Mfaktc keeps going after a factor is found

NBtarheel_33 · 2011-10-19, 08:58

Not sure if this is intended or not, but whenever mfaktc finds a factor in the middle of a bit level, it does not immediately stop and begin the next assignment (as Prime95 would); rather, it continues on to the end of the bit level, as if no factor were found.

TheJudger · 2011-10-19, 10:11

Hi,

works as expected.

check mfaktc.ini:

Code:

# possible values for StopAfterFactor:
# 0: Do not stop the current assignment after a factor was found.
# 1: When a factor was found for the current assignment stop after the
#    current bitlevel. This makes only sense when Stages is enabled.
# 2: When a factor was found for the current assignment stop after the
#    current class.

StopAfterFactor=1

Oliver

Chuck · 2011-10-19, 15:37

M602249941 has a factor: 86439049638640106399
M602249941 has a factor: 77482243219676340503
M602249941 has a factor: 22781228486416840393
found 3 factor(s) for M602249941 from 2^64 to 2^67 [mfaktc 0.17-Win barrett79_mul32]

In this sort of unusual case I was glad it kept going.

Chuck

delta_t · 2011-10-19, 19:49

If you set, StopAfterFactor=2, it will stop after the current class, but the output would read (notice the (partially tested)):

Code:

M490287241 has a factor: 23617822113724688119
found 1 factor(s) for M490287241 from 2^64 to 2^66 (partially tested) [mfaktc 0.17-Win barrett79_mul32]

As with Chuck, I'm glad I didn't have it set to =2 as that exponent had two factors in the tested range.

Code:

M490287241 has a factor: 23617822113724688119
M490287241 has a factor: 19751191822676004343
found 2 factor(s) for M490287241 from 2^64 to 2^66 [mfaktc 0.17-Win barrett79_mul32]

Some like to make sure the bit range is tested through instead of partially.

MikeBerlin · 2012-04-07, 08:25

Quote:

Originally Posted by Chuck

... I was glad it kept going....Chuck

But why, I dont understand, why it make sense to test a mersenne after finding a factor. In my opinion it's only lost time, isn´t it?

Karl M Johnson · 2012-04-07, 11:35

Nope.
Other mersenne numbers may be divisible by that factor, iirc.

Chuck · 2012-04-07, 13:33

After I joined GPU272, I changed the setting to stop when a factor is found. When I was doing those very high 600M TFs it only took a short time to check one, but now that I am working in the 59M range it takes 5 1/2 hours for each test, so it seems smarter to stop when a factor is found.

axn · 2012-04-07, 14:47

Quote:

Originally Posted by Karl M Johnson

Nope.
Other mersenne numbers may be divisible by that factor, iirc.

Nope. No other mersenne numbers will be divisible by that factor.

Karl M Johnson · 2012-04-07, 16:33

Quote:

Originally Posted by axn

Nope. No other mersenne numbers will be divisible by that factor.

So there's no real point in continuing a TF job after a factor was found ?

bcp19 · 2012-04-07, 16:44

Quote:

Originally Posted by Karl M Johnson

So there's no real point in continuing a TF job after a factor was found ?

Since a factor was found, it is safe to stop as you have disproved primality. If your goal is full factorization, letting it continue could add another factor.

axn · 2012-04-07, 17:04

Quote:

Originally Posted by Karl M Johnson

So there's no real point in continuing a TF job after a factor was found ?

Unless you want to find the smallest factor for a given number (which is midly interesting, mathematically speaking).

2011-10-19, 08:58	#1
NBtarheel_33 "Nathan" Jul 2008 Maryland, USA 5×223 Posts	Mfaktc keeps going after a factor is found Not sure if this is intended or not, but whenever mfaktc finds a factor in the middle of a bit level, it does not immediately stop and begin the next assignment (as Prime95 would); rather, it continues on to the end of the bit level, as if no factor were found.

2011-10-19, 15:37	#3
Chuck May 2011 Orange Park, FL 13·71 Posts	I was glad it kept going.... M602249941 has a factor: 86439049638640106399 M602249941 has a factor: 77482243219676340503 M602249941 has a factor: 22781228486416840393 found 3 factor(s) for M602249941 from 2^64 to 2^67 [mfaktc 0.17-Win barrett79_mul32] In this sort of unusual case I was glad it kept going. Chuck

2011-10-19, 19:49	#4
delta_t Nov 2002 Anchorage, AK 3·7·17 Posts	If you set, StopAfterFactor=2, it will stop after the current class, but the output would read (notice the (partially tested)): Code: M490287241 has a factor: 23617822113724688119 found 1 factor(s) for M490287241 from 2^64 to 2^66 (partially tested) [mfaktc 0.17-Win barrett79_mul32] As with Chuck, I'm glad I didn't have it set to =2 as that exponent had two factors in the tested range. Code: M490287241 has a factor: 23617822113724688119 M490287241 has a factor: 19751191822676004343 found 2 factor(s) for M490287241 from 2^64 to 2^66 [mfaktc 0.17-Win barrett79_mul32] Some like to make sure the bit range is tested through instead of partially. Last fiddled with by delta_t on 2011-10-19 at 19:50

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Factor found that should have been found by P-1	tha	Data	65	2020-08-05 21:11
F12 factor found?	johnadam74	FermatSearch	16	2016-11-03 12:10
mfaktc: Mis-reported composite factor	mattmill30	GPU Computing	1	2016-02-05 19:41
mfaktc TF credit 2x higher if factor found?	S34960zz	PrimeNet	10	2011-10-13 07:00
After a factor is found it keeps on going	jocelynl	Software	6	2004-08-07 01:31

2012-04-07, 11:35	#6
Karl M Johnson Mar 2010 3·137 Posts	Nope. