2005-09-23, 22:17 | #1 |
Oct 2003
Croatia
2^{3}×3×19 Posts |
Where to send ECM curves completed?
I just wonder, where should I send number of ECM curves I complete if it isn't for 2- and 2+ tables? I send my ECM curves for 2- and 2+ tables to George, but I would love to do some other tables as well.
Can I use Prime95 for other tables too? It seems that we can input k value in worktodo file. Thanks Edo |
2005-09-24, 13:31 | #2 |
"Nancy"
Aug 2002
Alexandria
2,467 Posts |
You can use Prime95 for numbers of the form a*2^n+-c, but not for numbers b^n+-1 if b!=2. So you can do ECM on Proth, SoB etc. numbers but for the Cunningham tables other than 2,n- and 2,n+ you'll need Dario Alpern's ECM applet or GMP-ECM at this time. (George said he has plans to extend Prime95 to b!=2 numbers)
Where to report curves and factors depends on which project hosts the info on the kind of number you are factoring. For Cunningham, send factors to Sam Wagstaff and tell Paul Zimmermann and our Cunningham subforum as well. Curve counts on Cunningham numbers should go to Paul and the forum (Sam doesn't keep track of them, only of factors). Alex |
2005-09-24, 14:04 | #3 |
Mar 2003
New Zealand
485_{16} Posts |
Prime95 ECM does work for a*b^n+c, but with b>2 it is slower than GMP-ECM for the current Cunningham composites which are all less than 400 digits.
Stage one can be faster than GMP-ECM at about 600-800 digits, depending on how many factors are already known. |
2005-09-24, 21:29 | #4 |
Oct 2003
Croatia
2^{3}×3×19 Posts |
Thanks for answers!
Is there any place where I can check number of curves already done for a*2^n+-c forms? Currently I run ECM on Mersenne numbers from George's list. Just recently find out how ECM is fun for crunching. Pure joy of crunching. :) Edo |
2005-09-25, 21:47 | #5 |
Mar 2003
New Zealand
13·89 Posts |
Here are a few projects that want numbers of this form, but I don't think they keep close track of how many curves have been done. You can look at the sizes of the factors being found to get an idea of what bounds to try.
k*2^n+-1 For odd k<14. n*2^n+-1 Cullen/Woodall numbers. n*b^n+-1 Generalized Cullen/Woodall numbers, b<13. There is no equivalent to lowm.txt for these forms, so add ContinueECM=1 to prime.ini to prevent Prime95 stopping after finding a known factor. You will also need to look through the results.txt output to check whether the factors are new or already known, so it is probably best to do a fair few curves on each candidate or else this becomes a hassle. If you just want to do one or two curves on each candidate then it is easier to use GMP-ECM. Prime95 works fastest on a*b^n+-c when a and b are minimised, so for the Generalized Cullen/Woodall convert the base 4 and base 8 forms into base 2. For example, the worktodo.ini entry for 500*8^500+1 should be ECM2=125,2,1502,1,<B1>,<B2>,<Curves> |
2005-09-26, 08:54 | #6 |
Oct 2003
Croatia
710_{8} Posts |
geoff, thanks for your detailed answer. Really appreciate it.
Edo |
Thread Tools | |
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
How to send a PM to Batalov | pinhodecarlos | Forum Feedback | 3 | 2016-06-08 02:43 |
Where do I send my PRP primes with large k? | Trilo | Riesel Prime Search | 3 | 2013-08-20 00:32 |
Completed 29M work not showing as completed in GPU72 | Chuck | GPU to 72 | 2 | 2013-02-02 03:25 |
Lost a) mod powers and b) ability to send PMs | 10metreh | Forum Feedback | 2 | 2010-09-08 18:06 |
Very odd error 7, send benchmark data. | Uncwilly | PrimeNet | 2 | 2008-12-29 04:48 |