Mersenne Digest Tuesday, October 12 1999 Volume 01 : Number 641 ---------------------------------------------------------------------- Date: Tue, 12 Oct 1999 12:54:34 +1000 From: Simon Burge <[EMAIL PROTECTED]> Subject: Re: Mersenne: Re: GIMPS "John R Pierce" wrote: > a year on one of these [a vax] wouldn't equal one day on a pentium-II. Probably a bit generous there even, given that older vaxen wouldn't have pipelined FPU's so you might get one result every 10 (or perhaps even 100) clocks, as opposed to one result almost every clock on modern hardware. Simon. _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 01:38:01 EDT From: [EMAIL PROTECTED] Subject: Mersenne: DOSville 'ello. This is quoting multiple people. They know who they are. Most of this is s'posed to be funny, by the way. :-D <<It suddenly links up to the server and downloads new exponens....I am sure testing LL 9xxxxxx is enough on a cyrix....its my baby!>> <<You shouldn't run LL tests on a Cyrix -- factoring will utilize your CPU much better (since Cyrixes are so bad at LL testing.). But you probably know that already.>> </JOKE> Intel is the one true path, the middle way to Nerdvana. All other chipmakers shall pale in comparison to the glorious might of Integrated Electronics! </JOKE> Seriously, I can't wait for the Merced. Better yet, McKinley. *drool* Watch Prime95 on those chips, whoo! <<Very dissappointed at the supplementation and interface....>> Hrm. GIMPS gives you several FAQs, websites, easy-to-contact designers, a mailing list, etc. Doc files aren't that good anyways. Disappointmentville. <<The GIMPS software was never designed to be user-friendly -- it was designed to be fast. If that doesn't suit you (and it doesn't suit many people), you are entirely free to choose a different project.>> Preach it, brother! Testify! Gooeys would make Prime95 bloated and (in all likelihood) slower. Not to mention create places for bugs to creep in. <<Now what is wrong with .txt files?>> .TXT files are the one true document file. Pure ASCII.... bring it on! Mathematical notation in .TXT isn't easy, though. It's a conspiracy. <<FOR GOD SAKE GIVE ME A USER FRIENDLY INTERFACE, PROPER HELP MENU....>> Give me a DOS-based command line (yet Win98 background-running capability) with a really nasty, God-awful set of fifteen switches that must be ordered in a precise way depending on the phase of the moon. :-P Now for the serious part: Lighten up. Ask questions 'bout whatever you're confused 'bout on the mailing list and the nice people here will help you. GIMPS is a great project. S. "Property of Billy and Andy" L. _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Mon, 11 Oct 1999 23:43:08 -0600 From: "Aaron Blosser" <[EMAIL PROTECTED]> Subject: RE: Mersenne: resuming a prime >8410531 63 1169884 23.3 30.0 46.0 27-Sep-99 00:53 18-Sep-99 18:08 > 3rd -- just reinstalled the gimps client on my home PC. How do I tell my > home PC to resume prime exponent 8410531, which it was working on before > the crash ? Just add a line to the worktodo.ini file (or make it if you haven't started the GIMPS client yet) that reads: Test=8410531,63 A doublecheck would look like: DoubleCheck=exponent,factoredbits That's all. The 8410531 is the exponent to test, and the 63 is how many bits it's already been factored to. That particular exponent would now be factored to 64 bits under the v19 client, so if you update the client to that version, expect it to factor up to 64 bits before starting on the LL test. I guess we'll really see how well v19 is doing. I just switched my machines over (most of 'em anyway) to v19, all running the NT service version. Tomorrow I'll try updating the Win98 machines I have with the new prime95. Harder to do that since I can't do it remotely as easily as with NT boxes. Someday, Windows 95/98 will be gone and my life will be easier. So far, all are working well. And I've got about 7 boxes running Windows 2000 of various builds. Some Beta 3, some RC1 and some RC2. A couple Advanced Server and others are Win2K Professional. All are running v19 just fine (and were running v18 just fine also). Aaron _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 13:49:55 +0200 (CEST) From: Attila Megyeri <[EMAIL PROTECTED]> Subject: Re: Mersenne: Mprime Error 2250 On Mon, 11 Oct 1999, Brian