Forgot your password?
Cyhist Apr 24 1997 A
Document Actions
Date: Thu, 24 Apr 1997 17:15:24 -0700
Reply-To: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@MAELSTROM.STJOHNS.EDU> Sender: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@MAELSTROM.STJOHNS.EDU> From: Mike O'Brien <obrien@ANTARES.AERO.ORG>
Subject: Reconstructing Colossus
In-Reply-To: Your message of "Tue, 15 Apr 1997 10:14:12 PDT."
<97Apr15.105427pdt.111228-3@aero.org>
______________________________________________________________________
Community Memory: Discussion List on the History of Cyberspace ______________________________________________________________________
I just got back from a trip to England, where I visited Bletchley Park. This is the installation where the German Enigma and Lorentz codes were broken during WW II.
I'm sure others are more up on the exact doings at Bletchley than I am, but since no one here has reported on the current effort, I thought I should say a few words.
Tony Sayle, director of the Bletchley museum of cryptography, has reconstructed part of Colossus, which many believe to be in line for the laurels of "first electronic digital computer." Tony was kind enough to give me an extensive tour of Colossus.
Colossus Mark II was constructed in 1945 to run correlation studies on wheel settings for the German Lorentz cypher machine. Most of the parts in Colossus are British Telecom phone switch parts, except for the 1700 or so vacuum tubes ("valves" to Colossus' British builders). The original instances of Colossus having been destroyed after the war, Sayle worked from photographs to reconstruct the main frame (and it is a frame) of the machine.
The front end of the machine is a 5,000 character-per-second photoreader. This part of the machine had to be redesigned from scratch, so Sayle found the original designer of the photoreader, now 80+ years old, and sat with him in front of a CAD/CAM system, where they redesigned the reader to the original specifications.
The paper tape that is Colossus' input is an endless loop tape of the message to be decrypted, rendered in five-level code. The sixth row of smaller sprocket holes is also detected, and these serve as the clock for the machine. Using this clock signal, decade counters in Colossus' back layer continually generate a sequence of possible wheel settings. These are combined with the data from the tape in Colossus' arithmetic unit, which includes a five-character buffer memory, to see if these wheel settings could have resulted in the encrypted text being read. After all correlations have been completed, the correlation statistics are printed out on a regular manual typewriter which has been fitted with relays on the number and carriage keys, indicating for each wheel setting, the degree of correlation to the message.
The keys with the highest correlation are then plugged in (literally, on a manual plugboard) on a machine called Tunney (which Sayle has also reconstructed), which is a logical equivalent of the Lorentz cypher machine. These keys are then tried against the encrypted message and the clear text read out.
Colossus is a highly parallel machine, working on each of the five tracks simultaneously. Two of the five parallel segments of Colossus have currently been reconstructed, with three more to go. The two-fifths of Colossus currently in existence work perfectly.
For those used to current logic levels: "1" in Colossus terms is +100 volts, "0" is -100 volts. This is not arithmetic to be lightly trifled with by the fumble-fingered. Colossus' logic runs off a 400-volt supply.
As with its original, Colossus is rarely if ever turned all the way off. The secret to sufficient reliability from over a thousand tubes is never to turn off the heater elements.
Colossus has never been nearly as widely discussed as ENIAC in the literature, mostly because most of its details remained classified until 1980. As it is, Tony Sayle had to get special permission for his own reconstruction project. Luckily, there are still lots of old British Telecom phone switches out there to use for parts scrounging, but he could use all the tubes (of the right kind) that anyone can find.
Neither ENIAC nor Colossus were stored-program machines, and I'm not competent to judge the merits of which one deserves the laurels of "First electronic digital computer." However, having wandered around the guts of Colossus, and seen it run with my own eyes, I admit to being extremely impressed.
It should be noted that Colossus runs faster than its own emulator, written in QBASIC on a Pentium 100. I have no idea what that proves.
Anyone in, or headed for, England should make it a point to go out to Bletchley Park and see Colossus for themselves. It certainly makes computer history come alive.
Bletchley Park has a web page at
http://www.cranfield.ac.uk/ccc/bpark. Note that it's only open every other weekend.
