I suspect your enthusiasm for the exploits of the USA in WW2 may be a little coloured by your nationality Myke :-). (Of course, as Carl Sagan was occasionally wont to say, I could be wrong). {I assume that you are a genuine US of A ite}. I have read a number of articles on the Bletchley Park efforts including a recent one which appeared in several magazines by someone who claimed to have been closely involved in some capacity with the original code breaking effort. If I find the article I'll send you the reference. As I recall the writer was, personally, the person who had written code for a Pentium to try to equal the speed of the original equipment, and had failed. If we presume (dangerously :-)) that he was competent at coding and had used a decent compiler, the explanation is probably based on the parallel and specialised nature of the task. If this is the reason then there is little doubt that a solution based on programmable array logic custom tailored to the problem, would easily beat the original equipment. I think that the original used thyratrons clocked at under 100 KHz so it would have to have had some special advantage (like massive parallelism) to compete with a Pentium. Certainly, the achievement was written up very sensationally in some magazines - along the lines of "Secret WW2 equipment had processing power in advance of modern pentiums" etc but the actual claim was not stated in urban folklore terms. The writer certainly didn't seem to think that the original decoding effort was US based but I'm sure he also could be wrong :-). There were 2 different decoding efforts, the enigma machine used by many German troops and the ??? (name escapes me) which was meant to be far more complex and secure and only used by the "high command" for critical uses. Even when "cracked" the latter still required vast quantities of deciphering effort to produce plain text output. The decoding task -----Original Message----- From: myke predko To: PICLIST@MITVMA.MIT.EDU Date: Tuesday, 10 February 1998 07:47 Subject: Re: Mark1 programming challenge >I'm just thinking, maybe an appropriate entry for this > >Keith Howell wrote: > >>The British built a computer to crack the messages encrypted >>by the German Enigma machine. I heard that this can be done >>by a Pentium PC of today, but nowhere near as fast as the >>electro-mechanical original machine. > >I don't know where you heard this about the decoding computers produced in >the early forties being very fast, but I think that's some kind of (luddite) >urban legend. > >Actually, it was the Americans who built the best "engines" during the war >and the ones sent to help the British were based on the "Magic" decrypters >used to defeat the Japanese Diplomatic/Military codes. > >The American engines were based on automatic telephone switching circuits. >The circuits were originally designed to count pulses and select circuits >based on the number of pulses. The American devices had the additional >advantage of being very reliable and using standard equipment (for fast >replication, easy repair and lower costs). > >>This demonstrates that custom hardware can beat generic >>problem solvers. The I thought, what about building an >>equivalent to the original in electronic logic? >>If it were implemented in programmable logic chips, >>would this be potentially faster? > >I agree with that comment, but not in this case. > >The engines used by the British, ran at about 10 checks (of a set of wheel >combinations) per second (the American's machines ran about about 50 cycles >per second). To check each "wheel" combination (the Enigna machines used 3 >wheels (out of about six), each wired differently and placed in different >locations in the machine (ie disks 1-2-3 would provide a different code from >disks 2-1-3). > >The total number of combinations/permutations was about 300 Billion. This >means, that to figure out a code, it would take one (britsh) machine about >six months to a year to crack a code. The machines were paralleled (of >course), but the issue for the British was their machines were basically >hand built and there was a lot of resistance to using the American built >machines which were mass produced using commercial components. > >Each wheel is wired uniquely, but each letter switch (ie "A" is output as >"M") also has the opposite wired true ("M" is output as "A"), so after >coding a message, to decode it, all you had to do was pass the data through >the wheels in reverse. After a character was coded/decoded, the wheels were >incremented (like an odometer on a car) to a new position to further >scramble the codes. > >I've written programs (years ago when I first start reading about this) with >an 8088 PC (running GW-BASIC) that could decode a three wheel Enigma code in >about an hour (assuming it understood significant parts of the message). I >presume a 200+ MHz Pentium Machine could do it in less than a minute >(although I haven't tried it, now I am intrigued about going back to them >(Now, if I can find the diskette they're saved on)). > >The decrypting of the code was carried out by knowing part of the message. >The Germans changed the codes (wheels and starting positions) from once a >month to once a quarter. Knowing part of the message was either by >monitoring specific stations with specific call signs (which were three >letters) or capturing unencrypted weather reports (which were the only way >they could do it for the U-Boat fleets). This series of letters was put >into the engine before it was run, stopping each time these letters were >found. > >When the letters were found, a human would go and look at what the machine >kicked out and see if a man-readable message was produced. So, with this, >obviously a lot of time was required looking at potential messages to see if >they were valid. The goal was to have at least fifty machines working on >the codes at the same time, so they could be cracked in something >approaching real time (ie within weeks of when they were changed) - at the >start of the war, typically the best they could do was about ten machines at >a time. > >>BTW, is there any truth in the rumour that while the >>government were sending spies to find out about >>the Enigma machine, they could have done so far more >>safely by asking the London patent office? >>machine > >No. While the mechanics were well known (and used as the basis for the >decryption engines), it was the wheels that had to be procured and the >German Navel machines which used four wheels and were a complete mystery >until later in the war when a sub was captured and boarded before the radio >officer had a chance to destroy his enigma machine (the Navy also used >different wheels that the British didn't have access to). > >The original Enigma machine was brought to the British by workers from the >plant in Poland where they were being manufactured. Even though the British >knew what the machines looked like, they didn't begin to try and figure out >the codes until they received a sample machine and wheels. > >Actually, it was the Americans that figured out the Japanese codes and >hardware from scratch. Apparently, none of the cyptographers saw a working >one except for a crushed parts of one that was found buried in the ground of >the old Japanese embassey in Washington after the war. > > >It's a really fascinating part of the history of the Second World War and >there are a ton of books written about it (including detailing how the >hardware works). Robert Harris' novel "Enigma" is not a bad introduction as >to how the codes were figured out. > >If you want to look into it, look for "The Ultra Secret" at Amazon.com for a >list of books (although many of the really good books are out of print) - >"Enigma" just brings in too many incorrect references. "Ultra Secret" was >the code name for the program to decode the "Enigma" box. > >myke > >Opus: There's a 465 pound woman pruning her azelias while wearing a pink >stretch bodysuit. > >911 Operator: So what's the emergency? > >Opus: From a taste perspective, it's a crisis of biblical proportions! >