Computer History: Building the UNIVAC 1108 Computer, Twin Cities (1965-1968) Sperry Rand, UNISYS
HTML-код
- Опубликовано: 11 сен 2024
- Late 1960's video showing the Sperry UNIVAC data processing facility in Twin Cities, Minnesota, manufacturing the UNIVAC 1108 computer. Great footage of creating a CORE Memory block, and manufacturing of Printed Circuit Boards ("PCB"), and the quality assurance lab in action. The original film was a bit rough, but cleaned it up as much as possible. (We are looking for a better copy of the film, so please let us know if you are aware of any.) Run time about 16 minutes. Hope you enjoy this inside look at creating the UNIVAC 1108! (Computer History Archives Project)
Sperry announced the UNIVAC 1108 in the summer of 1964 and delivered the first one in late 1965.
An improved version of the 1107, it used a combination of transistors and integrated circuits.
This film shows a rare look inside the Twin Cities Data Processing Division, Minnesota.
Topics included are
- Wiring Core Memory Planes
- Assembling Printed Circuit Cards
- Hand Assembly Manufacturing Line
- Automated Wire Wrap Machines
- Quality Assurance Testing
- UNIVAC 1108 Console Testing Room
- Peripheral Devices; Magnetic Drum Memory
- an Image Gallery.
For more information on the UNIVAC 1108, see also:
Wiki:
en.wikipedia.o...
Unisys History Newsletter, Volume 3, Number 2, April 1999 (rev. Mar. 2000)
by George Gray (this link loads slowly)
web.archive.or...
Adapted from the Sperry UNIVAC film "Electronic Servants" of the late 1960's.
(C) Unisys Corporation
with additional thanks to:
Hagley Museum and Library,
www.Hagley.org
Computer History Museum,
www.computermus...
VIP Club - Information Technology (IT) Pioneers
Retirees and former employees of Unisys, Lockheed Martin, and their heritage companies
www.vipclubmn.org
I spent many years working at Univac (Unisys) Roseville,MN. 1108’s were the first computer I ever worked on as a tech. Eventuality wrote software for this computer and it’s many variants. I also assembled and tested the very first CRT display on a 1108. My mother in law worked there also and worked stringing those tiny wires through the core memories sitting at a microscope many hours!
What programming language do you used?
I started working on a Univac 418 III. It was used by the city. Started out writing Fortran. But also did work on the machine in Cobol and assembly.
While yet in high school in 1969-70, I taught myself FORTRAN IV on a UNIVAC 1108 at Collins Radio Company in Newport Beach, California, using a book called "Fortran IV, Self-Taught" by Mario V. Farina. I was fortunate to be one of the only ones in our school to have such computer access, more so than any of the teachers. Collins sponsored a Boy Scout technical Explorer Post, and provided access to amazing technical resources. I had 24-hour on-site access to all the computer time I wanted. The actual Univac 1108 was is Dallas, Texas, but connected by high-speed data lines to a Collins-built I/O processing computer that managed the very high speed card reader and line printer. And they had several IBM Model 29 card punches which I used to punch many hundreds of cards for the programs I wrote, and that another fellow scout collaborated on. During the second week of a two-week Christmas respite from school, my friend and I wrote and ran a 500-card Fortran digital logic simulation program. I still have the card stack and the printouts from the 132 column line printer.
Hi dwbogardus, teaching yourself FORTRAN is no small feat! Very cool. Thanks for the story! ~ Victor, at CHAP
Watching them painstakingly running a needle and copper thread through the cores or making those little copper-wound donuts _by hand_ is just... wow, I can't imagine more tedious, mentally wearisome tasks.
For a short while I worked in a factory making radio aerials. I put five parts in a jig, pulled a handle, took the aerial out, put five parts in a jig, pulled a handle...
Those types of delicate and repetitive operations would be unbearably tedious to many of us, and most men, but it seems that many women have both the skill and the temperament to tolerate such jobs. Many have what is called "tweezer dexterity", referring to their precise fine motor skills. Were it not for those skills, ability to concentrate uninterrupted for long periods, patience and precision, women would never be able to the the lace-making, crocheting, needle-point, and knitting they have long been justly famous for. Core memory wire weaving fits right in to that same tradition.
