How to start?

Thomas11 · 2003-02-19, 12:03

Hi all,

I would like to do trialfactoring on some range, but I need a bit more information on how to do that. I did some test runs using Goerge Woltmans mprime code on Linux and observed that an AdvancedFactor job is much slower than "normal" factoring using the FactorOverride option and giving one line for each exponent in the worktodo.ini file, e.g. AdvancedFactoring from 2^31 to 2^32 takes as long as normal Factoring from 2^62 to 2^63.

If I run a factoring job like the following:

Factor=21866239,0

then the screen output starts around 2^57. Is it because the factoring below that value is done that fast or doesn't the factoring start at 2^0?

So my question is what type of factoring is the right one and what are you using?

Thanks in advance,

Thomas.

tha · 2003-02-19, 12:55

Quote:

Originally Posted by Thomas11

I would like to do trialfactoring on some range, but I need a bit more information on how to do that.

Hi Thomas,

Let me tell you and others who will read this later how most of us work.

Usually we use P95 for Windows or MPrime for Linux downloaded from
http://www.mersenne.org/freesoft.htm
In the Test|Primenet menuoption we make sure that 'Use Primenet' is unchecked, we don't use Primenet. Then from the page
http://www.mersenne.org/status.htm
we download decomp.zip and nofactor.zip. Read the readme file of decomp and then if you want to have all exponents between 52.5 million and 53 million or so use the following setting:
decomp -w 52500000 53000000
This creates an output file that you have to rename to worktodo.ini and move it to the directory in which you have MPrime or P95.
In the file prime.ini that come with MPrime or P95 add the following two lines:
FactorOverride=60
WellBehavedWork=1
This will stop factoring above 2^60, you may want to use 58 or 64, whatever your choice. WellBehavedWork will make sure your I/O is not stressed too much as a result of the FactorOverride setting.
Send in the results every now and then to George Woltman. Download a new nofactor.zip when available (weekly) and needed for choosing a new range. A graphical overview of how much factoring has been done can be found at
http://home.planet.nl/~tha/mersenne/index.html

Post any more questions you have in this forum. Also look in this forum to see which ranges have been taken by others and post your own range over here.

YotN,

Henk

Thomas11 · 2003-02-19, 16:49

Thank you, Henk.

I've started trialfactoring for the range 44M-45M up to 2^60.

With best wishes,

Thomas.

flasholinos · 2003-06-17, 10:23

where we pt the decomp -w 52500000 53000000 ?????

S80780 · 2003-06-17, 15:27

You got to type this in at the command line (Ensure you unzipped decomp.zip and nofactor.zip to the same directory and that your current dirotory is the one you unzipped to.).
If you want to LL-test a certain range, type in " -t " instead of " -w ".
The program creates a file called "nofactor" which you rename in "worktodo.ini" and move to the directory, you installed prime95 / mprime to, as stated by Henk.
You should add the "WellBehavedWork=1" line to prime.ini in both cases. The "FactorOverride=n" line is only of interest, if you trialfactor a range. If you LL-test a range, it doesn't matter.

Benjamin

tom11784 · 2003-08-29, 19:21

command line of what file? sorry, but I'm a bit lost in trying to get a range of my own to test

garo · 2003-08-29, 19:29

commandline = dos prompt

leifbk · 2005-01-31, 18:09

I've followed tha's instructions, and downloaded decomp.zip and nofactor.zip, and unpacked them on my Linux machine. As I couldn't get decomp.c to compile, I sent decomp.exe and nofactor.cmp to my Windows laptop. This machine don't currently run Prime95, but has version 23.8.1 installed. I ran

Code:

decomp -w 29600000 29700000

this produced a nofactor file of 50,500 bytes, that I copied to worktodo.ini in the Prime95 folder. I've also made the required changes to prime.ini:

Code:

FactorOverride=60
WellBehavedWork=1

as well as unchecking the "Use PrimeNet" option. But as soon as I start Prime95 with this worktodo.ini, I get an endless row of

Code:

Error: Work-to-do file contained bad factoring assignment: XXXXXXXX,60

where XXXXXXXX is any number in the range. When it's done, the worktodo.ini is empty.

I've tried a couple of other ranges as well, but I get the same result.

The nofactor file looks perfectly normal eg. with more<nofactor, with the first line looking like this:

Code:

Factor=29600027,60

Can anybody suggest a cure for this situation? I've probably done something silly again

regards, Leif.

ric · 2005-01-31, 18:36

Quote:

Originally Posted by leifbk

I've also made the required changes to prime.ini:

Code:

FactorOverride=60

All of those candidates have already been factored to 2^60, so you'll have to raise FactorOverride to 62 (or whatever other depth you'll be LMH-ing to).

ric

garo · 2005-01-31, 18:44

ric beat me to the response there!

