Talk:UNIVAC

Univac page consolidation
Moved from Village pump on Thursday, June 12th, 02003.

Help! Help! Help! I'm drowning! :-S I thought it would be a good idea to direct all the precursors and descendants of Univac to that central article, since it seemed to provide a comprehensive overview of how all those companies merged... But now when I read closer, it seems that 'Sperry Corporation' separated from the pack, only to swerwe back into the fray very late in the game. It definitely isn't as clear cut as I Thought it would be. I Really hope I haven't gone overboard; It just seemed that many of those articles were just linking into each and every other of each other, when they really only covered the same historical ground. But I starting to get doubts, did I simplify things too much. Any comments wellcome. -- Cimon Avaro on a pogo stick 13:27 9 Jun 2003 (UTC)


 * It was also a rather interesting exercise a while back when I untangled all the Univac, UNIVAC, and UNIVAC I links that pointed to each other. I sure this is far from complete and often gets retangled. -- RTC 18:29 23 Jun 2003 (UTC)


 * Still unraveling these names, I hope they settle down soon. The current usage is:


 * UNIVAC - Article on the company
 * Univac - Redirect to the article on the company
 * UNIVAC I - Article on the computer (although it was originally called UNIVAC, when it was the only computer they made)
 * -- RTC 19:43, 11 May 2005 (UTC)


 * I think that the article needs to be rewritten to eliminate the first person (e.g. "I joined ECC... " I put {wikify}, but I'm not sure if that is the correct term.   Bubba73 3 July 2005 23:34 (UTC)

There seems to be a Sperry article, a Unisys article and a Univac article. My father worked for this company for over 30 years, and it seems to me that before Unisys became Unisys, it was Sperry Univac. Hate to throw more mud into the name game here....I'll have to research that. NickBurns 15:50, 21 July 2006 (UTC)

As far as i know (working for unisys) there was no company calles univac, the name univac was a synonym for computer back in these days. Due that Univac is a product line there are some strange issues because of all that mergering in unisys history, first there was Remingtong Rand, and they took over the Eckert Mauchly Corp, which produced the eniac, first of all they introduced the Univac 1 wich was the first commercial mainframe after that. in 1955 sperry rand an remington rand merged then the univac product came from sperry rand, aka sperry (they dropped the rand somewhere) in 1986 the merged with bouroughs and became unisys see also http://www.unisys.com/about__unisys/history/ after that Univac was dropped and became the sperry/unisys 2200 series. which was renamed to Clearpath in 1995 (still compatible with the univac operating systems and also to the bourroughs operating systems)

and as far as i know there was a business division of Sperry called Sperry UNIVAC, but it was juts a business division

greetz digital


 * There most assuredly was a company called Sperry UNIVAC. I worked there for a time until it was sold off to Control Data. — Quicksilver (Hydrargyrum)T @ 23:35, 17 January 2017 (UTC)

UNIVAC Solid State is NOT an upgraded UNIVAC II
The UNIVAC Solid State and the UNIVAC II are totally different machines with no relationship. The design of what came to be known as the UNIVAC Solid State 80-90 was began under contract to the USAF at the Univac laboratories in Philadelphia beginning around 1955. The engineers at UNIVAC knew that the vacuum tube era was over, and that solid state devices of some sort would be the next wave. At Univac's laboratories in Philadelphia, PA, work had been in progress for some time on the application of transistors and of magnetic amplifiers to computer circuits. When the contract was let for the new machine, the germanium transistors available were in short supply. In fact Univac had bought the entire production of Philco's product for another computer, the one-of-a-kind Livermore Advanced Research Computer or LARC. Experience with these led to the decision to use magnetic amplifiers in the Solid State 80-90. The SS 80-90 is believed to have been the first computer based on printed circuit cards. Each card, about 4" x 6" had four magnetic amplifiers on it. They were inserted into edge connectors with all logic in the backplane wiring. The amplifiers operated on a clock of 707 Khz. The first machine was contained in a cabinet about six feet long, five feet high and about 18" wide. There was a separate power supply cabinet about the size of a file cabinet. It was delivered to the USAF at Lawrence G. Hanscom Field near Cambridge, MA in the Fall of 1957 for use in communications processing. It was set up and running just a few days after delivery. The USAF asked that it be moved to a hangar to show to the public during Veteran's Day, November 11th, 1957. The author of this paragraph, the project engineer on the computer, likes to think that the SS 80-90 thus became the first portable computer, being moved on a fork lift truck. —Preceding unsigned comment added by Co-RegistrarL23 (talk • contribs) 00:35, 16 November 2007 (UTC)

