Delivered-To: luke@ndatech.com Date: Wed, 14 Nov 2001 12:57:09 -0800 From: mersenne-digest-invalid-reply-address@base.com (Mersenne Digest) To: mersenne-digest@base.com Subject: Mersenne Digest V1 #905 Reply-To: mersenne@base.com Sender: mersenne-digest-invalid-reply-address@base.com Mersenne Digest Wednesday, November 14 2001 Volume 01 : Number 905 ---------------------------------------------------------------------- Date: Mon, 12 Nov 2001 20:44:22 -0000 From: bjb@bbhvig.uklinux.net Subject: Re: Mersenne: 1st 6 P90CPU yrs jump! On 11 Nov 2001, at 20:12, Nathan Russell wrote: > >The output from the script is interesting, but it can be very > >misleading for those of us who have recently upgraded equipment. Yes, those who have recently upgraded equipment - or added more systems - will find their "hours per day" very far from static. I've been with GIMPS/PrimeNet for 3.5 years now, but for various reasons my hours/day has increased by 10% over the last two or three months. I think perhaps the hours/day reported should reflect recently submitted work - say over the last 3 months - rather than the historical average since the user registered with PrimeNet. > > Well, the way I see it the "P-90 Standard" itself is less than > perfect. It was computed on a machine that can no longer even be > purchased new in some stores, Hmm. It's a Long Time since I saw a P90 system in a store. My guess is you'd be lucky to get more than $5 for a P90 system unit at an auction on e.g. eBay. If you _really_ want a P90 system unit, you'd probably be best advised to try your local landfill. The point is NOT whether a particular system still exists, it's to define a "benchmark" which can be applied to _any_ system. Obviously you end up with an obsolete benchmark system - you can compare a "new" system with an existing one, but there's no way George could have used a 2GHz P4 as a benchmark system in 1995. > on a version of the program well before the present version. I thought the timings had been scaled (at least to v19); the current benchmark system is a PII-400 (?) On a personal note, I just passed 100 P90 CPU years accredited by PrimeNet - but I didn't get a telegram from the Queen, so presumably the benchmarks _are_ wrong! Regards Brian Beesley _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Mon, 12 Nov 2001 17:25:26 -0500 From: "Carleton Garrison" Subject: Re: Mersenne: Fw: The Mersenne Newsletter, issue #18 I wouldn't worry about it. It is a lot easier to catch up double-checks than it is to get ahead doing first times. Carleton Garrison - ----- Original Message ----- From: "Daran" To: Sent: Sunday, November 11, 2001 10:37 PM Subject: Mersenne: Fw: The Mersenne Newsletter, issue #18 > > * Over 70,000 double-checks completed. > > * Over 100,000 exponents tested for the first time! > > Does this mean that we're not doing enough double-checking? _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Mon, 12 Nov 2001 23:34:50 -0600 (CST) From: ribwoods@execpc.com Subject: Mersenne: Fw: The Mersenne Newsletter, issue #18 "Daran" wrote: > > * Over 70,000 double-checks completed. > > * Over 100,000 exponents tested for the first time! > > Does this mean that we're not doing enough double-checking? Many recent double-check assignments involve a second round of factoring because (a) the trial-factoring breakpoints are higher now than they were when the first L-L test was assigned, and/or (b) P-1 factoring had not yet been implemented in Prime95 when they were assigned for initial L-L tests. Consequent elimination of once-L-Led Mnumbers by second-round factoring would account for some of the difference, though I doubt there've been 30,000. In addition, some folks have concentrated on factoring Mnumbers that have been L-L tested but not yet DC'd, specifically in order to reduce the number of necessary double-checks. _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 13 Nov 2001 22:03:40 -0000 From: bjb@bbhvig.uklinux.net Subject: Re: Mersenne: Fw: The Mersenne Newsletter, issue #18 On 12 Nov 2001, at 23:34, ribwoods@execpc.com wrote: > Many recent double-check assignments involve a second round of > factoring because (a) the trial-factoring breakpoints are higher now > than they were when the first L-L test was assigned, and/or (b) P-1 > factoring had not yet been implemented in Prime95 when they were > assigned for initial L-L tests. Consequent elimination of once-L-Led > Mnumbers by second-round factoring would account for some of the > difference, though I doubt there've been 30,000. Nothing like 30,000! The probability of finding a factor by going one bit deeper (given that trial factoring is already completed to ~60 bits) is less than 0.02; the probability of finding a factor using P-1 with "double checking" limits is typically 0.025 (variable depending on how deeply trial factoring has been done). 