P-1 coordination thread - discussion

[glennpat]

I'm been running linux lately so have not been using Prime95 much. The newer Prime95 version 25.3 has changed some on the file names. From the whatsnew:

Several files have been renamed so that they are not changed by Windows
save/restore. Prime.ini is now prime.txt. Local.ini is now local.txt.
Worktodo.ini is now worktodo.txt. The primenet.ini file has been
deleted - it is now a section within the prime.txt file.

[Death]

hmmmm.... maybe I should check this out.

[hhh]

thank you.

and another question. I try to make p-1 for riesel, just to see some stats for my team.

and latest version of prime95 doesn't "continue" after makewdt.exe

what's wrong with it and what version do you use?

P-1 for Riesel will not be worth the CPU-cycles: They have sieved deeper than we have, and have even lower n.

H.

[Death]

This is make just for fun. I want to see team Ukraine in stats, don't care about place or scores, just presence. And also try to understand how to run p-1, this knowledge can be useful in other projects.

[hhh]

14485000-14490000 Cow_tipping [Reserved]

Your results are going to be useless, given the advance of firstpass, see here or here. A reservation above 15M will make more sense. H.

PS: Yes, I agree, somebody could clean up thread. H.

[Max Dettweiler]

Your results are going to be useless, given the advance of firstpass, see here or here. A reservation above 15M will make more sense. H.

PS: Yes, I agree, somebody could clean up thread. H.

Won't they still be useful for secondpass, though?

[hhh]

Won't they still be useful for secondpass, though?

Yes, but the whole purpose of P-1 is to save tests at a last stand, isn't it? H.

[Max Dettweiler]

Yes, but the whole purpose of P-1 is to save tests at a last stand, isn't it? H.

I would think that even if the number has been tested once, a factor for that number can still save us from having to do the secondpass test.

[hhh]

I would think that even if the number has been tested once, a factor for that number can still save us from having to do the secondpass test.

That's right. Yet, it will happen only in a few years, and by that time, there is a good chance that a prime will be found or that sieving would have found that factor. And then, P-1 is not superfast. By the time it takes to find this factor, one could have done this one LLR test; but not two.

So, if one would want to look for factors of Doublecheck numbers, one would do that right before doublecheck hits them, and then again, one would not do it at all, because you are better off with LLR directly.

I hope I look not too wise-assed now... H.

[balachmar]

Does mprime have to do the second pass tests as well, because on this machine I have it running for a while, but I didn't change the memory setting, so it doesn't do second pass tests. How bad is that?
On my other machine, I actually opted for a range that seems to be passed by PRP. But then it will be finished by now, so I will send the results in this evening.

[hhh]

Does mprime have to do the second pass tests as well, because on this machine I have it running for a while, but I didn't change the memory setting, so it doesn't do second pass tests. How bad is that?
On my other machine, I actually opted for a range that seems to be passed by PRP. But then it will be finished by now, so I will send the results in this evening.

Stage2 is needed to find factors. In your mprime directory, look for a bunch of files called (one letter)(long number), I hope you get what I mean. it could be something like
z5123875 or
f1329834

if they are there, you probably can just put in the stuff for worktodo.txt again, change your memory settings, and it will start to do all the stage2 for the tests. If that fails, it is proably better to drop that range and to do another one with larger memory settings, rather than redoing all the stage 1 work.

H.

[balachmar]

OK, thanks for the clarification. I know what you mean with the files.
I will copy the lot to another machine, since this machine is doing some other work at the moment. I hope that the other machine will be able to finish it. OK, I first tried to get it to run on this machine and that works.
So I will complete it normally.

[balachmar]

I'll post here not to trash the other thread, but my range resulted in one factor, not two!
14920000 14921000 balachmar 1 [complete]

[Joh14vers6]

I'll post here not to trash the other thread, but my range resulted in one factor, not two!
14920000 14921000 balachmar 1 [complete]

I can not see the factor in the stats. Is the factor big? Then post it here.

[glennpat]

I had posted that I had 2 factors in the range 15060000 15070000. Maybe it should of only really been 1. The factor 8058533177897161|24737*2^14910223+1 I turned in was new and saved to the data base, but I see today I did not show up in my stats. I looked in the stats for everyone and saw:

5189.734T 21181 15307772 3800310.858 Sat 12-Jul-2008 Mon 21-Jul-2008 35.000 PrimeGrid

Mine was found after the one PrimeGrid had found so it didn't really count for me.

