beamjockey: Drawing of Bill of the Heterodyne Boys by Phil Foglio. (Default)
[personal profile] beamjockey
Once upon a time, the physicists who were attempting to create an atomic bomb were trying to solve some very hard mathematical problems.

So they sent a genius out to shop for a computer.

In the files of Los Alamos National Laboratory, I've learned, is a 1944 memo from the great mathematician John von Neumann to J. Robert Oppenheimer, scientific director of the Manhattan Project, reviewing the features, performance, and price of the machine. (It's a PDF.)

This may not be the first computer review ever. But it's certainly the earliest one I've ever read. (It was classified, of course.)

The device in question was the Bell Labs Relay Computer, also known as the Complex Number Calculator George Stibitz and Samuel Williams. Operating on numbers of 7 significant digits, it could add, subtract, multiply, divide, or extract square roots. A multiplication took 800 milliseconds.

Von Neumann writes:
An instruction on the control-tape therefore looks like this: "Take the contents of register a, also the contents of register b, add (or subtract, or multiply, etc.) and put the result into register c." At the same time it must be specified, whether the content of a (or b) must be held or cleared after this step. If it is to be cleared, then c may coincide with a (or b).

Stibitz states that a cube rooting device could be added with relative ease.

Von Neumann was involved with multiple computer projectsin the 1940s. With colleagues in Philadelphia, he came up with the "von Neumann architecture" used in the machine on which you are reading this. Had he lived in another age, he might have written hardware reviews for a computer magazine...

Date: 2008-04-05 04:07 pm (UTC)
From: [identity profile] mmcirvin.livejournal.com
Interesting that it's coarsely parallel, and that the review talks about some of the same engineering choices with regard to instruction sets that come up today. 15 float registers is not bad, really.

I wonder what he means when he says all data on the multiplier are for domestic use only. (Maybe that, in addition to the classification requirements, he was under a commercial NDA?)

Date: 2008-04-06 02:24 am (UTC)
From: [identity profile] mmcirvin.livejournal.com
...Though, of course, that's 15 float registers with no RAM. Though the handwritten note about the "punching relays" suggests that it could at least programmatically punch output cards that could be fed back into the input later.

Date: 2008-04-06 01:51 pm (UTC)
From: [identity profile] serge-lj.livejournal.com
I always liked the memo displayed in the Los Alamos museum, in which Oppenheimer requisitionned a nail. To hang his hat.

Profile

beamjockey: Drawing of Bill of the Heterodyne Boys by Phil Foglio. (Default)
beamjockey

May 2024

S M T W T F S
   1234
56789 1011
12131415161718
19202122232425
262728293031 

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Apr. 2nd, 2026 06:20 am
Powered by Dreamwidth Studios