70,000 double check assignments would be expected to find about 2,000 factors. > > In addition, some folks have concentrated on factoring Mnumbers that > have been L-L tested but not yet DC'd, specifically in order to reduce > the number of necessary double-checks. Eh? Doesn't it make more sense to concentrate on factoring Mnumbers that haven't yet been L-L tested? That way "success" in finding a factor reduces the number of LL tests, as well as (eventually) the number of double checks. Regards Brian Beesley _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Wed, 14 Nov 2001 10:09:53 +0100 From: Henk Stokhorst Subject: Mersenne: prime L.S., from the status.shtml page: ------- Mersenne Exponent Test State ------- Assigned in Tests Cleared Since Last Synchronization Factoring only : 8683 Factored composite : 13505 Lucas-Lehmer testing : 27314 Lucas-Lehmer composite: 50306 Double-checking LL : 8158 Double-checked LL : 34091 Prime, VERIFIED : 1 Prime, UNVERIFIED : 1 ---------------------- ------- ---------------------- ------- TOTAL : 44155 TOTAL : 97904 YotN, Henk _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Wed, 14 Nov 2001 14:32:28 -0000 From: bjb@bbhvig.uklinux.net Subject: Re: Mersenne: prime On 14 Nov 2001, at 10:09, Henk Stokhorst wrote: > Prime, VERIFIED : 1 > Prime, UNVERIFIED : 1 Yes, I saw this in the report produced at 07:00 GMT this morning. Certainly wasn't there yesterday am! Last week I joked "of course, we could find two tomorrow". I have no knowledge as to what has triggered this report but the "obvious" explanation - that two primes have been found on the same day, and one has already been verified - sounds implausible. Assuming that a single prime has been found, there could be a problem with the (automatic) PrimeNet status page generation: after all, this situation doesn't often arise... Case 1) A prime is reported; the user gets excited and reports it again using the manual testing page. Could the PrimeNet automatic report generator consider this to be a double-check, and therefore verified - but also keep the "unverified" record? Note, normally double-check verification is done be George using the PrimeNet server transaction logs when he updates the database - usually on a Saturday - George's method eliminates multiple notifications, which will have the same offset when one of the Prime95 family of programs is used, or the same program, user & computer ID otherwise. Case 2) Rather simpler - a prime is discovered during a double- check; the report generator "knows" there is no prior notification that the exponent yields a Mersenne prime (so it must be unverified) yet the result "must be a verification" because the assignment is a double-check. So it gets counted twice, once in each category. Case 3) No primes have been discovered, but someone has discovered how to insert false data into the server. As I said above, I have no knowledge as to which of these possible explanations (if any) is true. Yet congratulations are obviously in order to anyone who really has discovered a new Mersenne prime (or primes!) - we've waited long enough! Please remember that, if the new prime is in the EFF prize range, publicity pending verification could jeopardize a claim on the prize, so it is understandable that any information which might be released in the near future may be incomplete to the point of sketchiness, just as it was during June 1999. Regards Brian Beesley _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Wed, 14 Nov 2001 10:19:11 -0500 From: George Woltman Subject: Re: Mersenne: prime Hello all, At 10:09 AM 11/14/2001 +0100, Henk Stokhorst wrote: > ------- Mersenne Exponent Test State ------- > Prime, VERIFIED : 1 > Prime, UNVERIFIED : 1 > ---------------------- ------- ---------------------- ------- > TOTAL : 44155 TOTAL : 97904 It's too early for me to comment much. The line "Prime, VERIFIED : 1" is obviously a bug in the reporting system. There has been an unverified prime reported! It passes the 32-bit security code that comes on every results.txt line. This is not overly difficult to forge though. The user reporting the prime has completed 3 other LL tests and seems to have signed up with a valid looking email address. I'll start a verification run on this Mersenne number of over 3,500,000 digits. With fingers crossed, George _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Wed, 14 Nov 2001 12:24:24 -0500 From: George Woltman Subject: Mersenne: Mersenne prime #39 Hi all, I've exchanged email with the discoverer. It looks like we've found the 39th known Mersenne prime - not a hacking attempt. Well done everyone! My verification will complete November 24th. I'll see if I can't get Ernst Mayer to do the official "different program - different CPU architecture" verification. Having a good time, George _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Wed, 14 Nov 2001 12:51:05 -0500 From: Jeff Woods Subject: Re: Mersenne: prime At 10:19 AM 11/14/01 -0500, you wrote: >There has been an unverified prime reported! It passes the 32-bit security >code that comes on every