J. Beesley wrote: > Tip for diagnosis of problems: entropia.com's IP address is > 207.104.25.155 (though this isn't fixed, it shouldn't change often). > If you're getting problems connecting to entropia.com, try pinging > the numeric IP address; if that works, try pinging by name. If the > first fails, there is a network fault; if the second fails, there is > a fault in the DNS configuration or server; if both work, you *may* > have an application error (though there's still a chance that the > server isn't running the appropriate protocol for your application). I've just tried to use sprime v19 under Debian 2.1 (kernel 2.2.12) and also get the following error message immediately following 'mprime -c': ^^^^^^^^^^^ - --------------------------------------- Updating user information on the server ERROR 2250: Server unavailable The FAQ at http://www.entropia.com/ips/faq.html may have more information. - --------------------------------------- Pinging entropia.com: - ---------------------------------------- PING entropia.com (207.104.25.155): 56 data bytes 64 bytes from 207.104.25.155: icmp_seq=0 ttl=110 time=591.7 ms 64 bytes from 207.104.25.155: icmp_seq=1 ttl=110 time=602.7 ms 64 bytes from 207.104.25.155: icmp_seq=3 ttl=110 time=541.7 ms 64 bytes from 207.104.25.155: icmp_seq=4 ttl=110 time=533.9 ms - --- entropia.com ping statistics --- 5 packets transmitted, 4 packets received, 20% packet loss round-trip min/avg/max = 533.9/567.5/602.7 ms - ----------------------------------------- Also the same with IP numbers. It seems that the connection error may not be caused by network or DNS configuration errors. v18 connects without any problems on the same machine. My /etc/nsswitch.conf file: - ----------------------------------------- passwd: compat group: compat shadow: compat hosts: files dns networks: files protocols: db files services: db files ethers: db files rpc: db files netgroup: db files - ----------------------------------------- '/etc/host.conf' file: - ----------------------------------------- order hosts,bind multi on - ----------------------------------------- Any suggestions? Regards Attila Megyeri _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 15:30:59 +0300 From: Jukka Santala <[EMAIL PROTECTED]> Subject: Mersenne: Factoring numbers... Is it just me, or does factoring smaller Mersenne numbers take propotionally much longer? I would expect M727 to be much faster than the 33M range to a fixed depth, yet the opposite _seems_ to be true. Ofcourse, I can't be sure about this, because the real complaint I have is that factoring numbers to depths beyond the "default" seems nearly impossible. The manual factoring assignment seems like the only possibility to force these, yet it doesn't work like the normal factoring (Doing one bit depth at time) and is a pain on a dual-CPU machine. Is it possible we'd get a third parameter to the Factor= work-line specifying the intended depth for the factorization? Also, usually for some reason Prime95 seems to reject most (all?) Factor= statements I've tried, could we get some more detailed instructions on this? -Donwulff _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 14:31:34 +0100 From: Robin Stevens <[EMAIL PROTECTED]> Subject: Re: Mersenne: glitches in mprime v19? On Mon, Oct 11, 1999 at 12:06:07PM -0500, jason wrote: > Well, I also downloaded mprime.tar.gz, and I had a slightly different > problem trying to connect to the PrimeNet server. To summarize: > > mprime.tar.gz v19.0.2: results in "ERROR: Primenet error 1" > sprime.tar.gz v19.0.2: results in "Error 2250: Server unavailable" > mprime.tar.gz v18.1.2: No problems. > > I wonder why the two versions give different error messages (and why I'm > getting them at all)? Anyway, I just compiled mprime from source > (debugging not enabled), and it's working just fine. Except for the fact > that I will not receive credit for my work, since I compiled it myself... > so much for initiative. :-) Indeed - this is the problem I reported a little while ago (again with sprime, since I don't have glibc 2.1 as yet). If v19 works faster, I might as well use it even if I can't currently report results. My temporary solution has been to grab 90 days' of work, switch the machines to run as dialup hosts even if they aren't, and wait for someone else