mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Conjectures 'R Us (https://www.mersenneforum.org/forumdisplay.php?f=81)
-   -   Riesel base 3 reservations/statuses/primes (https://www.mersenneforum.org/showthread.php?t=11151)

KEP 2008-09-27 10:11

[QUOTE=henryzz;143916]sounds good
what scripts do u use for this
do u just manually use srfile to remove ks from a file when llring or does a script do it for you[/QUOTE]

Well Michaf created a script, which test any even k in a range defined by a min-k and a max-k to a max-n. This script runs under WinPFGW and it PRP tests the entire range that you defines in the script file, before running the script.

The primes above, I just used the calculator to find, but once a range gets to n<=25000 I use a spreadsheet, to remove any k which can be reduced to a lower existing k at n<=25000 through a 3^q division. By the way, do you only LLR the primes or do you also verify the primes, using WinPFGW with the -tp flag in the command line?

Regards

Kenneth

henryzz 2008-09-27 12:01

i have been proving them with pfgw using -tp or -tc

i meant after pfgw has done its work and u are using srsieve and llr how do u remove ur ks from the sieved file do u have an automated way

3677878 and 4223272 are at 52k

KEP 2008-09-27 12:39

[QUOTE=henryzz;143922]i have been proving them with pfgw using -tp or -tc

i meant after pfgw has done its work and u are using srsieve and llr how do u remove ur ks from the sieved file do u have an automated way

3677878 and 4223272 are at 52k[/QUOTE]

Ups, my bad :smile:

After PFGW has done its work, I use this script in the first line of the NewPGen file that srfile creates from the output file srsieve creates. This line replaces the NewPGen coding that srfile creates using -G when completed:

ABC $a*3^$b-1 // {number_primes,$a,1}

$a represents the k-values and the coding between {} tells PFGW to only find one prime per k even if further n's should remain untested.

$b represents the n-value.

If a k is primed, PFGW still reads the k/n pairs, before jumping straight to the next k. I've found that searching the n<=25K is way easier this way, since LLR don't know how to skip testing of k's already primed :smile:

So to answer your question shortly, there really is no need to remove k's manually from the input file, since PFGW skips the tests itself aslong as it gets the above coding :smile:

But srfile does have a -d function which removes the k*b^n sequence that you have already primed. Sadly the coding that anon send me some time ago doesn't seem to work, since the '^' is not recognized by srfile, so in fact I've never really been succesfull in removing k sequences other ways than by hand :smile:

Regards

Kenneth

Ps. If this doesn't answered your question or there is something else you need to know or feels like wanting to know, please feel free to ask further questions, this is after all a very helpfull community :smile:

henryzz 2008-09-27 13:11

that answered it perfectly i didnt realize u didnt use llr for your second pass

KEP 2008-09-27 13:49

[QUOTE=henryzz;143929]that answered it perfectly i didnt realize u didnt use llr for your second pass[/QUOTE]

well I don't use LLR since it is 30 % faster (at least), but using the previously given script line to you, saves hundreds of tests that LLR has to do, which in the end will make PFGW faster than running LLR on these small k/n pairs.

Glad I could be of assistance :smile:

em99010pepe 2008-09-27 22:28

Within two or three days I would like to help you KEP. Can you send me work for 4 cores?

KEP 2008-09-27 23:37

[QUOTE=em99010pepe;143961]Within two or three days I would like to help you KEP. Can you send me work for 4 cores?[/QUOTE]

Glad to see your interest in the Riesel base 3. My hope and dream is that all k's can be tested to n<=25000 within the next 1435 days (1. september 2012). So thanks for whatever you decide to do for the effort :smile:

I'm not aware if you has Michafs script. However it would be need if you could figure out how many Million/Billion of k's you can/would like to take to n<=25000, then I'll just start my next ranges from where you tells me you will stop working. Please notice that the least availeable k to PRP test today is k=500,000,002 and above :smile:

Its late here, so in case of false answer or confusion, please feel free to ask again :smile:

KEP

em99010pepe 2008-09-28 08:36

I'll put it in a simple matter, I want work for a week to be done on a Q6600. You decide what to run...

KEP 2008-09-28 17:46

[QUOTE=em99010pepe;143987]I'll put it in a simple matter, I want work for a week to be done on a Q6600. You decide what to run...[/QUOTE]

Then just wait untill Gary has prepared the mini-drive for Riesel base 3, because I really doesn't has any tasks on this stage and with what I'm doing, that can be completed in under a week. But in case I get work that can be done in less than a week, I will send it to you on e-mail :)

KEP

em99010pepe 2008-09-28 18:12

[quote=KEP;144001]Then just wait untill Gary has prepared the mini-drive for Riesel base 3, because I really doesn't has any tasks on this stage and with what I'm doing, that can be completed in under a week. But in case I get work that can be done in less than a week, I will send it to you on e-mail :)

KEP[/quote]

When I am done with Sierp base 3 reservation I want to boost your work. You can't have all the credits for yourself...lol

em99010pepe 2008-09-28 20:53

KEP,

Got your email but I thought I was only going to PRP the candidates like the mini-drive and not generation the file and sieve. I'll stay with Sierp Base 6 mini-drive.

Carlos


All times are UTC. The time now is 00:55.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.