Mersenne Digest Tuesday, October 9 2001 Volume 01 : Number 890 ---------------------------------------------------------------------- Date: Sat, 6 Oct 2001 09:03:05 +0100 From: "Tony Gott" Subject: Mersenne: MLucas benchmarks How does the new version 2.8c compare with Prime95 beta-version 21.4 as far as benchmarks is concerned? It would be just fantastic if the full power of the Altivec in Apple G4 could be utilised as it has been in the RC5 distributed project. Is this ever likely? Tony _________________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Sat, 06 Oct 2001 10:35:03 -0700 From: "Robert Braunwart" Subject: Mersenne: Re: Mersenne Digest V1 #889 I second this suggestion. I would be very interested in seeing a calculator page, or at least a detailed formula. I neglected to keep track of the individual credit for my earliest numbers, and now I am curious. These go back to v. 15, I believe. - --Bob Braunwart >Date: Fri, 5 Oct 2001 17:36:14 -0400 >From: "Digital Concepts" >Subject: Mersenne: What are the PrimeNet formulas for calculating LL and >Factoring credit? > >This is a multi-part message in MIME format. > >- ------=_NextPart_000_02A6_01C14DC4.3B0A3CE0 >Content-Type: text/plain; > charset="iso-8859-1" >Content-Transfer-Encoding: quoted-printable > >First a short memory recollection... A few months back I wrote in = >explaining that I wanted to compare contributions (work done in CPU = >years) between clients running under a team umbrella. I was privately = >replied to and referred to the benchmark page as the way they would do = >it. More recently, someone asked the same thing and I believe George = >referred them to the status page. I've done it both ways and they come = >out very close, but I assume the status page is the one that is going to = >be kept current as new releases and timings come along, and the one = >George uses. > >Either time though, it was mentioned how difficult it was to calculate = >factoring years, and I see why. Yet there must be a formula. Who here = >is a PrimeNet rep? > >With this info, maybe someone might volunteer a calculator page, where = >someone could enter an exponent and get the LL value, and optionally = >whatever criteria is stored in factoring messages to PrimeNet, to = >determine the Factoring value. > >Thanks for you consideration, >Carleton Garrison LL#164 F#291 G#253 www.teamprimerib.com > _________________________________________________________________ Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp _________________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Sun, 7 Oct 2001 15:28:06 -0400 (EDT) From: Jason Stratos Papadopoulos Subject: Re: Mersenne: MLucas benchmarks On Sat, 6 Oct 2001, Tony Gott wrote: > It would be just fantastic if the full power of the Altivec in Apple G4 > could be utilised as it has been in the RC5 distributed project. Is this > ever likely? Not for a floating point FFT; the native Altivec floating point precision is too low to be of use for Mersenne computations. Using Altivec for an all-integer FFT is a possibility, but I don't know for the life of me how best to use it. Experience shows that it's usually not a good idea to use multimedia hardware to perform all-integer multiple-precision work; the resulting code is extremely complex and usually too slow. jasonp _________________________________________________________________________ Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers ------------------------------ Date: Tue, 9 Oct 2001 23:23:19 +0200 (MET DST) From: Reto Keiser Subject: Mersenne: Re: Factoring Failure? Hi all, Daniel Swanson wrote: > 7019297 57 DF 160100125459121849 27-Sep-01 22:52 > 7020641 58 DF 226230108157229263 30-Sep-01 02:05 > 7025987 56 DF 74052063365823791 30-Sep-01 01:12 > 7027303 55 DF 31090234297428433 30-Sep-01 22:14 > 7028947 58 DF 203918491658210359 01-Oct-01 03:11 > 7033963 56 DF 100945633281264553 03-Oct-01 07:36 > 7036409 58 DF 321885922408857601 04-Oct-01 02:54 'ribwoods' wrote: > I'm pretty sure the second column is the previous trial-factoring > limit, in units of power-of-2. Actually, when a factor is found, the primenet server returns the size of the factor in bits, but it rounds the value Log2(factor) mathematically to the next integer and not always upwards what would make more sense. Therefore the factor can be bigger than 2^n shown in the second column. There are some people doing broadband factoring, what means that they extend the factoring limit on a big range of numbers by 1 or a few bits using the factoroverride option. Hence, it is very likely, that the fault happened during that process and not by the person who was doing the first LLTest (At that time the P-1 Test was not available yet, and the factoring limit beyond the factor). While completing some trial factoring in the 60 million range I noticed, that the prime95 checksum of factoring from 65 to 66 bits and from 66 to 68 bits are the same (I did this two parts of the same exponent on different computers). That means that it is easily possible that some mistakes can happen, when someone writes the 'Factor=' line into the worktodo file. Another reason might be, that one person ran a broadband factoring range from 55 to 58 bits on one computer, from 58 to 59 on another (to split up the work) and forgot to check in the former results. As there is no information about the starting bit in the result file, the primenet server is not able to detect that problem. I think the easiest way to fix this problem is to recheck these ranges, which were completed by the user(s) where more than one factor appeared. As the server also logs the id, that would not be too difficult to realize. Probably that problem might become worse in future: George told me, that primenet shows the clients, that a p-1 is already done, by adding 0.5 to the factoring limit. Prime95 detects that correctly, but in the reports these values are rounded to the next digit (64.5 -> 65 and so on). If someone takes the factoring information from these lists, if would cause a gap of one bit in the factoring process. I suggest to change the rounding behavior of the primenet server or to decrease the value which indicates a completed p-1 test from 0.5 to 0.4. I hope that we will find the missing factors without too much effort. Regards, Reto _________________________________________________________________________ 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 #890 ******************************