76M - 77M to 2^60

Boulder · 2003-05-26, 18:17

Joining the club ;)

Hopefully no one's doing this range, I'll probably move on to 2^61 as soon as the batch is done.

guido72 · 2003-05-27, 00:40

Quote:

Originally Posted by Boulder

Joining the club ;)

Hopefully no one's doing this range, I'll probably move on to 2^61 as soon as the batch is done.

Welcome!
I didn't understand what you mean, precisely... Did you get these huge exponents from Primenet? (Don't think so...) Or do you like to LL-test (or factoring just to 2^60) some of them?
There are few crazy People doing some of these dirty jobs, actually. They asked George for them...
From undoc.txt file:
"You can limit how far the program tries to factor a number. This feature
should not be used with the Primenet server. To prevent the factoring code from factoring higher than 2^n, enter this in prime.ini: FactorOverride=n"
Regards

Boulder · 2003-05-27, 05:28

Quote:

Originally Posted by guido72

Welcome!
I didn't understand what you mean, precisely... Did you get these huge exponents from Primenet? (Don't think so...) Or do you like to LL-test (or factoring just to 2^60) some of them?
There are few crazy People doing some of these dirty jobs, actually. They asked George for them...
From undoc.txt file:
"You can limit how far the program tries to factor a number. This feature
should not be used with the Primenet server. To prevent the factoring code from factoring higher than 2^n, enter this in prime.ini: FactorOverride=n"
Regards

I've been reading the threads for a while and decided to join the factoring effort. I was off the GIMPS for some time and now that I've built a decent P4 system, I decided to get on the train again. I just noticed that the range was not factored beyond 2^59 and wanted to give it a shot just for the hell of it.

Do I have to email George directly and ask him for the permission or is it okay to do the job and let people know here in the forum? I know that some of the people have asked for a permission but as this range is not available via PrimeNet, it probably won't hurt to crunch it.

markr · 2003-05-27, 05:57

Welcome to the club, Boulder!

You are correct: you don't have to ask George for any range as long as it's outside Primenet. Posting here is sufficient. (But send George your results.txt, of course!) Generally someone will correct any inadvertent overlapping ranges or whatever. I see garo has included your selection in the sticky topic listing the ranges currently out, so your choice is okay.

Boulder · 2003-05-27, 06:41

I should have a decent amount of results by the end of the week and I'll send them to George probably on Sunday. I'm thinking about going through 76-76.1M to 2^61 as soon as this batch is done to see how much time it would take to factor the whole 76-77M range to 2^61.

But that will have to wait and I'll post my progress here and email the results to George, of course.

garo · 2003-05-27, 07:07

Boulder,
Welcome to the effort. But one caveat. Though I know it's your computer and you can do as you wish with it please do not use a P4 for factoring under 63 bits. P4s are really bad at this. My Athlon 1333 beats a P4 2.53 If you have a P4 use it for first time LL tests or doublechecking tests.

Boulder · 2003-05-27, 07:26

Quote:

Originally Posted by garo

Boulder,
Welcome to the effort. But one caveat. Though I know it's your computer and you can do as you wish with it please do not use a P4 for factoring under 63 bits. P4s are really bad at this. My Athlon 1333 beats a P4 2.53 If you have a P4 use it for first time LL tests or doublechecking tests.

Hmm, I didn't know that. Are there any LMH factoring jobs available that would be more useful to the project?

EDIT: would it be useful to do P-1 factoring on some range and how would I do that?

garo · 2003-05-27, 19:59

There aren't any factoring jobs with LMH that would be really suitable for a P4. LMH by definition is for people with "relatively" slower computers who still want to participate.

P-1 factoring is slightly more complex. It is usually done after a number has been tested using trial factoring to the appropriate limit (67 bits for 22M exponents) and as such only available via Primenet. Are there any specific reasons why you don't want to use Primenet? If so, let us know and we will suggest what may be appropriate.
Garo

PS: If you choose to factor on Primenet (ie the 22M range and thereabouts most of the time will be spent on factoring above 64 bits for which the P4 is great! However, you cannot just choose numbers to factor in this range and must contact primenet for it to assign numbers to you.)

trif · 2003-05-27, 20:21

Certainly, someone could make a project to factor a range to 64 bits, and then hand them over to the P4's. It's quite possible for a P4 owner to have the factoring bug. :)

trif · 2003-05-27, 20:32

As an example, markr is currently engaged in taking the 33.6M range to 64 bits. This is the next range destined to go to Primenet for LL testing in the 10M digit range, but these go so slowly that there might well be enough time to factor them completely to 67 bits before they go to Primenet, although it would need more than one P4. Other P4 owners with the factoring bug might be interested.

cheesehead · 2003-05-27, 20:53

Quote:

Originally Posted by Boulder

would it be useful to do P-1 factoring on some range and how would I do that?

