How to only do Trial Factoring? - Page 2

michael · 2004-01-05, 10:48

My LL-test will be done in a few days, my worktodo.ini file got an extra line: Factor=24347537,60

Does this mean it will only factor up to 60bits? seems a bit low for that kind of number, i would at least expect 67 or 68...perhaps it updates during the factoring, i didn't take a close look on my previous exponent though it says Test=22063543,67,1 which makes me think it trialfactored up to 67 bits...

If i want to factor to say 72 bits, do i just change the line there(worktodo.ini)? Or do i need to change something in the prime.ini file?

-michael

S00113 · 2004-01-05, 10:57

Quote:

Originally posted by michael
My LL-test will be done in a few days, my worktodo.ini file got an extra line: Factor=24347537,60

Does this mean it will only factor up to 60bits?

It means that someone already tried all possible factors up ti 2^60. You will start at 2^60 and work up to 67 or 68, depending on CPU and if the number has been LL tested before.

Quote:

If i want to factor to say 72 bits, do i just change the line there(worktodo.ini)? Or do i need to change something in the prime.ini file?

You need to set FactorOverride prime.ini, but don't do that if you use PrimeNet. Read about FactorOverride in undoc.txt.

michael · 2004-01-05, 11:40

undoc.txt tells me:
-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

This doesn't tell me what i should use to be able to trialfactor higher...can anyone help me out on that?

-michael

lycorn · 2004-01-05, 12:19

That feature allows you to set an upper limit for Trial Factoring. I am not sure how far you can go, i.e. if it allows you to set a limit higher than the default for that particular range (although I think does). It is normally used to set a lower limit.
Now two important things:

1. NEVER use that feature for exponents you got from Primenet. If the exponent was assigned by the server, either you allow it to go all the way to the default limit, or you simply unreserve it, if you don´t want it anymore (losing the work done to that point). Manually limiting the upper level confuses the server and leads to the "stuck exponents" discussed in other threads of the forum.

2. It appears that you are interested in Trial Factoring. You may wish to read about Lone Mersenne Hunters, to get exclusively that type of work. LMH works outside Primenet ranges, so it is safe to use the FactorOverride feature, and that´s in fact what is normally done, as people in LMH use to get large ranges of exponents and then TF them to a depth of 1 or 2 bits more. Now, if you have a P4 machine, I don´t recommend you join LMH, for most of the time you would be doing TF below 64 bits, which is a waste for P4s, as the SSE2 instructions won´t be used below that level. Of course you could get just a few large exponents and TF them to the default limit, which can be as high as 72 bits.

From the point of view of project benefit, though, it would be better if you stick with Primenet, getting Trial Factoring assignments for the 23/24M range (those will be the next ranges in Primenet, and many exponents will be done by P4s, so raising the TF to the default 67 bit limit would spare P4s calculating the lower levels. (Unless you also own a P4, oh God...

)

The two really important things to retain are:
Don´t use FactorOverride with Primenet, and avoid using a P4 to systematically TF under 64bits.

HTH

michael · 2004-01-05, 18:54

Yes, i got a P4-2400.

When i look at the cleared exponents database i see some numbers even trialfactored up to 100 bits...i have no intention to do this, but i wonder ...

I like that idea of LMH too, there is alot of trialfactoring i could do 64-74 bitwise (as i understand P4's only are an advantage starting from 64 bits). The program they asked me to download fails to run on my computer though, no clue why or what i'm doing wrong

In the end i guess it's best to just stick to the exponents primenet provides me...

-michael

nfortino · 2004-01-05, 18:59

Quote:

Originally posted by michael
Yes, i got a P4-2400.

When i look at the cleared exponents database i see some numbers even trialfactored up to 100 bits...i have no intention to do this, but i wonder ...
-michael

The exponents you see in cleared are not trial factored up to 100 bits (that would take forever with current exponents), instead, those are large factors found with the p-1 test.

lycorn · 2004-01-06, 00:15

Quote:

Originally posted by michael

The program they asked me to download fails to run on my computer though, no clue why or what i'm doing wrong

Which program you mean? Decomp? That is the program used to extract exponents to test. What you have to do is downloading the file decomp.zip from GIMPS Status page, then unzip it. Download also nofactor.zip and unzip it to the same directory. The file you obtain, nofactor.cmp, is still in a compressed format, and that´s the role of decomp to decompress it. Decomp has to be run on a DOS window (Command prompt on Win XP). Having decomp.exe and nofactor.cmp in the same directory, you just have to type at the command prompt decomp -w xxxxxx yyyyyy, and will obtain a text file named nofactor with lines like Factor=<exponent to test> (i.e. in a form suitable to be copied and pasted to a worktodo.ini file), ranging from exponent xxxxxx to yyyyyy. Upon decompressing decomp.zip read the included readme file for complementary information. Don´t forget to uncheck the UsePrimenet... box under Test->Primenet on Prime95 window menu, and to tweak Primenet.ini as per the indications given on http://www.mersenneforum.org/showthr...=&threadid=379. It should work fine

michael · 2004-01-06, 00:24

Yes it's the decomp.exe that doesn't want to open...
I got all the files in 1 folder, when i double click decomp.exe the doswindow pops up for about a millisecond...then disappears.
I tried to get there through start/run/command but that wouldn't work out either.