to fix it, reporting via the web interface if necessary :-) - -- Robin Stevens <[EMAIL PROTECTED]> http://www-astro.physics.ox.ac.uk/~rejs/ (+44) (0)1865: 273212 (work) 726796 (home) 273275 (fax) Pager: 0839 629682 Oxford University Computing Services, 13 Banbury Road, Oxford OX2 6NN, UK "Life, loathe it or ignore it, you can't like it." -- Marvin _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 16:00:24 +0200 (CEST) From: Henrik Olsen <[EMAIL PROTECTED]> Subject: Re: Mersenne: Factoring numbers... On Tue, 12 Oct 1999, Jukka Santala wrote: > Is it just me, or does factoring smaller Mersenne numbers take > propotionally much longer? I would expect M727 to be much faster than > the 33M range to a fixed depth, yet the opposite _seems_ to be true. It's not just you, it's a natural consequence of the property that all factors of Mp must be of the form 2kp+1, so with a fixed depth and larger p there are fewer possible factors to check. > > Ofcourse, I can't be sure about this, because the real complaint I have > is that factoring numbers to depths beyond the "default" seems nearly > impossible. The manual factoring assignment seems like the only > possibility to force these, yet it doesn't work like the normal > factoring (Doing one bit depth at time) and is a pain on a dual-CPU > machine. Is it possible we'd get a third parameter to the Factor= > work-line specifying the intended depth for the factorization? Also, > usually for some reason Prime95 seems to reject most (all?) Factor= > statements I've tried, could we get some more detailed instructions on > this? > > -Donwulff > > _________________________________________________________________ > Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm > Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers > - -- Henrik Olsen, Dawn Solutions I/S URL=http://www.iaeste.dk/~henrik/ No, power off on shutdown is not SMP safe. It kind of happens to work on a lot of boards. If making that APM call reformats your disk and plays tetris on an SMP box, the bios vendor is within spec (if a little peculiar). Alan Cox, on the Linux Kernel list _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 10:51:30 -0400 From: George Woltman <[EMAIL PROTECTED]> Subject: Re: Mersenne: Factoring numbers... Hi, At 03:30 PM 10/12/99 +0300, Jukka Santala wrote: >Is it just me, or does factoring smaller Mersenne numbers take >propotionally much longer? I would expect M727 to be much faster than >the 33M range to a fixed depth, yet the opposite _seems_ to be true. Since factors are of the form 2kp+1 there are many more factors to test when you factoring M727. Yes each factor can be tested in less time, but this is overwhelmed by extra factors to test. >the real complaint I have >is that factoring numbers to depths beyond the "default" seems nearly >impossible. Yes, the only way to trial factor to a greater depth is with the Advanced/Factor menu choice - and it is difficult to use. Improving this is in the wish list, but didn't make the cut for v19. The biggest reason is: More trial factoring is likely a waste of time. You will find more factors using ECM. Compare the cost of running 700 curves at B1=250000 (this will find most 30-digit factors) against trial factoring to 99 bits! It is true that ECM can miss a factor, but for M727 which has had thousands of curves run at bounds as high as 50,000,000 - the chance of finding a factor smaller than 30 digits is.....well, zero. Regards, George _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 11:00:05 -0400 (EDT) From: Darxus <[EMAIL PROTECTED]> Subject: Mersenne: splitting up 10m digit primes 704.5 days to go on this 10m digit prime my computer at home is working on. P2 233. 