Happy 50th Birthday, UNIVAC 1

Frums writes: "Today is the 50th birthday of the UNIVAC I (UNIversal Automatic Computer), the first commercial computer. It was quite a beast: 16,000 lbs, 5000 vacuum tubes measuring 9 inches by 2 inches, and an amazing 1000 instructions executed per second! The first UNIVAC was sold to the US Census bureau where it revolutionized data storage from them. No longer did they have to use punch cards, UNIVAC supported storage on metal tape! The US Census bureau still maintains a plaque commemorating the computer. It reads "Bureau of the Census dedicated the world's first electronic general purpose data processing computer, UNIVAC I, on June 14, 1951. Eckert-Mauchly Computer Corporation" Happy Birthday, UNIVAC I!" Wired has a brief story about it.

Comments

Made out of depleted uranium?!?

[Just Some Guy]

...16,000 lbs, 5000 vacuum tubes measuring 9 inches by 2 inches...

I originally parsed that as: "16,000 lbs, 5000 vacuum tubes, and 9"x2" in size.

Holy crap! The world's heaviest palmtop!d

UNIVAC I experiences

[Animats]

Case Institute of Technology had a UNIVAC I running when I went to visit them before enrolling. But by the time I got into college, they'd upgraded to a Univac 1107. Al Misek, who had maintained the UNIVAC I, told me that in the years the machine had been at CASE, they'd never had a tube failure during operation. Every morning, the tubes were run on "high margin", with elevated voltages, which would burn out all tubes near failure. Those were then replaced, allowing a day of uninterrupted operation. The self-checking dual CPUs would catch any errors.

This was a decimal machine. Early computing used decimal machines for business, and binary machines for scientific work. There's still a residue of this in the decimal instructions of the x86 and of IBM mainframes.

As a kid, I came across a junked UNIVAC I, including console, at Alert Surplus Sales (920 W St. NW), in Washington, D.C. Got to poke around the insides a bit. The tape drive's reel motors were driven by standard McIntosh audio amplifiers. The console switches were all telephone lever switches. There's no display on the console other than lights.

Working at the Census Bureau in the late 1960s, I met many people who'd used the UNIVAC I machines. They also still had lots of punched-card tabulating gear, but prior to the UNIVAC I, they'd had acres of IBM tabulators. All the IBM gear was on rental; IBM didn't sell their machines. So, once the UNIVAC I was up and running, one day the IBM sales rep was called in and told that Census was cancelling most of the tab gear. It was the biggest return in IBM history, and the event that made T.J. Watson get IBM into computers.

Census still had two UNIVAC 1105 machines running; the biggest vacuum-tube machines ever sold commercially. They still had lots of UNIVAC I tape. The original UNISERVO I tape was 8 track (6 data, 1 parity, one clock), 50 BPI and steel. Not steel on plastic, the tape was a ribbon of steel. Plastic tape, and an upgrade to 200 BPI, came with the UNIVAC 1105 and the UNISERVO II. Bad spots had to be found manually, and a tape with a bad spot could be rewritten if you manually punched a hole in the tape on either side of the bad spot. I still have a reel of this stuff from my years at Case.

The UNIVAC I was operated as a tape-in, tape-out machine. Other standalone systems, each the size of a mainframe computer, did card-to-tape, tape-to-card, and tape-to-printer operations. The keyboard on the console had no display other than the console lights. Typically, UNIVAC I machines spent most of the day sorting, spinning tapes back and forth merging subsorts together. This was inefficient by modern standards, but far, far better than sorting hundreds of millions of punched cards. The sorting job alone justified the machines for Census.

The UNIVAC I was basically the first commercial computer good enough to routinely use for business data processing.

First Digital Computer?

[clary]

For another view of the early days of the computer, see this link: John Vincent Atanasoff and the Birth of the Digital Computer.

Disclaimer: Iowa State University is one of my alma maters, so I am naturally biased.

Re:...but not the first stored program computer

[UncleFluffy]

Well, to be accurate:

"The Manchester Machine (aka Manchester Mk 1)" (1949) was the *second* stored program computer, and was a general purpose machine.

"The Baby" (1947) (also from Manchester) was the *first* stored program computer, and was a general purpose machine.

"Colossus" (1943) was built at Bletchley Park, and was neither a stored program computer nor a general purpose machine.

"ENIAC" (1945) was built at the University of Pennsylvania and was (almost) a general purpose machine, but not a stored program computer.

"Ferranti Mark 1" (February 1951) was the world's *first* commercial computer.

"UNIVAC" (March 1951) was the world's *second* commercial computer.

(I'm not familiar enough with Zuse's contributions to place them accurately, but will acknowledge that they exist)

Nice PR move by Unisys

[hillct]

It really ammounts to a nice PR move by Unisys. Vary slick. Remind the world that 50 years ago the company was an inovator, well What have you done for me lately?

I was a little disappointed with their spokesman Mr. Esnouf:"My son, for example, plays this game called 'I'm Going In,'" Esnouf said. "He spends all Sunday morning shooting people on the computer. We've invented this whole virtual reality. It's great, isn't it?" Is that really the best light he could put computer gaming in? I'm all for computer games and I'd say 'spending sunday morning shooting people' is a bit harsh. But all in all, Unisys pulled off a vary nice PR move without having to produce announce, or unveil a new product. Good deal for them...

--CTH


---

My Uncle...

[stonewolf]

One of my Uncles wrote code some of the code for the UNIVAC I that was used to predict the out come of the 1952 presidential election. He would tell the story every time he was around his techie kids and relations. Seems most of the folks on the project were hard core Democrats and they all believed deep down inside that the other guy (Stevenson?) was going to win.

When they got the first numbers and ran the analysis the UNIVAC said that Eisenhower was going to win by a landslide. Well, the programmers didn't believe it so they started looking for the bugs in their code. They looked really hard and fixed a couple of bugs and they got new data and they reran the analysis and it said Eisenhower by a landslide. So they went looking for more bugs...

Finally they had to report their results and they did with great embarrassment because nobody, including the press, believed the Eisenhower could possibly win.

Eisenhower won in a landslide.

My Uncle would always end the story with a moral about having to trust the results of experiments even when they disagreed with your personal belief. He's a great guy, I wish I knew him better.

StoneWolf

Not the machine so much as the people

[Penfield Zoat]

We in the tech business sometimes have a habit of elevating technology to a sort of self-causing status. For example, we are now celebrating the birthday of UNIVAC, a non-living being which has done nothing of its own accord. I have to ask: where are the humans?

Neither the wired article nor Slashdot so much as mention those who made this, and all subsequent computers a reality. Sure Linus Torveldes buys a taco and it's written up in every Linux rag, but try to give some credit to the people who gave him his opprertunity, and you come up empty! (Never mind that these guys were doing original, really original work, and Linus was just copying).

I demand that Slashdot's editors actually bother to find out who created this machine and publish it. It think that as a computer user, you owe them that much of a memory. Not to mention it might put a stop to all this senseless gadget-worship.

UNIVAC 1 users manual on line

[Al Kossow]

In honor of UNIVAC 1's birthday, I've placed a scan of the users manual at http://www.spies.com/aek/pdf/univac/Univac1_OperMa n.pdf There is a nice picture of the front panel at the back. re: simulators I seriously doubt any UNIVAC 1 code survives to RUN on a simulator. It appears that there is very little second generation computer software left, much less first generation.