You don't have to go through Primenet to do P-1 factoring.

Just manually add the

[code:1]Pminus1=eeeeeeee,B1,B2,0,0[/code:1]
command to your worktodo.ini file (where eeeeeeee is the exponent, and B1 and B2 are the B1 and B2 bounds for the P-1 run). Example: Pminus1=13453547,150000,2100000,0,0

Then you can cut-n-paste your results from results.txt to the manual entry area on the Primenet manual test report page.

The current PMINUS1.TXT file will tell you which unfactored Mnumbers have already been P-1ed and what the highest B1/B2 were for them. (Note that for P-1, "highest B1/B2" means highest B1 or, if the B1 for two different runs is the same, then highest B2. So a 11000,55000 run would be considered "higher" than a 10000,990000 run or a 11000,44000 run, but lower than a 11000,66000 run or a 12500,12500 run.)

However, absence of an exponent in the PMINUS1.TXT file does not necessarily mean it needs P-1 factoring! It might already have been factored. Whenever a Mnumber is factored by any means, it is removed from the PMINUS1.TXT file (and nofactor.cmp). So be sure to check the factors.cmp file to see whether the Mnumber you're considering has already been factored!!

Of course, take precautions to avoid barging in on an exponent already being worked-on by someone else. Check the current assignments report on the Primenet status page to make sure the exponent you want to P-1 isn't currently assigned. Check the lucas_v.txt file to see whether the exponent has been L-L double-checked (if it's already been verified, no one else is likely to be working on it), and stay out of the ranges open for assignment on Primenet.

Also, it's best to avoid Mnumbers that someone else has just recently L-Led (i.e., LL-completed Mnumbers not far behind the ranges currently being assigned by Primenet). If your P-1 run factors a Mnumber, then whoever has L-Led it will lose GIMPS (but not Primenet) credit for that L-L work. So let folks enjoy their L-L credits for a while before running P-1 on those.

Then post your intended choices here in order to coordinate, and so someone can warn you off if necessary.

2003-05-26, 18:17	#1
Boulder May 2003 3·13 Posts	76M - 77M to 2^60 Joining the club ;) Hopefully no one's doing this range, I'll probably move on to 2^61 as soon as the batch is done.

2003-05-27, 05:57	#4
markr "Mark" Feb 2003 Sydney 3·191 Posts	Welcome to the club, Boulder! You are correct: you don't have to ask George for any range as long as it's outside Primenet. Posting here is sufficient. (But send George your results.txt, of course!) Generally someone will correct any inadvertent overlapping ranges or whatever. I see garo has included your selection in the sticky topic listing the ranges currently out, so your choice is okay.

2003-05-27, 06:41	#5
Boulder May 2003 3·13 Posts	I should have a decent amount of results by the end of the week and I'll send them to George probably on Sunday. I'm thinking about going through 76-76.1M to 2^61 as soon as this batch is done to see how much time it would take to factor the whole 76-77M range to 2^61. But that will have to wait and I'll post my progress here and email the results to George, of course.

2003-05-27, 07:07	#6
garo Aug 2002 Termonfeckin, IE 2⁴×173 Posts	Boulder, Welcome to the effort. But one caveat. Though I know it's your computer and you can do as you wish with it please do not use a P4 for factoring under 63 bits. P4s are really bad at this. My Athlon 1333 beats a P4 2.53 If you have a P4 use it for first time LL tests or doublechecking tests.

2003-05-27, 19:59	#8
garo Aug 2002 Termonfeckin, IE 2⁴·173 Posts	There aren't any factoring jobs with LMH that would be really suitable for a P4. LMH by definition is for people with "relatively" slower computers who still want to participate. P-1 factoring is slightly more complex. It is usually done after a number has been tested using trial factoring to the appropriate limit (67 bits for 22M exponents) and as such only available via Primenet. Are there any specific reasons why you don't want to use Primenet? If so, let us know and we will suggest what may be appropriate. Garo PS: If you choose to factor on Primenet (ie the 22M range and thereabouts most of the time will be spent on factoring above 64 bits for which the P4 is great! However, you cannot just choose numbers to factor in this range and must contact primenet for it to assign numbers to you.)

2003-05-27, 20:21	#9
trif Aug 2002 CA₁₆ Posts	Certainly, someone could make a project to factor a range to 64 bits, and then hand them over to the P4's. It's quite possible for a P4 owner to have the factoring bug. :)

2003-05-27, 20:32	#10
trif Aug 2002 2·101 Posts	As an example, markr is currently engaged in taking the 33.6M range to 64 bits. This is the next range destined to go to Primenet for LL testing in the 10M digit range, but these go so slowly that there might well be enough time to factor them completely to 67 bits before they go to Primenet, although it would need more than one P4. Other P4 owners with the factoring bug might be interested.