1.93 years. I think it said 1 in 250,000 chance if finding a prime. So.. on average, it would probably take that one computer, by itself, 241,250 years to find a 10m digit prime. Right ? I'm okay with that. But I think, if possible, it'd be good to break up primes into like, 1 month chunks, & distribute them. I'm sure it'd be possible, I just don't know if/how much it'd impact speed. I also think it would have been better to award $5k per new prime, of any length. But that's just my opinion. And how is the probability of finding a prime calculated ? I wanna do a comparison of the prize money to probability ratio between distributed.net's rc5-64 project ($2k prize), & GIMPS 10m digit prime ($55k prize). But it'll take me a chunk of time. Any estimates ? __________________________________________________________________ PGP fingerprint = 03 5B 9B A0 16 33 91 2F A5 77 BC EE 43 71 98 D4 [EMAIL PROTECTED] / http://www.op.net/~darxus Join the Great Internet Mersenne Prime Search http://www.mersenne.org/prime.htm _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 12:03:01 -0400 From: Jud McCranie <[EMAIL PROTECTED]> Subject: Re: Mersenne: Re: GIMPS At 12:54 PM 10/12/99 +1000, Simon Burge wrote: >"John R Pierce" wrote: > > > a year on one of these [a vax] wouldn't equal one day on a pentium-II. > >Probably a bit generous there even, given that older vaxen wouldn't >have pipelined FPU's so you might get one result every 10 (or perhaps >even 100) clocks, as opposed to one result almost every clock on modern >hardware. In his book "Programming Pearls", Jon Bentley gives the figure of 570 seconds for sorting 10,000 integers on a VAX-11/750. My PII-300 does it 1160 times faster and my C-400 is 1780 times faster. But that is comparing working with integers instead of FP. +---------------------------------------------------------+ | Jud McCranie | | | | Programming Achieved with Structure, Clarity, And Logic | +---------------------------------------------------------+ _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 13:24:38 -0400 From: George Woltman <[EMAIL PROTECTED]> Subject: Re: Mersenne: splitting up 10m digit primes Hi, At 11:00 AM 10/12/99 -0400, Darxus wrote: >704.5 days to go on this 10m digit prime my computer at home is working >on. P2 233. 1.93 years. I admire your patience! > I think it said 1 in 250,000 chance if finding >a prime. So.. on average, it would probably take that one computer, by >itself, 241,250 years to find a 10m digit prime. Right ? Define "probably". 241,250 years gives you a 50% chance. Actually it will take longer since the exponents get bigger and bigger. >it'd be good to break up >primes into like, 1 month chunks, & distribute them. A good idea but the Lucas-Lehmer primality test is a "serial" algorithm. That is, I can't have 33 machines each do a million iterations and get the answer in a month. The second million iterations can't start until the first million complete. It would be nice if someone invented a primality test that could be done in parallel. >I also think it would have been better to award $5k per new prime, of any >length. But that's just my opinion. The prize fund was set up by the EFF and the anonymous donor. I agree with you and have tried to encourage an orderly progression by awarding $5,000 to all smaller Mersenne primes (but only if we also find the 10 million digit prime). >And how is the probability of finding a prime calculated ? It is roughly how-far-factored-in-bits * 2 / exponent >I wanna do a comparison of the prize money to probability ratio between >distributed.net's rc5-64 project ($2k prize), & GIMPS 10m digit prime >($55k prize). But it'll take me a chunk of time. Any estimates ? There is now a prize for factoring Fermat numbers too. However, the best investment is probably to turn off your computer and pocket the $30 you save in electricity! Of course, that's no fun. So pick whichever project gives you the biggest thrill. Regards, George _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 12:49:14 -0500 (CDT) From: jason <[EMAIL PROTECTED]> Subject: Re: Mersenne: glitches in mprime v19? OK, now the dynamically-linked v19.0.2 version is working fine. I guess the "ERROR: Primenet error 1" was occurring while PrimeNet was down, as Scott pointed out, not because there was anything buggy with the client. It was just bad timing that I tried connecting then. :-) The static executable still gives "Error 2250: Server unavailable", though. Now I'm curious. v19 is linked against glibc2.1, right? I'm running a glibc2 (not glibc2.1) machine (redhat 5.1 base, but almost everything has been upgraded by compiling from source for a long time now). But the v19 dynamically-linked client is running just fine for me. Does anyone know why? (I'm not complaining! :-) Anyway, in light of my success here, I would suggest to those of you trying unsuccessfully to run the static executable