results.txt line. This is not overly difficult >to forge >though. The user reporting the prime has completed 3 other LL tests and >seems to have signed up with a valid looking email address. > >I'll start a verification run on this Mersenne number of over 3,500,000 >digits. FYI, folks, this would make "p", the exponent/assignment, greater than 11,626,750. It will take a Pentium 4 of 1.8 Ghz about 5.5 days to double-check this result. Of course, if it isn't a P4, or isn't that fast, it won't be this fast. (i.e. if the testing machine was WinIntel, the double-check will likely be run on Linux using mprime, or some other not-identical system). Don't expect an answer this week, or even next (unless George can fill us in on the iron being used for the double-check). Patience is a virtue. Some infonuggets: There are only 2,718 people who have tested exactly 3 or 4 LL tests, though, so that narrows it down a *bit*. It wasn't erroneously listed in the cleared list this time (like it was for M37), but a REALLY enterprising person (with access to the assigned exponents list for November TWELFTH) could probably narrow it down to a handful of candidates... ;-) Does this "general area" of somewhat greater than 11.6MM tend to fit the "Mersenne Island" conjecture? It seems to me that it would... unless George is being intentionally secretive about the size (i.e. > 3,500,000 digits COULD be 6,000,000 digits -- I'm assuming it is of the close order of that range, though). _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Wed, 14 Nov 2001 13:47:18 -0500 From: Jud McCranie Subject: Re: Mersenne: prime > >>There has been an unverified prime reported! It passes the 32-bit security >>code that comes on every results.txt line. This is not overly difficult >>to forge >>though. The user reporting the prime has completed 3 other LL tests I think it is quite interesting if this is only his/her 4th LL test! Of course we all appreciate the people who are able to devote several CPUs to the job, but this is encouraging to the "little guys". It shows that every little bit helps! +---------------------------------------------------------+ | Jud McCranie | | | | Programming Achieved with Structure, Clarity, And Logic | +---------------------------------------------------------+ _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Wed, 14 Nov 2001 11:00:32 -0800 From: "Aaron Blosser" Subject: RE: Mersenne: prime Bummer... it doesn't show that anymore. :( George/Scott: can one of you verify what the situation is? Finding a new one, even unverified, is big news, but if it's a glitch, we'd want to know so we don't get our hopes up. :) Aaron > -----Original Message----- > From: mersenne-invalid-reply-address@base.com [mailto:mersenne-invalid- > reply-address@base.com] On Behalf Of Henk Stokhorst > Sent: Wednesday, November 14, 2001 1:10 AM > To: mersenne@base.com > Subject: Mersenne: prime > > L.S., > > from the status.shtml page: > > > ------- Mersenne Exponent Test State ------- > > Assigned in Tests Cleared Since Last Synchronization > > Factoring only : 8683 Factored composite : 13505 > Lucas-Lehmer testing : 27314 Lucas-Lehmer composite: 50306 > Double-checking LL : 8158 Double-checked LL : 34091 > Prime, VERIFIED : 1 > Prime, UNVERIFIED : 1 > ---------------------- ------- ---------------------- - ------- > TOTAL : 44155 TOTAL : 97904 > > YotN, > > Henk > > > ________________________________________________________________________ _ > Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm > Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Wed, 14 Nov 2001 15:54:00 -0500 From: George Woltman Subject: RE: Mersenne: prime Hi Aaron, At 11:00 AM 11/14/2001 -0800, Aaron Blosser wrote: >Bummer... it doesn't show that anymore. :( > >George/Scott: can one of you verify what the situation is? Finding a >new one, even unverified, is big news, but if it's a glitch, we'd want >to know so we don't get our hopes up. :) Unfortunately, we don't get to test out the "a prime has been reported" code very often :) The primenet server is supposed to email me, Chris Caldwell, Richard Crandall and discoverers of previous Mersenne primes when a new prime is reported. However, Scott reports that something about the SMTP setup got screwed up in the two years since M#38 was found. In fact, one of the reasons I was a little suspicious of the report this morning was that I had not received the primenet email. It seems that the eagle-eyed Henk Stokhorst may have been the first to know that M#39 was uncovered. Brad at Entropia.com is trying to fix the bug in the primenet reports that showed both a VERIFIED and UNVERIFIED prime discovered. It should read one UNVERIFIED prime has been found. Give him a little more time to get it just right. Having a good day, George _________________________________________________________________________ Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ End of Mersenne Digest V1 #905 ******************************