Some more 1970s details for UNIVAC
I worked in a shop (from 1972 - 1977) that had a UNIVAC 9400 mainframe. The UNIVAC 9400 CPU was very similar to an IBM 360/30. We even used the IBM assembler language textbook to learn how to program it.

Facts that I can recall from memory about the 9400...

o It used Plated Wire memory (64K per cabinet, we started out with 2 cabinets (128K) and ended up with 3 (192K)). IIRC, the UNIVAC 9480 was the same as the 9400, except it used very early solid state memory.


 * The 9400 has a multitasking OS (we could run several batch jobs at one time)
 * We had two UNIVAC 1004 systems (with channel adapters installed) as unit-record devices for the 9400.
 * Original tape drives were 800/1600 BPI, and later we upgraded to 800/1600/6250 BPI
 * We had removable hard disk drive (20MB and maybe 30MB iirc)
 * UNIVAC 0768 line printer (rotating drum, just like the printer in the 1004's)
 * Some 1004's read (and punched) 90-column cards. The two that we had were the 80-column versions (rectangular holes, compatible with IBM systems). 90-column cards had round holes, 45 columns across the top and 45 across the bottom edge. They were the same card dimension as the 80-column cards, just different types of holes and distributions.
 * Our system supported assembler, Fortran, COBOL and RPG.
 * The operating system (called a supervisor) was custom assembled from macro statements. You coded a series of statements defining your IO devices and desired options, then assembled the supervisor and loaded it onto the proper location on a disk pack for loading.
 * There were switches on the front of the CPU that defined the boot target address (channel and device). We usually booted from disk, but I did try a tape boot one time and it worked (albeit slowly).
 * The operating system for the 9400/9480 was called OS/4
 * vaguely, I seem to recall that UNIVAC also made a 9300, although I never saw one.

I worked for Sperry UNIVAC in 1971-1972 on the AMME project. We developed on and for the 9400, and made occasional use of 9200s and 9300s. I expanded the article to include some of what I remember, but I did not cite sources. We seem to have a dearth of sources in general. There was also a 9400ASP if I recall correctly, but I never workes on it. It was a 9400 re-purposed as an intelligent peripheral controller for the 110x machines. Arch dude 04:06, 23 September 2006 (UTC)

UNIAC
Should the article also refer to the alternative "UNIAC" terminology? I have seen this machine called UNIAC in a number of places,.


 * That information appears to be bogus. The name UNIVAC has been used since the days

of Eckert-Mauchly Computer. --John Nagle 20:35, 17 May 2006 (UTC)