on a non-glibc2.1 machine to give the dynamically-linked executable a try. Of course, your mileage may vary. (If you still have problems, feel free to email me, and we'll try to sort out how our systems differ. I'm sure I'm not the only one who'd like to document what's making or breaking mprime.) - -jason On Tue, 12 Oct 1999, Robin Stevens wrote: > On Mon, Oct 11, 1999 at 12:06:07PM -0500, jason wrote: > > Well, I also downloaded mprime.tar.gz, and I had a slightly different > > problem trying to connect to the PrimeNet server. To summarize: > > > > mprime.tar.gz v19.0.2: results in "ERROR: Primenet error 1" > > sprime.tar.gz v19.0.2: results in "Error 2250: Server unavailable" > > mprime.tar.gz v18.1.2: No problems. > > > > I wonder why the two versions give different error messages (and why I'm > > getting them at all)? Anyway, I just compiled mprime from source > > (debugging not enabled), and it's working just fine. Except for the fact > > that I will not receive credit for my work, since I compiled it myself... > > so much for initiative. :-) > > Indeed - this is the problem I reported a little while ago (again with > sprime, since I don't have glibc 2.1 as yet). If v19 works faster, I might > as well use it even if I can't currently report results. My temporary > solution has been to grab 90 days' of work, switch the machines to run as > dialup hosts even if they aren't, and wait for someone else to fix it, > reporting via the web interface if necessary :-) > -- > Robin Stevens <[EMAIL PROTECTED]> http://www-astro.physics.ox.ac.uk/~rejs/ > (+44) (0)1865: 273212 (work) 726796 (home) 273275 (fax) Pager: 0839 629682 > Oxford University Computing Services, 13 Banbury Road, Oxford OX2 6NN, UK > "Life, loathe it or ignore it, you can't like it." -- Marvin > _________________________________________________________________ > Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm > Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers > _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 14:07:08 -0400 From: Jeff Woods <[EMAIL PROTECTED]> Subject: Re: Mersenne: splitting up 10m digit primes At 11:00 AM 10/12/99 -0400, you wrote: >I'm okay with that. But I think, if possible, it'd be good to break up >primes into like, 1 month chunks, & distribute them. I'm sure it'd be >possible, I just don't know if/how much it'd impact speed. Not possible. Well, POSSIBLE, but it would actually slow the effort down. The problem is that the test you're performing uses the results of the last "loop" to compute the next value. To make the math simpler, I won't use the actual LL test, but something much simpler: X = 0; DO 32,722,124 times X = X + 1; Now, if you're running this prgram, forget for a moment that you and I can easily look ahead and see that after a thousand times, X is 1000. Unfortunately, there's no "shortcut" in the Lucas-Lehmer test whereby one can look ahead. You have to do this: X = 0 Prior result was 0 so 0 + 1 = 1 Prior result was 1 so 1 + 1 = 2 Prior result was 2 so 2 + 1 = 3 Prior result was 3 so 3 + 1 = 4 Prior result was 4 so 4 + 1 = 5 Prior result was 5 so 5 + 1 = 6 etc. Note that you cannot just "jump ahead" to the 120th loop, because you need to know what the RESULT of the 119th loop was before you can start. And you need the 118th loop to start the 119th, etc, all the way back to loop # 1. If you're testing an exponent of 32,361,147 (for example), you *must* do all 32,361,144 calculations, and they have to be done in order. If you tried to "break them up", with, say, 100,000 "iterations" per user, then the person with 100,001 through 200,000 cannot even start 100,001 until all 100,000 iterations of the last person had been finished. If you tried this, you'd only slow things down, because you'd have to send the "mid-test results" back to Entropia, and then out to the next tester, before they can start. That's what would slow things down -- the back-and-forth of the intermediate results. Yes, it takes a while, but its very much best for one person to just do the whole test, even though it's going to take two years. _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 22:41:37 +0300 From: Jukka Santala <[EMAIL PROTECTED]> Subject: Re: Mersenne: Factoring numbers... George Woltman wrote: > Since factors are of the form 2kp+1 there