1 mhz, 1k ram ... let's go to the moon!

Chmee · Post by **Chmee** » 2005-02-10 06:53pm

February 10, 2005
In Computer Years, Apollo Replica's an Antique
By CYRUS FARIVAR

SOME people climb mountains to achieve greatness. Some people try to win championship sports games. John Pultorak built a working replica of a 40-year-old computer.

Late last year, Mr. Pultorak, of Highlands Ranch, Colo., completed a four-year project, a reconstruction of the Apollo Guidance Computer.

The A.G.C. was onboard for many of the Apollo space missions from 1969 to 1972. It was the computer that Neil Armstrong, Buzz Aldrin and Michael Collins used in the command and lunar modules during the Apollo 11 mission, which landed on the moon on July 20, 1969.

"I was looking for something that would be a challenging thing, a really outrageous kind of project," said Mr. Pultorak, 51, a software engineer for the space systems group at Lockheed Martin. He compared his quest to that of a person who aspires to climb a mountain "just to see if they can achieve it."

"I wanted to build something that was really distinctive - the computerized version of Mount Everest or the Super Bowl," he said.

The A.G.C. is a piece of computing history. It had a one-megahertz processor, one kilobyte of random-access memory and 12 kilobytes of read-only memory. By contrast, typical desktop computers today have about 1,000 times the processor speed and about 500,000 times the RAM, and have dropped ROM for hard drives with millions of times the capacity.

Mr. Pultorak spent close to $3,000 and worked about 10 hours a week for four years, mostly in the evenings and on weekends. When it was finished, in October, he spent two months detailing his travails in documents that totaled over 1,000 pages, and he made them available at his Web site, starfish.osfn.org/AGCreplica/.

Mr. Pultorak said he had to make one small compromise. The microchips in the original A.G.C. are no longer available, so he had to use something slightly more modern: chips from the late 1960's. Otherwise, the computer is the same.

"I couldn't do the project at all with the original chips," he said in a telephone interview. "In my documentation I have a whole section about how my machine differs from the original. It has the same architecture, it has the same control signals, it has the same micro-instructions - hey, this is an Apollo Guidance Computer."

When he started, Mr. Pultorak had to dig around for schematics, documentation and any other relevant material he could find. In November 2001, nearly a year after he began, he found that the Massachusetts Institute of Technology had a repository of information on the project (hrst.mit.edu/hrs/apollo/public). M.I.T. helped design the A.G.C. nearly 40 years ago.

From there he was able to find parts of the source code - instructions written in a programming language the computer can understand - for the main piece of software that ran on the A.G.C. in the command module of many of the Apollo missions. The program was called Colossus. While working on the A.G.C. source code, Mr. Pultorak found references to Margaret Hamilton, one of the hundreds of programmers who worked on the software. "It's been a couple of years since I've been in the A.G.C. source code, but I remember laughing at those references," he said. "There would be something Margaret had done - you feel like you're working with them, you feel touched by it, you feel like a member of the team in some way."

When he finished the project, Mr. Pultorak trudged up his basement stairs and informed his wife, Sue, matter-of-factly.

"He's a very modest, low-key guy," Sue Pultorak said. "He just came up and said, 'I guess I finished it.' We all went down and watched it blink. It was awesome."

Now that the construction is complete, Mr. Pultorak has pondered what to do with the 70 pounds of electronics crammed into a three-by-five-foot flat box. He has had an offer from somebody who wanted to buy it, but he cannot bring himself to part with the A.G.C.

His 20-year-old son, Andrew, a senior at the University of Colorado, Denver, helped with the project over the summer. He, too, said he did not want to see the computer sold.

"No matter what he got for it, it wouldn't be worth the kind of time and energy that he put into it," he said.

But John Pultorak said that once the building was over, he wanted to free up space in the house. He has moved the replica from his basement workshop to his mother's house in Littleton, Colo., a Denver suburb nearby.

"I built it in my basement, but I don't want to have it - I just wanted to build it," he said. "It's like climbing a mountain. You want to climb the mountain, you don't want to live on top of it."

Link

phongn · Post by **phongn** » 2005-02-10 06:56pm

The AGC didn't need that much computing power, really. You can do a lot on a little bit of hardware.

Jalinth · Post by **Jalinth** » 2005-02-10 08:17pm

Destructionator XIII wrote:Wow, and I thought my P1 was slow....
I would love to have one of those old ones, just to say I do.

All that this computer shows is that the interfacts sucks up lots of power. All Hail to Windows.

But the moon shots were fairly basic physics problems in terms of the math. So you don't need huge amounts of computing power to figure them out.

The engineering side seemed much more difficult

Praxis · Post by **Praxis** » 2005-02-11 11:19am

Our computers are now 3000 times faster, and run software that is 5000 times slower!

SPOOFE · Post by **SPOOFE** » 2005-02-11 05:13pm

Our computers are now 3000 times faster, and run software that is 5000 times slower!

I don't even see how you can compare them. Because Windows can't natively calculate the trajectory of an artillery shell, does that mean modern computers are completely outclassed by room-sized, vacuum-tubed computers of the '50s?

Chris OFarrell · Post by **Chris OFarrell** » 2005-02-12 04:53am

And I'm sure with the right program windows would be quite happily calculate the trajectory of an artilery shell.

Master of Ossus · Post by **Master of Ossus** » 2005-02-12 01:40pm

I'm actually surprised the processor was that good, compared to the memory and the RAM.

Praxis · Post by **Praxis** » 2005-02-12 02:16pm

SPOOFE wrote:
Our computers are now 3000 times faster, and run software that is 5000 times slower!
I don't even see how you can compare them. Because Windows can't natively calculate the trajectory of an artillery shell, does that mean modern computers are completely outclassed by room-sized, vacuum-tubed computers of the '50s?

I'm just being sarcastic. I don't actually think that a 1 MHz computer is faster :p