n=390000 discussions and old reservations

TimSorbet · 2010-05-03, 22:34

Not sure if interest in this will continue, but I'll help a bit.

Reserving k=500k-1M. This is a rather small and easy chunk of work. One core of my i5, taking 0.640 ms per iteration, (or 4 minutes and 9.6 seconds per candidate) should finish all 262 candidates in about 18 hours and 10 minutes. A multiple-million reservation would be a more typically-sized piece of work for most people.

TimSorbet · 2010-05-04, 18:41

This file is too undersieved to do primality testing already. With NewPGen (I'm running an i5 on 32-bit XP; tpsieve with one thread is around half the speed) I can eliminate about one k every 2 seconds compared to one LLR test every 250 seconds. Unless I'm mistaken, (or shouldn't be ignoring things like accounting for double checks) that means that the sieve depth needs roughly 7 doublings (2 with 7 doublings is 2*2^7=256, just about 250) before it's optimally sieved for 32-bit (if you can get 64-bit sievers, sieve speed might approximately double, which might mean about another doubling). That puts it at roughly 400-450T (3.33*2^7=426.24) before primality tests ought to be run on this range.

I'm not putting a serious effort to sieving this at the moment, but I sieved it to 3.38T (107 factors in 267 seconds) with NewPGen. The new file is attached. (you might need to rename it to remove the ".zip" before it'll open, as the forum doesn't like .7z but only .7z will compress to under the attachment limit)

Oddball · 2010-05-05, 04:57

Hmmm, it looks like I can't edit the first post. Maybe that's because I'm not a mod. But anyway, I'm reserving 1M-1.2M. So far, we have:

Code:

Range		Reserved by		Status		Primes found
0.0M-0.3M	Oddball			complete	0
0.3M-0.5M	Oddball			complete	0
0.5M-1.0M	Mini-Geek		complete	0
1.0M-1.2M	Oddball			in progress	?

I'm planning to sieve more, so that range might not get done for some time.

gribozavr · 2010-05-06, 17:45

My Core i5 can help sieving with all 4 cores (running 64-bit Linux of course). I just want to ensure that neither my nor anyone else's CPU hours would be wasted.

Oddball · 2010-05-06, 20:51

Quote:

Originally Posted by gribozavr

My Core i5 can help sieving with all 4 cores (running 64-bit Linux of course). I just want to ensure that neither my nor anyone else's CPU hours would be wasted.

Thanks for offering to help out!

Right now, just sieve 10G-150G to 8.1T. When you're done, merge that file into the 1M-10G file (which is also sieved to 8.1T) uploaded here:

http://www.sendspace.com/file/dfee42

You can then sieve the whole 1M-150G file at once from that point on. I couldn't do that because my PC didn't have enough RAM, but the RAM that usually comes with a core i5 should be enough to do the job.

I'll keep the old system for the credits the same. If/when a twin is found, the finder shares the credit with both the top siever and the top LLR tester.

Oddball · 2010-05-06, 20:55

Those of you who want to jump right into LLR testing can start by downloading the attached file. It contains k=1M-10M, sieved to 8.1T. There are about 470 candidates per 1M.

gribozavr · 2010-05-07, 10:29

Quote:

Originally Posted by Oddball

Right now, just sieve 10G-150G to 8.1T.

Could you point me to the file? I don't seem to see it.

TimSorbet · 2010-05-07, 11:46

Quote:

Originally Posted by gribozavr

Could you point me to the file? I don't seem to see it.

I don't think it exists yet, you'll have to start it from the beginning.
Apparently Oddball decided to change the scope of this. The only older files posted were for k=200k-200M, and now he's talking about 1M-10G and 10G-150G.

Historian · 2010-05-07, 15:16

I haven't been on mersenneforum for quite some time, but I'll give this project a try. Reserving 1.2M-1.4M.

Quote:

Originally Posted by Mini-Geek

Apparently Oddball decided to change the scope of this. The only older files posted were for k=200k-200M, and now he's talking about 1M-10G and 10G-150G.

