From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Computing / Early (Rated Start-class, High-importance)
WikiProject icon This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Start-Class article Start  This article has been rated as Start-Class on the project's quality scale.
 High  This article has been rated as High-importance on the project's importance scale.
Taskforce icon
This article is supported by Early computers task force (marked as Top-importance).

Section sequence[edit]

My rationale for placing the History sec above the Tech description sec was that I think the majority of readers will be primarily interested in the UNIVAC's practical impact and installation history (which institutions and companies bought those machines, what did they use them for). Those of us who are equally/more interested in the tech stuff are probably going to read the History sec anyway, and then dive into the juicy details. :-) --Wernher 05:36, 30 March 2006 (UTC)

Misc UNIVAC 1 info[edit]

  • The Tektronix oscilloscope wasn't part of the computer and wasn't permanently attached to it. It was a standalone maintenance tool.
  • The Unityper was entirely mechanical. It wrote magnetic tapes by pushing permanent magnets against the tape and then advancing the tape.
  • The CPU had full redundancy and checking.
  • Case Institute of Technology had their UNIVAC I running until 1965.
  • The UNISERVO tape drives had motors powered by McIntosh audio amplifiers.

--John Nagle (talk) 04:05, 29 July 2008 (UTC)

A Univac I was installed at the Carborundum Company in Niagara Falls, NY. about 1956. In addition to the High Speed Sheppard printer which printed a 130 character line at 600 lines per minute the installation included an off line paper tape to magnetic tape converter. The paper tape was read at a rate of 400 cps by a Feranti photo sensor. The processor was later modified to allow three instructions to be included per each word. I can't give any citation for the above other than I served as installation manager in 1956. We worked closely with Dr. Grace Hoppers group who aided us with the systems programming. A set of program(s) written in B-0 (later to be known as COBOL) was put into use in 1957 or so. I have a fairly extensive write up on that installatin that could be of interest. We used Frieden Flexowriters to create orders at remote sales offices. The business systems concepts were generated by two achademics from Warton and were refered to as "The Total Systems Concept". (talk) 22:36, 22 October 2008 (UTC) Ronald F. Denz

Machine used to predict 1952 election[edit]

The article lead says: The fifth machine (built for the U.S. Atomic Energy Commission) was used by CBS to predict the result of the 1952 presidential election. However, according to this article, "The Univac in Philadelphia was connected to a teletype machine at the CBS studios in New York City. As the first precincts reported on election night, technicians used Unityper machines to encode the data onto paper tape to feed into Univac.", and the article indicates that the fifth machine was delivered in 1953 to the Atomic Energy Commission in Livermore, CA. Was Eckert–Mauchly testing the fifth machine in Philadelphia in Nov. 1952, prior to its 1953 delivery to California? Clarification in the article would be helpful. (I think the wired article has an error. Unityper machines encoded data onto magnetic tape, not paper tape) --Wbm1058 (talk) 23:53, 9 February 2012 (UTC)

Additional Information from Original Texts[edit]

While the description here states that if a non-digit was encountered during an arithmetic operation, the machine would pass it to the output unchanged. However according to the Univac Programming Manual, 1st Revision 1953, it states that during add/subtract, if a non-numeral is encountered as data, the machine will stop and light a neon on the SC to indicate the type of fault.

Other tidbits- while reading from tape can be performed both forward and reverse, writing can only take place during forward motion. And the programmer must know which direction he wishes to read from tape as the instructions are specific. The UNIVAC could write at 20 char/inch for uniprinter intended output. Or at 100 char/inch for use by the machine proper at a later time. (The programmer must specify via instruction). Output was buffered in 60 word blocks. This information from my ’53 Programming Manual.

The wiki page states recording density of 128 bits/inch but according to the 1954 Manual of Operations (Remington Rand EL-210 Copyright 1954), density is actually 128 DIGITS per inch. A digit is 4 bits ‘numeric’, 2 bits ‘zone’ and 1 bit ‘parity’.

A couple of other things I don’t see mentioned but find impressive, especially when compared against later machines: It could read in data from any one of 10 Uniservos, write data to 2 different Uniservos, and continue to perform computations all simultaneously.

With regards to error checking- the 4 primary arithmetic registers were all redundant and results were continually compared. Odd parity checks were also used to verify input and the memory tanks were scanned for parity as the serial words recirculated. If a fault was encountered, a neon specific to that "area" would light.

In regards to the remark about the Unityper being entirely mechanical, I’m curious how it handled clearing of erroneous bits from tape. I know the operator could “backspace” and write over a mis-typed character. Not sure how it was done. — Preceding unsigned comment added by (talk) 15:38, 16 July 2012 (UTC)

Someone who has the time should work in the information from the Univac patents. Notably:

  • US Patent 3133190 for a Universal Automatic Computer Utilizing Binary Coded Alphanumeric Characters, issued to Eckert, Weiner, Shaw, and Welsh, assigned to Sperry Rand, Filed 1952, Issued 1964.
  • US Patent 3784983 for an Information Handling System (followup to the above).
  • US Patent 2708554 for a Tape Drive and Recording Apparatus, issued to Welsh, Schrolner, Mock, Eckert, assigned to Remington Rand, Filed 1950, Issued 1955.
  • US Patent 2901734 for a Tape Drive and Recording Apparatus (followup to the above).
  • US Patent 3189290 for a Tape Drive and Recording Apparatus

(yet another followup). These patents are loaded with buzzwords like Uniservo and circuit diagrams full of vacuum tubes. The illustrations include some wonderful perspective views of computer rooms. Douglas W. Jones (talk) 21:44, 28 August 2012 (UTC).

Are you sure that the BINAC was really a commercial computer?[edit]

There is no citation in the first footnote for the assertion that the BINAC was actually a "commercial" computer.

As far as I can tell, the single BINAC sold never did much of anything for the customer, Northrop, so while its sale could be considered a commercial transaction, I question whether the product delivered was actually a functional computer at all. In contrast, clearly the Univac I was a functional computer line. — Preceding unsigned comment added by (talk) 17:04, 16 January 2015 (UTC)

Why not description of the memory in modern terms?[edit]

The section on memory has a lot of technical gobbledygook, but it doesn't translate that into a statement such as, "It had 1 megabyte of memory." Therefore, it is impossible to compare the UNIVAC memory to the memory of modern computers. — Preceding unsigned comment added by (talk) 22:43, 17 January 2016 (UTC)

The "technical gobbleygook" ... I know I'm just an IP address anon but ... I'm only commenting because your phrasing did cause rage in my hardware heart. Okay... the simple reason, why aren't UNIVAC specs expressed in "modern" units: UNIVAC predates the power of 2 convention that modern units of memory are based on. Note UNIVAC uses 11-bit chunks, unlike today's machines that took on a base of 8-bits to a byte, later 16, then 32, 64 etc. Also it's really incredibly hard to express how... mechanical types of memory work to someone expecting the nice discreet electronic logics of today. There is a HUGE reason that core memory was a Big Deal (tm), and an advancement from things like delay line (is that right? the mechanical coil) or mercury delay memory (both those, were brilliant but core brought memory to an electronic level) at a time before... any concept of a unit of memory compatible with today's "bits and bytes." (talk) 08:18, 27 January 2016 (UTC)