-michael

Xyzzy · 2004-01-06, 00:30

As Lycorn said, you have to use the command line...

Quote:

How to decompress the factors and nofactors data.

1) Download and unzip one or both of factors.zip and nofactor.zip.
2) To get factors, run "decomp -f start_exponent end_exponent"
3) To get the k in 2kp+1, run "decomp -k start_exponent end_exponent"
4) To get the nofactor data, run "decomp -n start_exponent end_exponent"
5) To get the nofactor data with "Test=" in front of each line, run "decomp -t start_exponent end_exponent"
6) To get the nofactor data with "Factor=" in front of each line, run "decomp -w start_exponent end_exponent"

michael · 2004-01-06, 00:37

I thought commandline meant dosprompt ... what am i missing here?

-michael

Xyzzy · 2004-01-06, 00:40

In XP:

Start Menu -> Programs -> Accessories -> Command Prompt

If decomp.exe is on your desktop, type "cd Desktop"... Then go to work...

2004-01-05, 12:19	#15
lycorn "GIMFS" Sep 2002 Oeiras, Portugal 2×7×113 Posts	That feature allows you to set an upper limit for Trial Factoring. I am not sure how far you can go, i.e. if it allows you to set a limit higher than the default for that particular range (although I think does). It is normally used to set a lower limit. Now two important things: 1. NEVER use that feature for exponents you got from Primenet. If the exponent was assigned by the server, either you allow it to go all the way to the default limit, or you simply unreserve it, if you don´t want it anymore (losing the work done to that point). Manually limiting the upper level confuses the server and leads to the "stuck exponents" discussed in other threads of the forum. 2. It appears that you are interested in Trial Factoring. You may wish to read about Lone Mersenne Hunters, to get exclusively that type of work. LMH works outside Primenet ranges, so it is safe to use the FactorOverride feature, and that´s in fact what is normally done, as people in LMH use to get large ranges of exponents and then TF them to a depth of 1 or 2 bits more. Now, if you have a P4 machine, I don´t recommend you join LMH, for most of the time you would be doing TF below 64 bits, which is a waste for P4s, as the SSE2 instructions won´t be used below that level. Of course you could get just a few large exponents and TF them to the default limit, which can be as high as 72 bits. From the point of view of project benefit, though, it would be better if you stick with Primenet, getting Trial Factoring assignments for the 23/24M range (those will be the next ranges in Primenet, and many exponents will be done by P4s, so raising the TF to the default 67 bit limit would spare P4s calculating the lower levels. (Unless you also own a P4, oh God... ) The two really important things to retain are: Don´t use FactorOverride with Primenet, and avoid using a P4 to systematically TF under 64bits. HTH Last fiddled with by lycorn on 2004-01-05 at 12:25

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Trial Factoring on AMD/ATI GPU's?	Stargate38	GPU Computing	9	2018-08-31 07:58
What is Trial Factoring?	Unregistered	Information & Answers	5	2012-08-02 03:47
How much Trial Factoring to do?	odin	Software	4	2010-08-08 20:23
How far to do trial factoring	S485122	PrimeNet	1	2007-09-06 00:52
About trial factoring	gbvalor	Math	4	2003-05-22 02:04

2004-01-05, 10:48	#12
michael Dec 2003 Belgium 65₁₀ Posts	My LL-test will be done in a few days, my worktodo.ini file got an extra line: Factor=24347537,60 Does this mean it will only factor up to 60bits? seems a bit low for that kind of number, i would at least expect 67 or 68...perhaps it updates during the factoring, i didn't take a close look on my previous exponent though it says Test=22063543,67,1 which makes me think it trialfactored up to 67 bits... If i want to factor to say 72 bits, do i just change the line there(worktodo.ini)? Or do i need to change something in the prime.ini file? -michael

2004-01-05, 11:40	#14
michael Dec 2003 Belgium 5×13 Posts	undoc.txt tells me: -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 This doesn't tell me what i should use to be able to trialfactor higher...can anyone help me out on that? -michael

2004-01-05, 18:54	#16
michael Dec 2003 Belgium 5·13 Posts	Yes, i got a P4-2400. When i look at the cleared exponents database i see some numbers even trialfactored up to 100 bits...i have no intention to do this, but i wonder ... I like that idea of LMH too, there is alot of trialfactoring i could do 64-74 bitwise (as i understand P4's only are an advantage starting from 64 bits). The program they asked me to download fails to run on my computer though, no clue why or what i'm doing wrong In the end i guess it's best to just stick to the exponents primenet provides me... -michael

2004-01-06, 00:24	#19
michael Dec 2003 Belgium 101₈ Posts	Yes it's the decomp.exe that doesn't want to open... I got all the files in 1 folder, when i double click decomp.exe the doswindow pops up for about a millisecond...then disappears. I tried to get there through start/run/command but that wouldn't work out either. -michael

2004-01-06, 00:37	#21
michael Dec 2003 Belgium 5·13 Posts	I thought commandline meant dosprompt ... what am i missing here? -michael

2004-01-06, 00:40	#22
Xyzzy Aug 2002 2·3²·13·37 Posts	In XP: Start Menu -> Programs -> Accessories -> Command Prompt If decomp.exe is on your desktop, type "cd Desktop"... Then go to work...