Sperry Univac Computers
There should be a separate category for Sperry Univac computers. There is a site page titled 30 Bit Computers that has quite a bit of history of these machines, including:
 * AN/USQ-17 (Naval Tactical Data System)
 * AN/USQ-20 (16 units built?)
 * CP-642A (UNIVAC Type #1206)(BUSHIPS)
 * CP-642B (UNIVAC type #1212) (Navy and Foreign Military Sales)
 * CP-667 (only single model built)
 * CP-808, (UNIVAC Type # 12130) (Marine Corps Marine Tactical Data System)
 * CP-823/U (Univac 1830) (Naval Air Development Center - only 1 prototype built for P-3 Orion)
 * 1830 Phoenix (Air Force F-111)
 * 1230 - CP-855/UYK (Apollo program)
 * CP-890 (Univac Type #18360 (Navy Poseidon Missile Boats)
 * CP-901 (Univac Type #1830A)(AN/ASQ-114 Anti-Submarine Warfare)
 * 1830B (German Navy Computer)
 * AN/UYK-8 (Marine Tactical Data System)
 * 1530 MTC (UNIVAC Type #1530) (Air Force Mobile Tactical Computer)
 * ARTS III IOP (Air Traffic Control)

So do we add these to this page, or create a separate Sperry Univac page? SunSw0rd (talk) 14:05, 25 March 2008 (UTC)


 * New pages should be created when they can be filled with appropriate encyclopedic content from verifiable sources (preferably multiple ones--pages that are taken from just one source tend to be AfD-bait). A list of machines produced by a company (or trivia about them) is not justification for an article--Wikipedia is not a collection of indiscriminate information.  A good article not just contains information but gives some sense about why the information is notable: what impact did the article subject have? why was/is it significant? etc.  If you can author an article that addresses the cultural, historical, or scientific meaning of the machines listed above, by all means create the new article and see how it fares.  But if your intention is a cut-and-paste job of a bunch of stats, you're probably setting yourself up for the disappointment of seeing the article get an unwanted decoration in the form of an AfD banner. Robert K S (talk) 15:27, 25 March 2008 (UTC)

1004 core memory size
The article says 961 characters. At 6 bits per char., that is 5766 bits, which seems a strange number. As I recall, the basic core was 2KiB, with an optional 2Kib available, for a maximum of 4. I don't remember the memory configuration exactly, but how does 5766 square with 2K or 4K, even if those figures are simply rounded off for convenience (much like 4096 is rounded to 4K)?

CampKohler (talk) 10:58, 16 November 2008 (UTC) CampKohler


 * UNIVAC encoded the addresses of a 1005 in a bizarre form, taking the bottom 5 bits of two charrcters as row and column selectors, with 31 valid values for each. The selectors were not in numerical order. Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:08, 3 September 2020 (UTC)

90/60, /70 and /80 Not Same Series as 9200/9300/9400 or 90/30 etc.
There are three distinct series here (in chronological order):

A. 9200/9300/9400/9480 : Univac's direct competition to IBM's early 360s: same instruction set, I/O model, etc. -- ran OS/4

B. 90/60, 90/70 and later 90/80 : Someone else's (it might have been GE's computer division) direct competition to IBM's 370s (i.e. with virtual memory) which Univac bought out and relabeled 90/60 etc. -- ran VS/9

C. 90/30, 90/40 (also /25 and /35 I seem to recall) : replacement line to the 9200/9300/9400 -- ran OS/3 (no virtual memory)

Because of its very different origins, the B) series above was actually very different from either A) or C). Only the Univac sales-people claimed that upgrading from a 9480 to a 90/60 was easier than to any other IBM/360 clone (especially to IBM itself).

For series A) and C), Telecommunications (i.e. CRT terminals) used TMS (Transaction Management System, later TMS/90) which was similar to IBM's CICS, running on top of what was essentially an OS extension called ICAM (Integrated Communications Access Method). Database management was hierarchical and used a program called DBS (Database Management System, later DBS/90)

Virtually all development I knew of was in COBOL/68, Asm/360 and RPG.