Also why core memory was so expensive at that time.
@@kevinmiller4486 would be even more expensive now (relatively) if it was still made today, even though we'd make robots do the work nowadays
Great video - brings back memories. My first programming experience was on an 1108. Computers used to be so much more fun!
I was a Univac CE starting in January 1970 in Huntsville, AL at MSFC, moving to El Paso, TX in 1976 to continue with the 1108's at WSMR. We also had 1100/80 across the hill at a NASA site (WSTF) and 1100/60 at HAFB during these years into the 1990's. We had eight 1108's at WSMR. The last 1100 series computer at WSMR was the 1100/90, a water cooled computer. Starting with Remington Rand Univac it evolved into, Univac, then Sperry Univac, then Sperry, and in 1986 was purchased by Burroughs and became Unisys. My employment with Univac and Unisys lasted until the summer of 1994 - 24.5 years. A great career and memories. Wish I could do it all over again. I believe the core memories (64K) were $500K each and the 1108 CPU was $750K each. $2.75M for one CPU and four memory cabinets, plus peripherals.
Hi @ElPasoTubeAmps, thank you for the great feedback and some of your impressive tech background(!) It sounds like you had a great career, spanning many tech advancements and the evolution of RR Univac. I had to look up some of your abbreviations (I learned some new ones), so thought I'd share them below, for those who do not know them already. Thanks again! Glad you found our channel, hope you will explore it further! ~ Victor, at CHAP
MSFC - Marshall Space Flight Center Located on the Redstone Arsenal in Huntsville, AL, MSFC, professional engineers and scientists with expertise in rocket propulsion technology, and building launch vehicles, spacecraft, and scientific instruments. -
WSMR -- White Sands Missile Range - Originally part of the U.S. Army's White Sands Missile Range, what is now known as the White Sands Test Facility was transferred to NASA in 1962 to test the Apollo Command Module and lunar descent engines. ---
WSTF- White Sands Test Facility, offers numerous ambient pressure and altitude simulation stands to test rocket propulsion systems. -
@@ComputerHistoryArchivesProject It was a great video on the 1108. It seems like it all went by in the blink of an eye and it started for me 54 years ago. 20 years just at WSMR and now I have been gone from there almost 30 years. I did move on to other contracts at INS (Immigration and Naturalization Services) which was part of DOJ but after 911, DHS (Department of Homeland Security) was created in 2002 that combined Border Patrol and Customs and became, Border Patrol, ICE, USCIS, CBP and HSI (another bunch of abbreviations...) I was with them for 11 years and finished up my last two years as a technical writer on contract at Fort Bliss. I got to work at Sandia Labs and other installations and see the big CDC and Cray computers along with the Univac systems they had up there. Quite impressive. I am still active as a (Ham) amateur radio operator (WA4QGA) and build audio and RF amplifiers and have a RUclips channel, ElPaso TubeAmps. Thank you for your video. Very well done and brings back a lot of wonderful memories.
Hi @ElPasoTubeAmps, I have read that the creation of the Homeland Security Department was quite a massive reorganization, but somewhat largely one of name changes and wording changes in mission statements. I do not know first hand, but sounds fascinating. It would be great to know what different computer systems all these agencies had and how they did, or didn't talk together or provide real-time security data to those who needed it. An amazing scenario. - sounds like you are quite active in the tech area, as you say. A great way to keep the brain cells sharp and focused. Thanks again for the great (and educational) feedback! ~ Victor, CHAP
Thanks once again, CHA, for presenting these films to us. Seriously appreciated. BTW, that's one serious beehive hair style at 1:00 ! Looks like they are wiring the microcode at 6:34.
Hi @headpox5817, Glad you enjoyed it. Yes, beehive hairdo is classic! That wiring process must have given the lady tech some nightmares. The detail required is frightening.
Love the ‘unit of data’ - 36 newspaper pages in a third of a second. ;-)
'Speeds and feeds' - an expression we used a lot way back then.
