Work on which famous unsolved mathematical problem revealed that Intel had a problem of their own?

peasypod

Would it be a case of those pesky 'floating point numbers'? Jun 29 05, 5:54 PM

lanfranco

I believe that when Andrew Wiles was working on Fermat's Last Theorem (The Maven and I were at Princeton at the time), he ran into a problem with Intel's pentium chips. They were not capable of completing certain operations, as I recall. Cant find the specific reference, but here's a site in the meantime: http://en.wikipedia.org/wiki/Andrew_Wiles Jun 29 05, 5:59 PM

gmackematix

Nice try Frankie but while it was Intel's Pentium chip that was found to be flawed, it wasn't FLT that was being worked on at the time. And it was the floating point bug so, spell it out then Peasy? Jun 29 05, 6:11 PM

gmackematix

And do it nicely (like such a hint is needed). Jun 29 05, 6:12 PM

peasypod

I'm a bit lazy this morning, tax time and I've got too many illicit goings-on to fiddle with.... Here's my easy way out: http://www.willamette.edu/~mjaneba/pentprob.html Jun 29 05, 6:14 PM

gmackematix

A shame your easy way out doesn't actually mention which problem was being worked on but it should certainly be easier to find now. Care for another stab, Frankie, while Peasy's trying not to overcook the books? Jun 29 05, 6:28 PM

peasypod

824633702418 <= x <= 824633702449 is the famous email sent by Dr. Thomas R. Nicely. http://members.cox.net/srice1/pentbug/pentbug7.html Jun 29 05, 6:29 PM

peasypod

...and yes I'm aware.....sigh. Jun 29 05, 6:30 PM

peasypod

Sheeeesh. Tell me it's the Riemann Hypothesis.. Jun 29 05, 6:37 PM

lanfranco

All I can say is that Wiles' first draft of the proof of the FLT had the bug -- and that would have appeared in 1992/93: http://www.cc.gatech.edu/classes/AY2005/cs3220_spring/pentiumbug.pdf Jun 29 05, 6:39 PM

gmackematix

Not the Riemann Hypothesis either I'm afraid, Peasy. Your last post almost gives it away. I can't link that site Frankie, probably due to my lousy browser, but the Pentium bug didn't surface until 1994, a year after Wiles's proof. Jun 29 05, 7:43 PM

lanfranco

Well, you're right, the link isn't working. However, it does say that Wiles had the pentium bug in the first public draft of his proof, which appeared in 1993. My guess is that problem simply wasn't fully indentified until '94. Wiles finally solved his difficulties in '95. Here's yet another site, which provides some date information, though nothing specifically on the pentium chip: http://www.answers.com/main/ntquery?method=4&dsid=2220&dekey=s10.3.9&gwp=8&curtab=2220_1 Jun 29 05, 8:03 PM

peasypod

Where was my brain? Oh yes, diddling the Queen, that's right... Twin Prime Conjecture. http://mathworld.wolfram.com/TwinPrimes.html Jun 29 05, 9:18 PM

gmackematix

You got there Peasy so a well deserved yay! (I'll leave the rolls to your discretion) If Jeff Bridges can explain the Twin Prime Conjecture so that Barbara Streisand understands it (OK so it was only in a film), I should be able to do it here. It just states that there exist an infinite number of primes that differ by two (like 11 and 13). Jun 30 05, 2:57 AM

peasypod

I was pre-occupied with a paradox of my own.....Those rolls? Keep em' coming. Jun 30 05, 4:33 AM