Also, leif can you tell me what the error message was while compiling? I have successfully compiled decomp.c on Solaris and Linux. Did you use the -lm flag to tell it to use the math library?

leifbk · 2005-01-31, 18:48

Quote:

Originally Posted by ric

All of those candidates have already been factored to 2^60, so you'll have to raise FactorOverride to 62 (or whatever other depth you'll be LMH-ing to).

Right -- I said that I'd probably done something silly

Thanks. Then I'll try to factor 29600000-29700000 to 62. It seems to work now, and I've got some wrecks that I can spread this row around on.

regards, Leif.

2003-02-19, 12:03	#1
Thomas11 Feb 2003 2⁷×3×5 Posts	How to start? Hi all, I would like to do trialfactoring on some range, but I need a bit more information on how to do that. I did some test runs using Goerge Woltmans mprime code on Linux and observed that an AdvancedFactor job is much slower than "normal" factoring using the FactorOverride option and giving one line for each exponent in the worktodo.ini file, e.g. AdvancedFactoring from 2^31 to 2^32 takes as long as normal Factoring from 2^62 to 2^63. If I run a factoring job like the following: Factor=21866239,0 then the screen output starts around 2^57. Is it because the factoring below that value is done that fast or doesn't the factoring start at 2^0? So my question is what type of factoring is the right one and what are you using? Thanks in advance, Thomas.

2003-02-19, 16:49	#3
Thomas11 Feb 2003 2⁷·3·5 Posts	Re: How to start? Thank you, Henk. I've started trialfactoring for the range 44M-45M up to 2^60. With best wishes, Thomas.

2005-01-31, 18:09	#8
leifbk May 2004 Oslo, Norway 1111000₂ Posts	I've followed tha's instructions, and downloaded decomp.zip and nofactor.zip, and unpacked them on my Linux machine. As I couldn't get decomp.c to compile, I sent decomp.exe and nofactor.cmp to my Windows laptop. This machine don't currently run Prime95, but has version 23.8.1 installed. I ran Code: decomp -w 29600000 29700000 this produced a nofactor file of 50,500 bytes, that I copied to worktodo.ini in the Prime95 folder. I've also made the required changes to prime.ini: Code: FactorOverride=60 WellBehavedWork=1 as well as unchecking the "Use PrimeNet" option. But as soon as I start Prime95 with this worktodo.ini, I get an endless row of Code: Error: Work-to-do file contained bad factoring assignment: XXXXXXXX,60 where XXXXXXXX is any number in the range. When it's done, the worktodo.ini is empty. I've tried a couple of other ranges as well, but I get the same result. The nofactor file looks perfectly normal eg. with more<nofactor, with the first line looking like this: Code: Factor=29600027,60 Can anybody suggest a cure for this situation? I've probably done something silly again regards, Leif.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Where should I start?	christian_	Information & Answers	9	2016-01-22 19:28
Before you start...	jasonp	Operation Kibibit	65	2013-09-03 22:06
Where to start	Jellyfish420	Homework Help	46	2013-02-06 13:51
how to start with P-1?	ValerieVonck	Marin's Mersenne-aries	8	2006-04-29 22:21
How to start?	OmbooHankvald	Factoring	15	2005-09-03 13:42

2003-06-17, 10:23	#4
flasholinos Jun 2003 4₁₀ Posts	where we pt the decomp -w 52500000 53000000 ?????

2003-06-17, 15:27	#5
S80780 Jan 2003 far from M40 7D₁₆ Posts	You got to type this in at the command line (Ensure you unzipped decomp.zip and nofactor.zip to the same directory and that your current dirotory is the one you unzipped to.). If you want to LL-test a certain range, type in " -t " instead of " -w ". The program creates a file called "nofactor" which you rename in "worktodo.ini" and move to the directory, you installed prime95 / mprime to, as stated by Henk. You should add the "WellBehavedWork=1" line to prime.ini in both cases. The "FactorOverride=n" line is only of interest, if you trialfactor a range. If you LL-test a range, it doesn't matter. Benjamin

2003-08-29, 19:21	#6
tom11784 Aug 2003 Upstate NY, USA 2·163 Posts	command line of what file? sorry, but I'm a bit lost in trying to get a range of my own to test

2003-08-29, 19:29	#7
garo Aug 2002 Termonfeckin, IE AD0₁₆ Posts	commandline = dos prompt

2005-01-31, 18:44	#10
garo Aug 2002 Termonfeckin, IE 2⁴×173 Posts	ric beat me to the response there! Also, leif can you tell me what the error message was while compiling? I have successfully compiled decomp.c on Solaris and Linux. Did you use the -lm flag to tell it to use the math library?