Worked for Sperry Univac 1975 - 1982. Worked on 1108 at Computer Sciences Corporation, El Segundo, California.
I was a Univac site engineer on the predecessor to this model - the 1106, which had very similar architecture. We used to do component-level repairs so I remember the tool shown at the end, the card puller, which was often used with a card extender to allow access to the card components while the system was running. Despite the card connectors being gold plated, bad connections would sometimes cause faults that could be cured by simply pulling the card and plugging it back in again. Being a site engineer gave you great experience, as you were expected to fix everything from a Fastrand drum storage system to a punched card reader. I still have some of my old 1106 system manuals that I used way back in the late '70s.
Hi Pat, sounds like fascinating times! I always wondered about maintaining these giant machines. It must have been stressful for the service engineers to try and fix problems on-site, with the DP folks impatient to get the system up and running again. I bet you learned some great skills! Thanks for your feedback! ~ Victor, CHAP
@@ComputerHistoryArchivesProject It wasn't that bad Victor. Remember these were not "live" systems - the users submitted jobs to a queue, and were used to waiting several hours to get the output back. We ran diagnostics on the system every morning that often flagged issues before they became critical. Plenty of spares on-site along with excellent support by phone also helped, along with the knowledege that if the system was down for more than a few hours, a support engineer was getting on a plane somewhere to fly to your assistance. But yes, it was an fantastic learning experience.
Very cool! Thank you for sharing that. ~ Victor
With all due respect, sir, you are misremembering things.
The 1106 was *NOT* the predecessor to the 1108 ... the 1108 was the predecessor to the 1106.
The 1106 was basically a slowed down 1108 which could be sold for less.
When the core memory of the 1108 and 1106 were replaced with semiconductor memory, they then became the 1100/20 and 1100/10, respectively.
Similarly, when the core memory of the 1110 was replaced with semiconductor memory, it was re-designated the 1100/40.
All of the careful attention to detail and those long test cycles really contributed to the bulletproof reputation that so many mainframes (theirs, and others) earned over the years.
The original opening sequence is priceless. :)
7:40 "Reliability does not just happen, It must be earned."
That's wrong, reliability is made. LOL.
It talks about reputation. You don't 'make' reputation. You earn high reliability reputation. Component reliability is completely different.
I can watch these all day 🥇
I worked in that factory for 3 months in early 1975, rebuilding plated wire memory cabinets. First job in 1968 was repairing Univac II at Ontario Hydro, later Univac III and then 1108, 1106, 9200/9300, DCT2000, 1110, CS/P and finally 1100/80 in early 80's. Lots of great memories, troubleshooting down to the IC/transistor.
Hi Gerrit, thank you very much for your feedback and info on the video! Glad you had a chance to see some of the old factory again. I bet they were some very interesting times at the Univac plant! Thanks again and hope you enjoy our other Univac and vintage tech videos! Victor, at CHAP
I would give about 80% odds that my uncle shot the footage used in this film. And if so, I almost certainly posses the camera he used. I myself participated in a project to decommission a pair of 1108s with drum storage at RAF Croughton in the UK and replace them with refurbished Univac 1100/60 systems. I wish these films had credits that listed the people that contributed to them, but alas they do not.
That's a fascinating story. That must have been interesting work. You're right about the lack of credits. That would have been a nice touch to see everyone's name who put it together. ~ Vincent, at CHAP
The assembler made threading the magnetic core memory look easy!
Wonderful content.
Thanks for Sharing✨
Cheers
Thanks for visiting! Glad you enjoyed it. ~ Victor, at CHAP
I first worked on (dual) 1108s in 1970. Super multiprocessing machine with up to 3 CPUs and a number of independent I/O processors possible. Unfortunately the OS software was still catching up at that time (Exec level 26). Went on to work at several sites with 1108s and when I joined Univac in Roseville, MN in 1980 the slightly updated 1108 CPU was still in use as the 1100/20.
I still have my coffee mug given out on the manufacturing of the 1000th Type 3011 processor (basically the 1108 and its derivatives).
