Strange result with Prime95

[ShoeLace]

I tried to factor my range 7412000 -> 7413000 with prime95 just now


the worktodo.ini was created with the make_worktodo.pl with parameters

make_worktodo.pl 7412000 7413000  49 1.5

24 numbers with 7412000 <= n < 7413000
Searching for known factors in results.txt...
Removed 0 numbers using the factor file
24 numbers with 7412000 <= n < 7413000

i then ran this this with prime95 (widnows version i got from http://www.sslug.dk/~grove/sbfactor/)

and the only result i got was

10223*2^7412009+1 does not need P-1 factoring.
24737*2^7412071+1 does not need P-1 factoring.
24737*2^7412191+1 does not need P-1 factoring.
55459*2^7412194+1 does not need P-1 factoring.

etc..

and it was finished in like 1 min.

so i tried the sbfactor program and it found

605506658575261 | 28433*2^7412233+1

so for an expriemnt i tried
Pfactor=28433,2,7412233,1,52,9
and then it started doing something.


can i get some help/feedback on this?

[ceselb]

It looks like prime95 doesn't estimate correctly. It wasn't done with our numbers specifically in mind, so that's no huge surprise.

Have you input the correct factor depth? I haven't used prime95 myself, except for a short time while trialdividing at gimps a few years back.

[Mystwalker]

Correct me if I'm wrong, but AFAIK, the current version of SoBPM1 doesn't support floating point values as "the last value" (I've forgotten a suitable name).
Try it with a "2" instead, which means that a factor will save 2 PRP tests (first time check and double check). Maybe that does the trick...

[hc_grove]

Originally posted by ShoeLace

10223*2^7412009+1 does not need P-1 factoring.
24737*2^7412071+1 does not need P-1 factoring.
24737*2^7412191+1 does not need P-1 factoring.
55459*2^7412194+1 does not need P-1 factoring.

etc..

and it was finished in like 1 min.

Whenever I've experienced that, the reason has been that it was configured to use a max of 8MB of RAM, which just isn't enough.

I have no idea how to configure the windows version.

[ShoeLace]

excellent thank you.. i think it was the memory setting..


under Option->cpu there is a memory setting.. and it was defaulted to 8MB

ive set it to 512MB are recreated teh worktodo.ini with 49 and 2 and it seems to be working fine now.


NB: memory usage is saved in local.ini eg

OldCpuType=12
OldCpuSpeed=1795
CPUHours=24
DayMemory=768
NightMemory=768
DayStartTime=450
DayEndTime=141

thanks everyone

[ShoeLace]

last slightly dumb question..

where does prime95 output any factors that it finds?

are they in a format the teh submission page will accept etc..

???

[garo]

Prime95 puts all results in results.txt . This can be a problem since results.txt means something else for SOB. A work around is to insert his line in prime.ini

results.txt = factor.txt

where the results will now be stoerd in factor.txt

However the results are printed in a format that is different from what SoB expects. I wonder if anyone would write a script to convert the result format.

[Frodo42]

Originally posted by garo
However the results are printed in a format that is different from what SoB expects. I wonder if anyone would write a script to convert the result format.

I would love to have that script also.
It gets a bit tedious looking through al the tests done for factors, it was easier with the "old" fact.txt where factors had the correct format for submitting them

p|k*2^n+1

[hc_grove]

Originally posted by garo
However the results are printed in a format that is different from what SoB expects. I wonder if anyone would write a script to convert the result format.

I've been planning to write that script for a long time, now I did it:

#! /usr/bin/perl

use strict;
use warnings;

my $resultsfile = 'results.txt';

open(RESULTS,$resultsfile) or die "Couldn't open $resultsfile.\n";

open(FACT,'>fact.txt') or die "Couldn't open fact.txt.\n";

while (<RESULTS>) {
    if (m/^(\d+)\*2\^(\d+)\+1 has a factor: (\d+)/) {
	print FACT "$3 | $1*2^$2+1\n";
    }
}
close(RESULTS);
close(FACT);

It assumes that the results are actually written to results.txt, but that can be changed in line 6 (counting the blanks).

[Frodo42]

thanks hc_grove
I guess had you in the back of my head when talking of that script and again you get the job done
btw. people should be careful if they have named their output file from prime95 fact.txt then line 10 should also be changed.