Mike O'Brien
______________________________________________________________________
Reply-To: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@MAELSTROM.STJOHNS.EDU> Sender: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@MAELSTROM.STJOHNS.EDU> From: Mike O'Brien <obrien@ANTARES.AERO.ORG>
Subject: Reconstructing Colossus
In-Reply-To: Your message of "Tue, 15 Apr 1997 10:14:12 PDT."
<97Apr15.105427pdt.111228-3@aero.org>
______________________________________________________________________
Community Memory: Discussion List on the History of Cyberspace ______________________________________________________________________
I just got back from a trip to England, where I visited Bletchley Park. This is the installation where the German Enigma and Lorentz codes were broken during WW II.
I'm sure others are more up on the exact doings at Bletchley than I am, but since no one here has reported on the current effort, I thought I should say a few words.
Tony Sayle, director of the Bletchley museum of cryptography, has reconstructed part of Colossus, which many believe to be in line for the laurels of "first electronic digital computer." Tony was kind enough to give me an extensive tour of Colossus.
Colossus Mark II was constructed in 1945 to run correlation studies on wheel settings for the German Lorentz cypher machine. Most of the parts in Colossus are British Telecom phone switch parts, except for the 1700 or so vacuum tubes ("valves" to Colossus' British builders). The original instances of Colossus having been destroyed after the war, Sayle worked from photographs to reconstruct the main frame (and it is a frame) of the machine.
The front end of the machine is a 5,000 character-per-second photoreader. This part of the machine had to be redesigned from scratch, so Sayle found the original designer of the photoreader, now 80+ years old, and sat with him in front of a CAD/CAM system, where they redesigned the reader to the original specifications.
The paper tape that is Colossus' input is an endless loop tape of the message to be decrypted, rendered in five-level code. The sixth row of smaller sprocket holes is also detected, and these serve as the clock for the machine. Using this clock signal, decade counters in Colossus' back layer continually generate a sequence of possible wheel settings. These are combined with the data from the tape in Colossus' arithmetic unit, which includes a five-character buffer memory, to see if these wheel settings could have resulted in the encrypted text being read. After all correlations have been completed, the correlation statistics are printed out on a regular manual typewriter which has been fitted with relays on the number and carriage keys, indicating for each wheel setting, the degree of correlation to the message.
The keys with the highest correlation are then plugged in (literally, on a manual plugboard) on a machine called Tunney (which Sayle has also reconstructed), which is a logical equivalent of the Lorentz cypher machine. These keys are then tried against the encrypted message and the clear text read out.
Colossus is a highly parallel machine, working on each of the five tracks simultaneously. Two of the five parallel segments of Colossus have currently been reconstructed, with three more to go. The two-fifths of Colossus currently in existence work perfectly.
For those used to current logic levels: "1" in Colossus terms is +100 volts, "0" is -100 volts. This is not arithmetic to be lightly trifled with by the fumble-fingered. Colossus' logic runs off a 400-volt supply.
As with its original, Colossus is rarely if ever turned all the way off. The secret to sufficient reliability from over a thousand tubes is never to turn off the heater elements.
Colossus has never been nearly as widely discussed as ENIAC in the literature, mostly because most of its details remained classified until 1980. As it is, Tony Sayle had to get special permission for his own reconstruction project. Luckily, there are still lots of old British Telecom phone switches out there to use for parts scrounging, but he could use all the tubes (of the right kind) that anyone can find.
Neither ENIAC nor Colossus were stored-program machines, and I'm not competent to judge the merits of which one deserves the laurels of "First electronic digital computer." However, having wandered around the guts of Colossus, and seen it run with my own eyes, I admit to being extremely impressed.
It should be noted that Colossus runs faster than its own emulator, written in QBASIC on a Pentium 100. I have no idea what that proves.
Anyone in, or headed for, England should make it a point to go out to Bletchley Park and see Colossus for themselves. It certainly makes computer history come alive.
Bletchley Park has a web page at
http://www.cranfield.ac.uk/ccc/bpark. Note that it's only open every other weekend.
Mike O'Brien
______________________________________________________________________