Hi Steve, thank you. Sounds like you had first hand experience with these. I bet they were interesting times. ~ Victor, CHAP
Well, when you're young everything is exciting.
Fun to see the 494 computer there. A popular large computer originally derived from the 460 and military models. I still have the maintenance panel from one of the last in service in the 1970s in my garden shed.
Sounds like a good teaching aid or wall display. : )
My father worked at that Univac as well. Used to drop me off at college in Roseville on his way to work and pick me up after he was done working back in the early 80's.
Remember an "open house" there when I was little - got a cool computer print out of cartoon characters. Only time I ever got to see where my dad worked.
Him and his buddies bought Commodore 64's and shared floppy disk copies of the games. Had a whole binder filled with games to play.
He had a stroke during his lunch one day unfortunately - and never was able to go back to work.
My father worked for Sperry Univac/Unisys and I know he was at the Twin Cities site several times for training - really cool to see this video!
Hi Patrick, very glad you enjoyed the video! Thank you for visiting our channel. We have several Univac related videos, it is one of our favorite product lines!! ~ Victor, CHAP
Супер! Большое спасибо! Очень интересно!
Уникальные кадры!
Thank you very much. ~Vk. (Google translated: "Super! Many thanks! Very interesting!
Unique shots!.")
A sperry univac was the first computer I ever used back in the early 1970s My father was an executive of a company that made glass printed circuit boards and used a timeshared univac for some of the accounting systems. I , still in my single digits years, would go in to the office on Saturdays with him (probably to give my mother a break) some times the computer operators who ran the system would put me on a paper tty terminal and let me play with it. I think they had something like the early Eliza program running so it would appear to talk to me. Other times it was more of a calculator program (this was at a time when most “adding machines” were still more mechanical than electric.) so having a computer do even very simple math that an 8 year old would understand was a novelty.
Hi S. Back, excellent feedback. Thank you for sharing your experiences, sounds like some very interesting times! ~ Victor, at CHAP
Folks these days are long gone, this is when America was Great. We americans built and designed these computers while the rest of the world was still 3rd or 4th world. We did not think of sending off these device to be built in some foreign country. America today is NOTHING like it was during this time period, these folks in this video I am sure was a truly proud American.
Hi D Zee, those were quite the days. Yes.
3rd world my arse. The electronic computer was a British invention. Like the gas turbine, railways, television, the concept of a modern industrialised society itself...
Well, somewhat true. But even in 1960, semiconductor manufacturers were moving to low-wage areas like Hong Kong and Singapore. There was PLENTY of profit margin in a time when computers cost millions of dollars each, and enough to pay everyone very well. I remember my father who marvelled that the secretaries at his computer company would take long lunch breaks because they just HAD to go shopping at the upscale mall nearby. Not so anymore.
been a Univac fan since i been confronted with one of their datacenters at Frankfurt airport in 1978. Still have a bunch of Univac promotional prints. Stayed in IT ever since, and postpunk 'course
I also worked for Sperry Univac from June 76 to Sept 78 at the 2276 Highcrest Drive, Roseville facility building 3. Did software for 1110 and 1180 mainframes.
Hi Raj, sounds like fascinating work! Roseville was the place to be for Univac back then! : )
This was great. The Univac 1108 was the first large computer I ever used, starting in 1975, as a student at the University of Maryland, College Park. The 1100 series was strange by modern standards. It used 36-bit words, ones' compliment arithmetic, and the FIELDATA character set.
Core memory was a major enabling technology for electronic computers, which had been severely memory limited until it as developed. The degree of manual labor required to manufacture core is astonishing, which underscores just how vital it was. If there had been any other practical choice...
James, thanks very much for your comment and info. Glad you liked the video. A question for you -- Do you recall if the machine had any maintenance or heat issues? just wondering. Thanks. Victor, CHAP
@@ComputerHistoryArchivesProject I honestly can't remember how reliable it was but it would crash sometimes. A bunch of us would be in the terminal room working on our projects late at night. Suddenly, the terminals (these were hard copy terminals, not CRTs) would stop at the same time, there would be an ominous moment of silence, and then they'd all print "Host not responding, probably down." A collective groan would rise up in the room as we all lost our unsaved work.
