was all about KC: I recall that there was considerable appreciation secondary information. Primary information was for [the Dompier Music} at the time - that the to be imparted in face-to-face discussion, or on the brilliance ofi t was immediately recognized. phone, or by various non-broadcast means. Sitting down in a meeting, and not saying anything unless LF: Yes. I proposed that he be awarded the called upon, was the price you paid for some very Stripped Phillips Screw Award, for finding a use sparse information that pointed you to the person for something previously thought useless. That was you wanted to talk tOj but that only began the a turning point, in effect, and later, when he wrote communication, it didn't complete it. So I wanted the software for Processor Technology, he put to go through several cycles -mapping, then Dompier Music into the game program - Target random access, then mapping, then random access - that he wrote for the Altair with the VDM - and the more cycles we could go through, I [Video Display Module]. But I'm getting ahead of figured, the better we could converge. But they myself. never came back from the first random access HOMEBREW GETS A NAME session. Enough people had introduced themselves, and I said "Okay, now talk among yourselves," and The third, and significant, meeting of the Club, it was like a cocktail party with no booze - every which didn't have a name yet, was at Peninsula body knew who they wanted to talk to, and why, School. The fourth meeting - I believe - was in and that was the rest of the meeting. It wasn't pos the orange room at SLAC\ and that was when we sible to get them back together. So I decided that tried to figure out what this organization was would be the mode of organization, and it was called, and we settled on Homebrew Computer very productive indeed. Club. At either the fourth or the fifth meeting, Gordon French announced that he could no longer A PROCESS IN PROCESS run the meetings because he was going to Balti My job in all that was actually to do comic more to work on a long-term contract with the routines and, when necessary, disarm certain types Social Security Administration. Either at that same of people. There was always some guy who would meeting or the next one - probably the next one get up, sweating, and feel like this was his big since I don't recall that Gordon was there - Marty chance to say everything at once - as if he were Spergel in the audience suggested me [to succeed giving a sales pitch and wouldn't have another Gordon] and I didn't say no, and everyone else chance. I also had to prevent back-and-forth, yes it seemed happy with that, so I got up there and is/no it isn't, discussions, because you could see the began to run the meeting. I started out by going lights go out all over the audience when that around the room, having people introduce them happened. If anyone started dominating the con selves and describe what they were doing, and that versation with a single topic, half the audience was the origin of the convivial process that I de would go to sleep. So I was strictly the facilitator, veloped. But at a later meeting, which must have and I warned people, "Don't come up afterwards been in SLAC auditorium, I tried to do that and it and ask me who said this or that, because," and took so long - I think it took almost the whole thenI would slip into my comic persona, "I'm just meeting to go around the room. thinking about my image up here and I don't know who said what. So you look for people and keep track of them yourself." I had to say that every time, along with "Don't ask me, don't talk Stanford Linear Accelerator Center. 1 February 1996 The Analytical Engine Page 9 to me, you're talking to everybody - and keep it K C: Physicians? short, you have ninety seconds that I will enforce capriciously." Of course there were exceptions; Jim LF: Yes, medical doctors who had this as a hobby. Warren would start delivering this gossip column, Some of them probably should have been engi what he called his Core Dump, and we never en neers. There may have been people from other in forced a ninety-second limit on him. I consider that dustries who didn't identify themselves, but the process that one of my best designs, because we doctors always referred to themselves as doctors. learned that in that environment, the expertise was Here we all were, doctors included, putting com distributed but it was there and could be relied on. puters together, trying to get them to work, Sometimes we had speakers, and we'd put on the without much of a fixed goal as to what we wanted speaker first, then we would do the mapping, and to do with them once they worked. That's another then we'd have random access. But if the speaker important point. The creativity that resulted from didn't show up, I would say "Okay, we had a working without goals was a very broad creativity. speaker scheduled and he was going to talk on this Working with fixed goals could produce creativity topic. Who here knows something about this in depth, but it would be tightly focused - very topic?" One or two hands would go up, I would narrow. This was broad. So many people were at call on somebody, and they would recite a fact. I'd tending Club meetings, and yet so few of the say "Okay, what does that bring up for anyone people who got involved with personal computers else?" And someone else would qualify that fact could really have justified them on any rational and give another fact. Through this process we grounds. At that time, the primary entrance into could construct a lecture on almost any topic out personal computing was a white lie that you were of that audience. In effect it was rather like object going to spend $397 on an Altair kit from MITS, oriented programming, in that we had a collection put it together, do some programming and then of objects that included everything we needed, it keep the recipes or checkbook or control the was just a matter of figuring out how to organize house environment with it. In practice that was them; and in this case they organized themselves. certainly not true. Your typical real experience During this time, various elements got organized would involve spending several thousand dollars, as into twenty-three companies that we could iden well as a lot of time to master a huge amount of tify, one of which was Processor Technology. information. Some of the things that you ended up learning probably wouldn't be useful in any other "SATISFACTORY AND part of your life, but they were very interesting to DIRECT INVOLVEMENT" learn. KC: So these people were, on the one hand, not quite KC: "What time, exactly, are we talking about? ready to concede that they were building computers LF: The most fertile time for the Homebrew for the sake ofi t, but on the other hand, not at all Computer Club was from 1975 to 1977 or possibly clear on what they intended to do with the computer 1978. During that time there was no venture capi once they had it. tal coming into this industry. The whole idea of LF: Right. It was a mass learning experience with making money at [microcomputer design and pro definite justifications which were not easy to make, duction] was an alluring but very dubious proposi certainly not on economic grounds. The impulse tion. It was being done by people who loved to do was to have a certain amount of say in one's life, it and who wanted this to be their main chance. I certainly over the technological systems that af think most of the people attending were working fected us all, and they knew it was important to in the electronics or computer industry, at various make computers and the surrounding technology levels which gave them either no access to com accessible to as wide a range of people as possible. puters, or just enough access to be frustrating. This was an article of faith that People's Computer They were looking for a more satisfactory and Company had been promoting, but it had also direct involvement. Meanwhile, the largest contin been discussed for years in science-fiction litera gent I could identify outside the industry were ture, and science fiction had risen to become a kind physicians. of ideology related to computer activism. Many