mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > PrimeNet

Reply
 
Thread Tools
Old 2012-06-28, 07:23   #1
dbaugh
 
dbaugh's Avatar
 
Aug 2005

22×29 Posts
Default How to handle ECM results?

I would like to do some ECM work. 100 curves takes a long time and if I quit before all 100 are finished there is no results entry. I can do the modify worktodo trick to next curve and get say 73 curves reported in results. I would like to queue up a string of 10 curve sets in worktodo. This gives me a string of 10 curve sets in results. I don't want to junk up the history for an exponent with scores of ECM entries.

It appears to be very easy to edit the results file and report 100 curves completed instead of 10 sets of 10. If this is true then someone could just as easily report 1000 curves completed. What is considered to be the appropriate way to run say 600 curves on an exponent without nuisance reporting and without committing to a long uncertain voyage of running 600 curves?
dbaugh is offline   Reply With Quote
Old 2012-06-28, 09:14   #2
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

22·3·599 Posts
Default

Quote:
Originally Posted by dbaugh View Post
I don't want to junk up the history for an exponent with scores of ECM entries.
Don't worry about it; that's what data bases are for. Other ECMers do that all the time -- look at some low exponents where there've been hundreds of reports of 3 curves at a time.

If this "junk up" ever became a real problem, that could easily be handled on the server side, such as by appropriately combining reports on a line. You needn't make that your worry.

Quote:
What is considered to be the appropriate way to run say 600 curves on an exponent without nuisance reporting
If by "nuisance reporting" you mean reporting 600 curves only 10 curves at a time, so that there are 60 reports: don't worry about that. Just send GIMPS your reports after you've run the curves in whatever divvy-up is most convenient for you.

Quote:
and without committing to a long uncertain voyage of running 600 curves?
Lessee ... you want to run 600 curves, but it's scary to look at that total, right? So ... just break it up as you've already proposed.

There's no magic trick to be "appropriate", other than what you already know, except to relax and let the computers deal with the big numbers, like 600, while you attend to just 10 at a time. :-)

(Think of it as a "divide and conquer" strategy, just as is used in many mathematical algorithms.)
cheesehead is offline   Reply With Quote
Old 2012-11-09, 00:43   #3
dbaugh
 
dbaugh's Avatar
 
Aug 2005

11610 Posts
Default

Is there a slick way to identify exponents with extensive histories. Some, such as 1277, are very well deserved. I remember stumbling across a much larger one that seemed to have been caused by a reporting loop bug or something. I'd like to find it again and squash it with a factor.
dbaugh is offline   Reply With Quote
Old 2012-11-09, 03:42   #4
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

100011101010002 Posts
Default

http://www.mersenne.org/report_ECM/
Scroll to see the whole page.
Batalov is offline   Reply With Quote
Old 2012-11-09, 04:24   #5
dbaugh
 
dbaugh's Avatar
 
Aug 2005

11610 Posts
Default

That is a good site to find exponents that have had a lot of curves run. I am looking for ones that have had like 500 curves run, but that were submitted one at a time. Resulting in 500 history lines rather than one.
dbaugh is offline   Reply With Quote
Old 2012-11-09, 07:11   #6
aketilander
 
aketilander's Avatar
 
"Åke Tilander"
Apr 2011
Sandviken, Sweden

2·283 Posts
Thumbs down

Quote:
Originally Posted by dbaugh View Post
It appears to be very easy to edit the results file and report 100 curves completed instead of 10 sets of 10.
I would say that I think that you never ever should edit a results file, or rather the content of each resultline, since that may cast doubt on your entire work. Each resultline is the outpout from the program and if they should be trusted they should be written exactly as the program did to the last jota, so please don't change anything.
aketilander is offline   Reply With Quote
Old 2012-11-09, 19:27   #7
dbaugh
 
dbaugh's Avatar
 
Aug 2005

22·29 Posts
Default

I am not intending to edit any results lines. I just report the results lines as they are. What I am asking about now is how to find exponents where hundreds of single curves were run. My plan is to find a factor, no editing.
dbaugh is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
gfn results ET_ FermatSearch 8 2019-06-02 12:09
Where can I find a arbitrary precision Calculator online that can handle this # ONeil Information & Answers 9 2018-04-17 18:18
Religious extremism leading to conflict and slaughter: how to handle this? Brian-E Soap Box 15 2014-05-16 03:12
Can msieve handle c197's with SNFS? david314 Msieve 21 2012-07-29 15:21
New PC can handle two instances of Prime 95 Bundu Software 9 2004-08-21 02:29

All times are UTC. The time now is 17:40.

Wed Oct 28 17:40:13 UTC 2020 up 48 days, 14:51, 2 users, load averages: 1.99, 2.28, 2.31

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.