mersenneforum.org  

Go Back   mersenneforum.org > Other Stuff > Archived Projects > NFSNET Discussion

 
 
Thread Tools
Old 2004-09-15, 16:14   #34
Wacky
 
Wacky's Avatar
 
Jun 2003
The Texas Hill Country

32×112 Posts
Default

You cannot. Although the server infrastructure supports multiple projects, we generally work (publicly) on only one at a time. When projects require a month or more to collect an adequate number of relations, it is better to get one finished and then start the next.

However, in the present case, the numbers selected are "easy" (in comparison to those that we have been factoring). We could sieve each of them in a week. However, there are two problems in doing that. First, there is the immediate post-processing. I am just now completing the filtering for 11,199- and we are still working on 3,491+ to try to get a more reasonable matrix. Therefore, I don't need another one to finish just yet.

But there is also the efficiency aspect. For every new factorization, each siever must compute the factor base. That "cost" is then amoritized over the lines sieved. Therefore it is more efficient to allow each siever to run for a couple of weeks on one number and ignore the other one.

There is also a problem in collecting all of the work done. Most results are returned within a few hours. However, some routinely get delayed for a day or so. A few even show up weeks after a project is completed. Obviously we cannot know when, if ever, delayed results will show up. Therefore, some of those results, particularly near the end of a project, will not be used. This is more significant when the projects are very short.

For these reasons, I chose to split the sievers into two camps and work on a separate factorization in each one.

Richard
Wacky is offline  
Old 2004-09-16, 09:00   #35
Death
 
Death's Avatar
 
Mar 2004
Ukraine, Kiev

2×23 Posts
Default

well, who choose what project will be assigned to user?

if I delete current project and try to connect to server it gives me the same project or other one??
Death is offline  
Old 2004-09-16, 11:30   #36
junky
 
junky's Avatar
 
Jan 2004

7×19 Posts
Default

Death, i've some machines running 2_657P and some others running 7_233P. I was wondering that, cause that would be great if all my machines was on the same projects. And when a new computer join the projects, it could be in the other projects.
Both projects seems to go fast enuf, so both should be complete in short time, like Wacky said.
junky is offline  
Old 2004-10-01, 03:19   #37
junky
 
junky's Avatar
 
Jan 2004

7×19 Posts
Default

hi guys,
i had 2 machines which were running into 7_233P and they "seems" to be moved to 6_257P.
i was running the project 6_257P_2, i take a look into my Client, to see if everything goes well, and at my great surprise, i saw i was now running the 5_304P_3's project.
So that's really cool.
Is that mean that both project has been completed?
Is there any estimation of the time required for this project now ?
Thanks

Last fiddled with by junky on 2004-10-01 at 03:20
junky is offline  
Old 2004-10-01, 08:45   #38
xilman
Bamboozled!
 
xilman's Avatar
 
"π’‰Ίπ’ŒŒπ’‡·π’†·π’€­"
May 2003
Down not across

47·229 Posts
Default

Quote:
Originally Posted by junky
hi guys,
i had 2 machines which were running into 7_233P and they "seems" to be moved to 6_257P.
i was running the project 6_257P_2, i take a look into my Client, to see if everything goes well, and at my great surprise, i saw i was now running the 5_304P_3's project.
So that's really cool.
Is that mean that both project has been completed?
Is there any estimation of the time required for this project now ?
Thanks
Status update:

Everyone is now running 5_304P, also known as the 172-digit composite factor of 5^304+1. Ignoring the various cockups at the start (5_304_1 and 5_304P_1 arose from typos I made ) there are two sub-projects to be sieved. 5_304P_3 is being done first and 5_304P_2 will follow. For those who understand the basics of line sieving, the first subproject is sieving the lowest 5 million lines at twice the length of the subsequent 20 million in second subproject.

Sieving for both 7_233P and 6_257P has completed. The matrix for the former is presently well underway on my home system and should complete next week. I had to pause the 7_233P matrix for a few hours yesterday to start the filtering and merging of the 6_257P relations.

The 6_257P data looks pretty good. Further processing will very likely mean the matrix will need pausing again, as the matrix is taking 546M memory out of the 768M fitted and the filter stages can themselves take a few hundred megs.

The current project is somewhat harder than its two predecessors. I'm not sure at the moment exactly how much harder but my guess is about 50% more. It will take a few weeks to sieve at current rates of progress.

Paul
xilman is offline  
 

Thread Tools


All times are UTC. The time now is 01:47.


Tue Jul 27 01:47:15 UTC 2021 up 3 days, 20:16, 0 users, load averages: 1.46, 1.76, 1.78

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, 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.