[balachmar]

I can not see the factor in the stats. Is the factor big? Then post it here.

Too bad I only read this now. I don't think I have the values of the factor anymore...
And as far as I know the submission page accepted the factor...
mm strange...
Anyway, I'm getting a bit fed up with P-1 did a few like 40 lines of work and still no result... (As in points for the team and tests saved...)

[[DPC]Frentik]

Sorry if asked before but these are the steps I performed to do P-1 factoring and I want to make sure it's done properly

- Created account at www.seventeenorbust.com and joined my team there.
- downloaded latest prime95 version from www.mersenne.org
- reserved a range in the appropriate thread and saved it in the prime95 folder
- started prime95 and choose "join GIMPS" (then I had to fill in a username/email etc but i assume this is only used if you do mersenne prime finding? <- insecure on this part)
-stopped prime95, added the results.txt=fact.txt to the prime.ini file and restarted prime95.

It's running now and since its working on the first value from the worktodo.ini file I assume this part is correct so far.

Now when I find a factor I only have to submit it at www.seventeenorbust.com/sieve? And when the range is done, I'll post it in the coordination thread.

[Joe O]

If the factor is larger than 2^64 then you have to submit it here.

[Joh14vers6]

I switched over from prime v25.6 to v25.7.
Since I switched over I see an extra text at the result.txt.
With 25.6 I had results like:

[Fri Oct 10 20:25:04 2008]
21181*2^15854852+1 completed P-1, B1=130000, B2=2200000, Wd1: 9A15E8A8

And now with 25.7 I mostly get results like:

[Tue Oct 14 22:45:12 2008]
55459*2^16041334+1 completed P-1, B1=130000, B2=2200000, E=6, Wd1: 9F09E1E8

What does E=6 mean?

[opyrt]

I switched over from prime v25.6 to v25.7.
Since I switched over I see an extra text at the result.txt.
With 25.6 I had results like:

[Fri Oct 10 20:25:04 2008]
21181*2^15854852+1 completed P-1, B1=130000, B2=2200000, Wd1: 9A15E8A8

And now with 25.7 I mostly get results like:

[Tue Oct 14 22:45:12 2008]
55459*2^16041334+1 completed P-1, B1=130000, B2=2200000, E=6, Wd1: 9F09E1E8

What does E=6 mean?

I'm getting the same thing on my 32 bit client, but not on the 64 bit client.

Edit: I now see that this has been answered on mersenneforum: http://www.mersenneforum.org/showthread.php?t=10902

[Joh14vers6]

Today I found the factor:
P-1 found a factor in stage #2, B1=130000, B2=2200000.
21181*2^16496492+1 has a factor: 13825033371936547

But I will not get any credit for it because user 13796 (PrimeGrid) already found this factor.
I found that out here: http://www.seventeenorbust.com/sieve/results.txt.bz2
I searched for more tests with already found factors in my worktodo.txt and found another two already found factors by PrimeGrid. So I deleted them from the worktodo.txt.

How can I prevent PrimeGrid finding my factors first?

[shauge]

It is always annoying when not getting credit for a factor found. There is of course no way to prevent sieving finding factors in your factoring range. It is only possible check for found factors before starting factoring. The factor you post is 13.8P which is in a sieve range PrimeGrid did a some time ago and when I look in the "worktodo_16000000_16500000.zip" posted in the "P1 coordination thread", the test you post have been excluded. You should use an updated worktodo file.

[Joe O]

How can I prevent PrimeGrid finding my factors first?

By checking your worktodo.txt against results.txt on a daily basis and removing k n pairs from your worktodo.txt that have already been factored and posted in results.txt.

Edit:

You should use an updated worktodo file.

Thanks shauge. I wrote my answer before seeing your post. I did not consider that Joh14vers6 would have used an obsolete worktoto file. Actually, even my advice to check it daily would not always work. You are correct in saying that there is no way to prevent this from happening.

[shauge]

I can recommend the "make worktodo" site my team mate runesk made to get a worktodo file which eliminates the factors found in the current online results.txt.bz2 file and I will mention my own tool that checks a local worktodo.txt against a locally copied version of results.txt and displays already found factors (take care not to overwrite the prime95 output file "results.txt")

[Natenofs]

I just thought Id start a thread for those of us in the Northeast to start discussing routes and movements for the project. Can everyone who is interested in the project from the Northeast chime in? The main thread for this is under the general topics forum, cycling related ? not sure, but it starts with "Wouldnt it be cool if..."