are many more factors to test > when you factoring M727. Yes each factor can be tested in less time, > but this is overwhelmed by extra factors to test. Thanks to everybody pointing this one out, I missed the obivious implication of the form of factors earlier. It does make sense now, though. > You will find more factors using ECM. Compare the cost of running > 700 curves at B1=250000 (this will find most 30-digit factors) against > trial factoring to 99 bits! It is true that ECM can miss a factor, > but for M727 which has had thousands of curves run at bounds as high > as 50,000,000 - the chance of finding a factor smaller than 30 digits > is.....well, zero. By now it should be obivious I'm not too well versed in this stuff, however... I've always been told that ECM is "not guaranteed to find all factors", is it however expected (or guaranteed, if you will) to find all factors under X bits given _enough_ curves? And is the probability of finding any given factor by ECM only a function of it's number of bits, or do other things affect it as well..? -Donwulff _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 20:45:49 +0100 From: "Brian J. Beesley" <[EMAIL PROTECTED]> Subject: Re: Mersenne: Factoring numbers... On 12 Oct 99, at 15:30, Jukka Santala wrote: > Is it just me, or does factoring smaller Mersenne numbers take > propotionally much longer? I would expect M727 to be much faster than > the 33M range to a fixed depth, yet the opposite _seems_ to be true. Yes, it _does_ take _much_ longer. The time to test one factor is proportional to the logarithm to base 2 of the exponent, rounded up to the next integer. So each trial takes 10T for M727, but 25T for M33219281 - 2.5 times as long. But, since factors of Mp are of the form 2kp+1, there are almost 50,000 times as many candidate factors of M727 in any given interval. So trial factoring M727 over a given interval takes almost 20,000 times as long as trial factoring M33219281 over the same interval. Note - T above is fixed only for numbers which fit neatly into the same number of words. > > Ofcourse, I can't be sure about this, because the real complaint I have > is that factoring numbers to depths beyond the "default" seems nearly > impossible. The manual factoring assignment seems like the only > possibility to force these, yet it doesn't work like the normal > factoring (Doing one bit depth at time) and is a pain on a dual-CPU > machine. Is it possible we'd get a third parameter to the Factor= > work-line specifying the intended depth for the factorization? I guess it would be possible, but it might be a waste of time (in the long run). The factoring depth limit for each range of has been chosen carefully to balance the chance of finding a "small" factor (this saving running a LL test), which increases only slowly (less than linearly) with factoring depth, against the time spent trial factoring, which increases exponentially (doubling with each extra bit of factoring depth). > Also, > usually for some reason Prime95 seems to reject most (all?) Factor= > statements I've tried, could we get some more detailed instructions on > this? Prime95 will kick out factoring assignments when it considers that the exponent has already been factored far enough. There are definitely exponents in the database which have been LL tested (some even double-checked) but have not been trial factored as deeply as v19 would do. Suggest you pull down the database files nofactor.zip (which now needs a program called decomp.zip to decode it) & lucasv.zip. Make a list of exponents which have been double- checked & have been factored less than the v19 limits. Construct Factor= lines for these exponents. Should keep your system busy for a while! If you do decide to do this, you're on your own - don't be surprised if someone else does the same thing independently - there's no mechanism for coordination to avoid wasted effort. Of course, you could try ECM or P-1 factoring instead ... there is some coordination in this area, & in any case there's so much work to do that it's easy not to tread on anyone else's toes! Regards Brian Beesley _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 16:12:12 -0400 (EDT) From: Darxus <[EMAIL PROTECTED]> Subject: Re: Mersenne: splitting up 10m digit primes On Tue, 12 Oct 1999, George Woltman wrote: > I admire your patience! Thank you :) > > I think it said 1 in 250,000 chance if finding > >a prime. So.. on average, it would probably take that one computer, by > >itself, 241,250 years to find a 10m digit prime. Right ? > > Define "probably". 