mersenneforum.org (https://www.mersenneforum.org/index.php)
-   15k Search (https://www.mersenneforum.org/forumdisplay.php?f=16)
-   -   stats on 15k (https://www.mersenneforum.org/showthread.php?t=1893)

 jocelynl 2004-01-11 05:56

stats on 15k

Hi all,

I just updated the stats. I have tested small n`s with trial factoring to test primality, and found quite a few. Our newest best candidate is 2995125705 with 115 prime found to (200000)
and a curve 10^((0.1364*i+0.1019)^(3/5))

I think we should stop 210885 at (450000) or (500000)
since it hasn't been producing much lately.
We should take 2995125705 to (400000) and see how it goes.
The problem remains the sieving!
What should we do?

Joss L15.

 SlashDude 2004-01-11 11:56

I agree :)

I am planning on stopping 210885 at n=450,000. I have already started sieving 2995125705 for n=200,000 to 1,000,000.

Can someone send me ksieve? I can't get to fatphil's web site....I get the 403 error cause of my location...

Thanks!

 Thomas11 2004-01-11 14:34

Slashdude,

I have sent you an email containing the Linux and Windows binaries of Ksieve.

-- Thomas.

 jocelynl 2004-01-12 00:24

Slashdude will you make it available on 15k.org ?

Joss

 Kosmaj 2004-01-13 14:24

It sounds reasonable to try another k, but it will be nice if people who reserved ranges of k=210885 almost a month ago tell us their status, and when can we expect them to return results. There are still chances that 210885 is a good candidate if, for example, ET_ finds a prime in the 355-360K range and we find another one for n<450,000. OTOH, if there is nothing all the way to 440K (my range), then it looks like a large gap (the last prime was at 315K by SlashDude), and the next prime should be imminent (?) Since n=440K means 132,000 digits maybe we should continue until we find it. BTW, in my 435-440 range there is nothing up to 439,000. It will be completed in 1-2 days.

As for ksieve, it can work with a single k, but I heard it's more efficient when sieving several k's at the same time. I also wonder is it possible to switch from NewPgen to ksieve when sieving is in progress (say, sieved to 100bn).

 SlashDude 2004-01-13 15:45

[QUOTE][i]Originally posted by jocelynl [/i]
[B]Slashdude will you make it available on 15k.org ?

Joss [/B][/QUOTE]

If you can't download it from [url]http://fatphil.org/maths/ksieve/[/url] - Private message me, and I'll send you a link to the ksiece2bin.tgz file on [url]www.15k.org[/url] (Unless Phil doesn't want us to make his software available without using his web site - Phil, if you want me to remove the ksieve file, let me know and I will remove the easy to guess link. :)

[QUOTE][i]Originally posted by Kosmaj [/i]
[B]
As for ksieve, it can work with a single k, but I heard it's more efficient when sieving several k's at the same time. I also wonder is it possible to switch from NewPgen to ksieve when sieving is in progress (say, sieved to 100bn). [/B][/QUOTE]

I'm sure I can merge the file I already have from NwePGen with the results from Ksieve! I have sieved 2995125705 through p=310 billion. I am working on testing ksieve, and so far, everything looks good!

With the larger k's, ksieve is by far faster then NewPGen! I havenâ€™t tested it with multiple k's, but the single k ksieve for large k's (k>2^31) is by far faster - A P4 2.6GHz machine sieving k=2305225065 completed 40 billion p's in about 8 hours!

If I have time, I'll look at the timing for multiple k's. Right now, using single k ksieve is a huge speed increase!! I should have the group sieve for 2995125705 setup soon!

Thanks again everyone for your help!!!
-Mark

 All times are UTC. The time now is 20:45.