mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Twin Prime Search (https://www.mersenneforum.org/forumdisplay.php?f=65)
-   -   Restart ;) (https://www.mersenneforum.org/showthread.php?t=13425)

 Svenie25 2010-05-21 17:28

Restart ;)

Hi Folks.

After around one year of being absent, I am back. ;)

I would like to restart my small project to find a prime @ n=666777. I know, it is a lot of work...

And now my question is, if someone please could give me some hints how to use tpsieve for the start.

Am I right, that I have to do a short sieve with NewPGen?

Any hint would be very helpfull.

Sven S.

 Oddball 2010-05-21 21:44

On my PC, NewPGen is faster than tpsieve for one n.

If you still want to use tpsieve, click the start menu, click "Run...", and then type in "cmd". The command prompt will come up, and you should type "cd [file that tpsieve is in]". Once that's done, type in:

tpsieve -i (name of file) -p 10e9* -P 20e9** -N 666777

*lower limit of range. One example is to use 10e9 to begin sieving from p=10G.
**upper limit of range. One example is to use 20e9 to finish sieving at p=20G.

By the way, would you be interested in testing n=390000? The testing times are faster, and you have a greater chance of finding a twin. Reservations are available here:

edit: Unlike NewPGen, tpsieve only gives you a list of factors; it doesn't remove them from the file. You need to remove those factors from the sieve file manually or write a separate program to do so.

 Puzzle-Peter 2010-08-05 19:38

[quote=Oddball;215669]
edit: Unlike NewPGen, tpsieve only gives you a list of factors; it doesn't remove them from the file. You need to remove those factors from the sieve file manually [B]or write a separate program to do so[/B].[/quote]

Don't know about Windows, but in Linux it's only a few shell commands.

Good luck, btw.

 mdettweiler 2010-08-05 22:21

[quote=Puzzle-Peter;224180]Don't know about Windows, but in Linux it's only a few shell commands.

Good luck, btw.[/quote]
I usually just use srfile (included with the [url=http://sites.google.com/srsieve]srsieve[/url] sieving program), which can remove factors from a NewPGen formatted file with the following command:

srfile -G -k factors.txt sieve.txt

That will output the sieve file with factors removed to t17_b2.prp. I've never tried it with a twin sieve file before; the exact output file name may be different, but it should work properly.

 kar_bon 2010-08-05 23:04

srfile can't handle candidate-files for twin-search!

If you want to sieve for twins with n=666777 and with kmin=1 (or 2) / kmax=1M using tpsieve, you need a file with kmax < pmin, so for this example sieve with NewPGen upto p=1M first.

If you got a multicore, use a file called "tpconfig.txt" and insert a line containig the number of threads. So for a Quad insert
[code]
[/code]

After NewPGen'ing you got a file (named say '666777.txt') like this
[code]
2000002:T:0:2:3
75 666777
165 666777
765 666777
1239 666777
1431 666777
...
[/code]

and now you can call tpsieve with
[code]
tpsieve -i 666777.txt -p 2e6 -P 1e9
[/code]

Here 'pmin' is the the sieve-depth of the NewPGen file in the header line.

Now tpsieve will sieve in 4 threads upto p=1e9. I've not yet tested/compared the timings for NewPGen and tpsieve.

To consider: tpsieve will find multiple factors of one candidate, NewPGen don't!

To remove the factors tpsieve will find (in a file called 'tpfactors.txt'), you have to use a script!

 All times are UTC. The time now is 00:08.