241,250 years gives you a 50% chance. Actually it will > take longer since the exponents get bigger and bigger. Mathmatical probability... on average, it'll take 50% of the whole thing. At the time I didn't know it took longer toward the end. > >it'd be good to break up > >primes into like, 1 month chunks, & distribute them. > > A good idea but the Lucas-Lehmer primality test is a "serial" algorithm. > That is, I can't have 33 machines each do a million iterations and get the > answer in a month. The second million iterations can't start until the > first million complete. I see. Well, we need more people involved either way :) Other thing I gotta calculate... how long it would probably take us if we converted everyone from distributed.net to GIMPS. This is actually the biggest reason why I wanna calculate prize/probability ratio of the 2. > >I also think it would have been better to award $5k per new prime, of any > >length. But that's just my opinion. > > The prize fund was set up by the EFF and the anonymous donor. I agree with > you and have tried to encourage an orderly progression by awarding $5,000 > to all smaller Mersenne primes (but only if we also find the 10 million > digit prime). I understand the contest was not set up by you, but it's good to hear you're trying to incourage $5k/prime. I doubt there are many people who's input could be more valued for this. > >And how is the probability of finding a prime calculated ? > > It is roughly how-far-factored-in-bits * 2 / exponent Okay.. what's "how-far-factored-in-bits" mean ? > >I wanna do a comparison of the prize money to probability ratio between > >distributed.net's rc5-64 project ($2k prize), & GIMPS 10m digit prime > >($55k prize). But it'll take me a chunk of time. Any estimates ? > > There is now a prize for factoring Fermat numbers too. Neat. Where's the info ? Also, the link pertaining to the EFF $100k award on the GIMPS page goes directly to the EFF rules page, instead of http://www.mersenne.org/prize.htm. I think it'd be nice to have a link from the main GIMPS page to something w/ info on all prizes which could be won with GIMPS. You were probably planning on doing that though... > However, the best investment is probably to turn off your computer > and pocket the $30 you save in electricity! Of course, that's no fun. > So pick whichever project gives you the biggest thrill. Alright then, it's a gamble, not an invenstment. Wait, no, I'd leave my computer on all the time anyway. It's still an investment. Yeah yeah, so if I didn't run one of these, the CPU would be off less & draw less power. It doesn't matter, I like distributed processing. I'm not in it for the money. __________________________________________________________________ PGP fingerprint = 03 5B 9B A0 16 33 91 2F A5 77 BC EE 43 71 98 D4 [EMAIL PROTECTED] / http://www.op.net/~darxus Join the Great Internet Mersenne Prime Search http://www.mersenne.org/prime.htm _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 16:15:11 EDT From: [EMAIL PROTECTED] Subject: Re: Mersenne: splitting up 10m digit primes In a message dated Tue, 12 Oct 1999 2:16:30 PM Eastern Daylight Time, Jeff Woods <[EMAIL PROTECTED]> writes: > If you tried to "break them up", with, say, 100,000 "iterations" per user, > then the person with 100,001 through 200,000 cannot even start 100,001 > until all 100,000 iterations of the last person had been finished. > > If you tried this, you'd only slow things down, because you'd have to send > the "mid-test results" back to Entropia, and then out to the next tester, > before they can start. That's what would slow things down -- the > back-and-forth of the intermediate results. > > Yes, it takes a while, but its very much best for one person to just do the > whole test, even though it's going to take two years. Well, as completion time gets longer and longer, it becomes more likely that a user will give up in disgust. This could be after several months of work is already complete. How about an option when you hit "QUIT GIMPS" to upload