See also http://www.kyber.biz/rants/UNIVAC%20history-2.php

/An old Univac 9000-series systems programmer

AlatarK (talk) 21:45, 2 June 2010 (UTC)

Another interpretation of the acronym
A former employee once told me that there was no truth to the story that UNIVAC was an acronym for "Usually Not In a Very Active Condition". Old_Wombat (talk) 09:36, 19 October 2011 (UTC)

Page lacks references to minicomputer line
While mentioned on the Sperry Corporation page, this UNIVAC page lacks the sales of Varian Data Machines minicomputers. 100s if not 1000s were sold. Even the Varian page is poorly covered. While I didn't necessarily like them, nor the 1108s I ran on, the minis should be covered. 143.232.210.38 (talk) 20:01, 23 July 2012 (UTC)

Are they still selling UNIVAC products?
The intro seems to suggest they do —Ebaychatter0 (talk) 22:53, 4 September 2012 (UTC)


 * Any hardware being sold today would be under the Unisys brand, which is what UNIVAC became after merging with Burroughs in 1986. It's unlikely that any UNIVAC hardware is being sold any more today, as the merger happened a very long time ago even for government procurement cycles and support requirements, which is usually no more than 10 years. &mdash; QuicksilverT @ 17:28, 3 October 2012 (UTC)

Link to the TV commercial
Links don't work the way this one was added to the article. I think this is the intended video. Bubba73 You talkin' to me? 05:17, 9 January 2015 (UTC)

Univac 1004 Emulator
Article says the 1004 could have a plugboard wired (called the Emulator) to load instructions from cards into memory and presumably run them from memory. Some more details or sources on that would be fascinating. - Rod57 (talk) 01:17, 15 January 2016 (UTC)

First general-purpose computer for commercial use?
This article states the following:

"The BINAC, built by the Eckert–Mauchly Computer Corporation, was the first general-purpose computer for commercial use."

If one follows the link in that sentence to the BINAC article it states the following:

"BINAC was their first product, the first stored-program computer in the United States; the BINAC is also sometimes wrongly claimed to be the world's first commercial digital computer [3][4] even though it was very limited in scope, never fully functional and always economically unviable."

Seems that these too articles contradict each other, so which is accurate and which should be corrected? — Preceding unsigned comment added by 74.250.143.134 (talk) 00:46, 25 January 2020 (UTC)

Confusion between 1005 and 1050
I recently noticed that the entry for 1005 had dates prior to the 1004, the predecessor to the 1005. Those dates looked plausible for the 1050, an earlier machine. I replaced the incorrect dates, but now wonder whether any of the 1005 narrative actually applies to the 1050. Could someone familiar with the military history of those machines check? Shmuel (Seymour J.) Metz Username:Chatul (talk) 17:19, 3 September 2020 (UTC)

Missing 36-bit machines
The article does not discuss the 36-bit UNIVAC 1101, 1102, 1103, 1103A or 1105, although there is a picture of the 1103. Shmuel (Seymour J.) Metz Username:Chatul (talk) 18:18, 15 December 2020 (UTC)

Dubious
The article mentions that the military and civilian Univac 1005 have different instruction sets etc. This seems dubious to me, I'm wondering if this ties into earlier apparent confusion between 1005 and 1050. Dóeltenga (talk) 12:30, 21 December 2023 (UTC)
 * Actually, bitsavers has manuals for both versions I only worked on the military (single address) version, but knew about the other. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 14:22, 21 December 2023 (UTC)

Organisation vs. Machine
Even though the page includes this at the top, it does feel that it was very much originally created as a page about EMCC and original UNIVAC (I). For example there is very little in the page about ERA even though it's machines were sold under UNIVAC brand post acquisition by Remington Rand and there's next to nothing about work done at Norwalk (where General Groves was based and VP of R&D). Ideally the sections around organisation should include at least some background on all three followed by what happened post merger into one Division in '55. I did add about Norris been appointed as first General Manager, there probably room to add in parts about internal RR politics which help drive Norris out (such as removal of manufacturing from his control etc.) Thoughts? --Dóeltenga (talk) 15:42, 27 June 2024 (UTC)


 * I'd be glad for you to clear all of that up. I don't know all of it myself.  Bubba73 You talkin' to me? 18:47, 27 June 2024 (UTC)