Hi James, thanks for that story. I can just envision this happening. Have seen PC networks go down like this, but usually due to main power outage, but the "ominous moment of silence" is just as deep. ~ Victor
In 1976, I worked with a DG (Data General) Nova II, which had core memory.
FWIW, I don't remember _any_ problems with that computer, other than it was slow, and only had 32KW of RAM. But it ran 4 VDTs (video display terminals, over RS232 9600 Baud serial connections), a 50 MB hard drive, a Kennedy 600/1200 BPI tape drive, a paper tape reader and two burpee paper tape punches.
I too was a student (EE dept.) at U of MD during the late 1960s. We started with the IBM 7094 II with cards. Then this fancy new Univac 1108 appeared. They installed TTY rooms in the computer science building and a remote site in the engineering building - I think it had 2 TTYs and a remote job entry machine. I had no classes that required its use but I got a student account so I could just experiment. I was not sure what commands to use. The operating system would sometimes remember my uploaded programs then the next day it would all be gone. One of my first programs was a lunar lander in Basic.
The narrator is the late Charlie Boone of WCCO radio in Minneapolis.
Thank you for spotting that! Good to know. ~ Victor, CHAP
I was a user on probability one of the last 1108 thru the early 1980s. We used it to mainly cross-compile and download 1219 code for a 1960s era Radar software we were maintaining. I think that the one we used was from the Pentagon. It took a team of 2 or 3 people to operate it keep it running for us.
I thought the name "Uniscope" was a little weird, until I realized that the monitor was not a CRT like a tv, but it was probably indeed an oscilloscope that was fed a very specific signal so it would generate text. No 540 or whatever lines, but each character was drawn separate. I wonder why, because television was already present in the 60's.
Those 3-wire magnetic cores, those are a cheap stripped-down version, the better and faster ones have 4 or 5 wires per core.
Hand-stuffing the PC boards was obsolete by 1968.
I tried using a Univac 1108 in 1972. It was unusable slow, taking up to 7 minutes for it to respond to each command. It was also totally insecure, anybody could read any file or crash the system.
Overall a very marginal combo of hardware and software.
incredible
These things must’ve been an absolute pain to make.
Can you imagine stringing the magnetic core donuts for 8 hours? ~ ~
Painful, I wonder if someone actually cut their hands on one of these.
Android smartphone is amazing good job Sir
My first job was at the 1108 BENCHMARK CENTER IN ST.PAUL.
I wonder by how many multiples was the 1108 faster and able to hold more bits in memory than the original Univac I ? Did Seymour Cray work on design of the 1108 before he set out to design his own super computer business ?
I'm not a Univac/Sperry Univac/Sperry/Unisys CPU designer and have never played one on TV, but assuming that the Wikipedia articles on Cray and the Univac I are correct, then no, Cray never worked on the 1108. He left long (in 1957) before the 1108 was being designed back in the days of the 1103 which was not at all like any of the 1100 Series computers like the 1108 (which came out in 1964).
Again, assuming that the Wikipedia article on the Univac I is correct, it had only 1K "words" of memory where a "word" was 12 "characters" where there's no mention of what a "character" is.
Since it mentions that each "character" was basically a decimal digit when used to represent numbers, I would *GUESS* that a character was something like Bardot (a 5-bit character) or Fieldata (a 6-bit character) which would mean that each "word" was more than the 36-bit words used on the 1108 and beyond and presumably represented numbers in BCD rather than ones complement.
The 1108 had about 262K words of memory except for some machines which were extended to use a bit in the Program Status Register (PSR) to act as an additional address bit so that 524K words of memory could be addressed.
So an apples-to-apples comparison of memory sizes isn't directly possible due to the difference in word size, but if you squint your eyes a bit (and call a Univac I word equivalent to an 1108 word), then the 1108's memory was about eight (8) to nine (9) powers of two greater larger than the Univac I.
