mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   FactorDB (https://www.mersenneforum.org/forumdisplay.php?f=94)
-   -   yafu.pl Usage Currently (https://www.mersenneforum.org/showthread.php?t=21556)

EdH 2016-09-07 02:41

yafu.pl Usage Currently
 
Are there many users working on the composites in the db currently? I'm just bringing some of my computers back on line and looking for a direction for the short term. I was thinking of running yafu.pl on some of them for a bit. Is there a thread dedicated to this area? Is there something I should do to avoid "stepping" on others' work? Is the randomness of the script enough to make that a rarity?

Would working on Primo certificates be of more use than composites?

henryzz 2016-09-07 09:53

[QUOTE=EdH;441793]Are there many users working on the composites in the db currently? I'm just bringing some of my computers back on line and looking for a direction for the short term. I was thinking of running yafu.pl on some of them for a bit. Is there a thread dedicated to this area? Is there something I should do to avoid "stepping" on others' work? Is the randomness of the script enough to make that a rarity?

Would working on Primo certificates be of more use than composites?[/QUOTE]

Does the database accept the latest Primo certificates yet?

EdH 2016-09-07 14:49

[QUOTE=henryzz;441813]Does the database accept the latest Primo certificates yet?[/QUOTE]
Well, since the published version of yafu.pl won't retrieve composites for me and Primo has updated past usefulness with the db, I guess I'll go play somewhere else...

chris2be8 2016-09-07 16:07

I'm running scripts to the same effect as yafu.pl (they use my GPU for ecm and polynomial selection stage 1, which complicates them) on a couple of old PCs. But I'm barely keeping up with the rate numbers are added to factordb so would be very glad for some help.

There's enough randomness in yafu.pl to make collisions rare enough not to matter.

Chris

henryzz 2016-09-07 16:58

[QUOTE=EdH;441833]Well, since the published version of yafu.pl won't retrieve composites for me and Primo has updated past usefulness with the db, I guess I'll go play somewhere else...[/QUOTE]
You do need to switch the script to factordb.com rather than the old address.

Puzzle-Peter 2016-09-07 18:41

[QUOTE=EdH;441833]Well, since the published version of yafu.pl won't retrieve composites for me and Primo has updated past usefulness with the db, I guess I'll go play somewhere else...[/QUOTE]

What about using an older primo version? 4.1.1 and older are working fine and factordb can read the certificates.

Antonio 2016-09-07 19:07

[QUOTE=EdH;441793]Are there many users working on the composites in the db currently? I'm just bringing some of my computers back on line and looking for a direction for the short term. I was thinking of running yafu.pl on some of them for a bit. Is there a thread dedicated to this area? Is there something I should do to avoid "stepping" on others' work? Is the randomness of the script enough to make that a rarity?

Would working on Primo certificates be of more use than composites?[/QUOTE]

You could always try the perl script mentioned here [URL]http://mersenneforum.org/showpost.php?p=401629&postcount=38[/URL], it will give you some control over the composites you work on.

EdH 2016-09-08 04:19

[QUOTE=chris2be8;441841]I'm running scripts to the same effect as yafu.pl (they use my GPU for ecm and polynomial selection stage 1, which complicates them) on a couple of old PCs. But I'm barely keeping up with the rate numbers are added to factordb so would be very glad for some help.

There's enough randomness in yafu.pl to make collisions rare enough not to matter.

Chris[/QUOTE]
I will run yafu.pl for a while and see if I can help knock down some composites. My machines are not yet up fully and not on 24/7 for now, but they still should help some.
[QUOTE=henryzz;441844]You do need to switch the script to factordb.com rather than the old address.[/QUOTE]
That was the first thing I did yesterday and it still wouldn't run. Oddly, though, tonight it worked flawlessly. I did have to remember to change both instances.
[QUOTE=Puzzle-Peter;441857]What about using an older primo version? 4.1.1 and older are working fine and factordb can read the certificates.[/QUOTE]
I thought of that and some of my machines that aren't yet back on are all set with earlier versions. Maybe when I bring them back up. At least for now, I have yafu.pl running.
[QUOTE=Antonio;441859]You could always try the perl script mentioned here [URL]http://mersenneforum.org/showpost.php?p=401629&postcount=38[/URL], it will give you some control over the composites you work on.[/QUOTE]
Thanks. For now I'll run yafu.pl, but look at yours and maybe swap when I have a chance.

Thanks everyone.

EdH 2016-09-08 22:15

Hmmm, I turn several machines toward composite factoring via yafu.pl and the composite list [B]increases[/B] by 100k! Am I running something backwards? Ok, let's bring some more on line and see if that causes an even larger increase...

pinhodecarlos 2016-09-09 01:11

Could you post a step by step guide how to run the Yafu.pl with links to programs?
Thank you in advance.
Carlos

EdH 2016-09-09 02:41

[QUOTE=pinhodecarlos;441985]Could you post a step by step guide how to run the Yafu.pl with links to programs?
Thank you in advance.
Carlos[/QUOTE]
All of the following is based on an Ubuntu and a Debian system and uses a terminal:

First, do you currently have a working version of YAFU on your system?

If, yes, then what I have done is make a directory for the work, which I have called yafupl, and placed a copy of the yafu binary and the yafu.ini file in the yafupl directory. I then copied the perl code from [URL="https://www.rechenkraft.net/wiki/index.php?title=Benutzer_Diskussion:Yoyo/factordb#yafu.pl"]here[/URL], and placed it in a file called yafu.pl in the same directory.

You might want to change the "-threads 8" to another value, or remove it altogether, as I have, to let your yafu.ini file "threads=" entry prevail. Make sure you leave the "|" character there. You may also want to change the "mindig=" from 10 to something higher than 69, since the db is taking care of <70d composites, if I understand correctly. Based on my next post, you may even want to work higher.

I then cd to the yafupl directory and issue the command:
[code]
perl yafu.pl
[/code]If you get an error about LWP missing from perl, you might want to look at [URL="http://lwp.interglacial.com/ch01_03.htm"]this page[/URL], unless someone else chimes in with better info.

As an alternative, you can look at the script mentioned a few posts earlier by Antonio. I will be looking at the possibility of switching over to it later on.


All times are UTC. The time now is 12:29.

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