your P and Q files to Primenet, so someone can at least finish the job? _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 23:14:56 +0300 From: Jukka Santala <[EMAIL PROTECTED]> Subject: Re: Mersenne: Factoring numbers... "Brian J. Beesley" wrote: // Regarding factoring Mersennes to v19 depth... > If you do decide to do this, you're on your own - don't be surprised > if someone else does the same thing independently - there's no > mechanism for coordination to avoid wasted effort. Partially. There now exists software to automate most of this, with intent to update the shared status files every month. Don't remember the URL, but little browsing of the Prime95 homepage should turn this stuff up. It's not exactly what I was looking for, though, I'm playing on filling some of the holes in Cunningham-tables and this means expending some extra effort on the lower Mersenne numbers. I'd like to have trial-factoring beyond the v19 limit among the options to go at this, altough I'm understanding this isn't apparently very fruitful method. Anyway, still waiting to hear if ECM will, eventually, find all factors or if it leaves "factors" in the range... -Donwulff _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 17:53:05 -0400 From: "Rick Pali" <[EMAIL PROTECTED]> Subject: RE: Mersenne: splitting up 10m digit primes From: [EMAIL PROTECTED] > How about an option when you hit "QUIT GIMPS" to > upload your P and Q files to Primenet, so someone > can at least finish the job? I'm running an exponent in the 33 million area and the save-files are over seven megabytes in size! That would require no small amount of server space... Rick. - -+--- [EMAIL PROTECTED] http://www.alienshore.com/ _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 18:06:35 -0400 (EDT) From: Lucas Wiman <[EMAIL PROTECTED]> Subject: Re: Mersenne: splitting up 10m digit primes > > There is now a prize for factoring Fermat numbers too. > > Neat. Where's the info ? I think Richard Crandall is offering a prize for Fermat factors (http://www.perfsci.com). John Selfridge is also anouncing a prize for factors of various numbers which "ought to be prime". I don't think that there is a website yet. I'll repost the list if you are interested. - -Lucas _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 23:09:07 +0200 From: "Steinar H. Gunderson" <[EMAIL PROTECTED]> Subject: Mersenne: Re: [Windows net utilities] On Mon, Oct 11, 1999 at 10:10:04PM +0100, Brian J. Beesley wrote: >Windows users might care to try a nice program called CyberKit, which >is freeware & does ping, traceroute & NS lookup (amongst other >things). I don't know if Windows does `other things', but it certainly has ping (ping) and traceroute (tracert). It lacks a decent `host', though, but ping will do most of the work for you. /* Steinar */ - -- Homepage: http://members.xoom.com/sneeze/ _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 12 Oct 1999 23:06:11 +0200 From: "Steinar H. Gunderson" <[EMAIL PROTECTED]> Subject: Mersenne: Re: splitting up 10m digit primes On Tue, Oct 12, 1999 at 04:15:11PM -0400, [EMAIL PROTECTED] wrote: >Well, as completion time gets longer and longer, it becomes more likely >that a user will give up in disgust. This could be after several months >of work is already complete. That's part the reason why searching for >10,000,000 digits is not yet default. >How about an option when you hit "QUIT GIMPS" to upload your P and Q >files to Primenet, so someone can at least finish the job? There are two problems with this approach. First, entropia.com will need some HD space to hold these files (they are big...). Second, they will also need bandwidth to receive all those big (and not very compressible) files. I think the discussion has been up before, and Scott (I think) pointed out that extremely few exponents were quit far out in the testing phase. So, I don't think it's worth it (I think that's the conclusion we reached last time, too). We might save a few P90 CPU-years, but it would mean lots of extra code, and less bandwidth and server space available. We shouldn't clutter up the Internet _too_ much :-) /* Steinar */ - -- Homepage: http://members.xoom.com/sneeze/ _________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ End of Mersenne Digest V1 #641 ******************************