That's roughly 2 to 3 powers of 10 for the more decimally minded.
As for speed, the Univac I was capable of about 2,000 operations per second (i.e. about 0.002 MIPS) versus about 1.5 MIPS for a single 1108 CPU (there could be up to three CPUs in a system).
That means that the latter was roughly 3 powers of 10 faster than the former in a single processor comparison.
Meanwhile, flying cars are just around the corner. And always will be.
True.... but in 1961, they had this... ruclips.net/video/E8m9Z010pM4/видео.html
All that is now in my hand😀
In 1985 I got the chance to tour a couple of these facilities in the Twin Cities. One not far from Ft Snelling was assembling some sort of custom computers for the Navy. The other was a brand new integrated circuit plant across the river from the airport. Is Univac still a going concern in the Cities?
Hi B. Laquisa, thanks for your comment. Sperry Univac was acquired by Burroughs Corp. in 1986, and they became UNISYS Corp. The "Univac" name is now owned by UNISYS Corporation. UNISYS does global IT problem solving, but I do not think they produce computer hardware anymore. They are based in Pennsylvania. ~ Thanks for watching~ Victor, at CHAP
There were a number of facilities in the Twin Cities at the time this video was made. Roseville, Eagan, the MSP airport facility whose name I don't recall.
Roseville and Eagan were sold off.
The last I heard, everything in sight is now near the airport although I don't know if it's the same facility back I was familiar with back in 1980.
As for hardware, they don't make any of the old mainframes anymore.
Both the 1100/2200 Series (Univac/Sperry Univac/Sperry) and A-Series (Burroughs) systems exist as emulation running on Intel servers.
thanks for sharing .
My pleasure, glad you enjoyed the video. ~ Victor, at CHAP
I was part of this scene once. I was hired by Honeywell in 1965 to train as a field engineer on their H200 mainframe computer. Spent 6 months in the Boston area for the initial training. While there, we had tours of the manufacturing facilities (some in old converted woolen mills) and saw lot of similar scenes as shown here. Once back in the field, would go on service calls with a big leather tool case with tools and vital spare parts in one hand and a portable oscilloscope in the other. It was an exciting time with always a 'next big thing' just around the corner. I moved on from hardware maintenance to customer support and finally software development at the end of that era.
flic.kr/p/7TEpjS
Hi "OldDogNewTrick", thank you very much for sharing this memory, cool photo too! Glad you enjoyed the video too. We've tried to find more on historic Honeywell computers online, tough to find, considering how much they were involved back in the day. ~ Victor, at CHAP
Amazing picture. I love computer history and honeywell is one of the greats. You should post storiesl about your career with honeywell.
Did i listen to this music when I was kid????
Were the cores 393216 bytes total from what I heard?
It sounded like there were eight (8) core planes of 4096 bits of memory. That would be 32,768 bits (not bytes and not words).
i thought this was a modern video with ppl building a vintage computer 🤣
It had one's complement arithmetic, was the weird thing.
Ummm, no it wasn't/isn't.
And I wish that people would stop trying to make it out as if it were (which is to say people who more often than not usually have no experience with it on an 1100/2200 Series system).
Yes, ones complement has two zeroes -- a positive one and a negative one -- but the 1100/2200 Series ones complement *SUBTRACTIVE ADDER* (meaning that one addend was complemented before being subtracted from the other addend to perform an addition) would automatically suppress the generation of negative zero unless one of the addends was negative zero to start with.
IOW, everything worked just the same as a twos complement adder in terms of the results.
Positive one (+1) minus positive one (+1) generates positive zero (+0) while positive one (+1) minus positive two (+2) generates negative one (-1), *NOT* negative zero (-0), just as one would expect.
Look at all that tedious hand work. We sometimes forget that automation is just as much to blame for the loss of manufacturing jobs in the USA as outsourcing is.
Jobs in Minnesota?
1:00
Is that Luke Skywalker and Matt Damon?
Had the same reaction(!) Thanks! ~ Victor at CHAP
The core memory was wired by hand... That's insanely tedious.
Only women make memories?