Another possibility could be that 1-10G was sieved, but he only chose to release 200k-200M because the whole 1-10G sieve file would take too long to download. As far as I can tell, he mentioned 1-150G all along, but he only sieved 1-10G because he lacked the resources to do so.

From: http://www.mersenneforum.org/showthread.php?t=13375

Quote:

I'm proposing a new goal: test n=390,000 from k=1-150G.

Oddball · 2010-05-08, 09:01

Quote:

Originally Posted by Historian

I haven't been on mersenneforum for quite some time, but I'll give this project a try. Reserving 1.2M-1.4M.

Welcome to TPS!

Quote:

Another possibility could be that 1-10G was sieved, but he only chose to release 200k-200M because the whole 1-10G sieve file would take too long to download. As far as I can tell, he mentioned 1-150G all along, but he only sieved 1-10G because he lacked the resources to do so.

That is correct.

Anyway, I'm done with my old range (results attached). No primes were found, so I'm getting 1.4M-1.7M. Here's the current status:

Code:

Range		Reserved by		Status		Primes found
0.0M-0.3M	Oddball			complete	0
0.3M-0.5M	Oddball			complete	0
0.5M-1.0M	Mini-Geek		complete	0
1.0M-1.2M	Oddball			complete	0
1.2M-1.4M	Historian		complete	0
1.4M-1.7M	Oddball			in progress	?
1.7M+		available

kar_bon · 2010-05-08, 09:13

Looking at the First Twin k (see bottom, first graph) the estimated k-value for n=333333 using the formula k=0.2784*n^1.9652 is k=19871455920.
The real values was k=65516468355.

So for n=390000 it's estimated for k=27053869866 or about 27G.

2010-05-03, 22:34	#1
TimSorbet Account Deleted "Tim Sorbera" Aug 2006 San Antonio, TX USA 11·389 Posts	n=390000 discussions and old reservations Not sure if interest in this will continue, but I'll help a bit. Reserving k=500k-1M. This is a rather small and easy chunk of work. One core of my i5, taking 0.640 ms per iteration, (or 4 minutes and 9.6 seconds per candidate) should finish all 262 candidates in about 18 hours and 10 minutes. A multiple-million reservation would be a more typically-sized piece of work for most people. Last fiddled with by Oddball on 2010-05-24 at 00:07 Reason: changing title

2010-05-05, 04:57	#3
Oddball May 2010 499₁₀ Posts	Hmmm, it looks like I can't edit the first post. Maybe that's because I'm not a mod. But anyway, I'm reserving 1M-1.2M. So far, we have: Code: Range Reserved by Status Primes found 0.0M-0.3M Oddball complete 0 0.3M-0.5M Oddball complete 0 0.5M-1.0M Mini-Geek complete 0 1.0M-1.2M Oddball in progress ? I'm planning to sieve more, so that range might not get done for some time. Last fiddled with by Oddball on 2010-05-05 at 05:11 Reason: typo

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Soapbox Discussions	only_human	Soap Box	41	2019-11-16 15:46
Archived sieve reservations and discussions	MooooMoo	Twin Prime Search	735	2011-12-09 16:33
Range reservation thread (n=390000)	Oddball	Twin Prime Search	8	2010-07-27 17:54
Primegrid discussions	pacionet	Twin Prime Search	17	2007-01-20 11:22
Automated PRP discussions	ltd	Sierpinski/Riesel Base 5	20	2006-09-02 22:19

2010-05-06, 17:45	#4
gribozavr Mar 2005 Internet; Ukraine, Kiev 110010111₂ Posts	My Core i5 can help sieving with all 4 cores (running 64-bit Linux of course). I just want to ensure that neither my nor anyone else's CPU hours would be wasted.

2010-05-08, 09:13	#11
kar_bon Mar 2006 Germany 2,999 Posts	Looking at the First Twin k (see bottom, first graph) the estimated k-value for n=333333 using the formula k=0.2784*n^1.9652 is k=19871455920. The real values was k=65516468355. So for n=390000 it's estimated for k=27053869866 or about 27G.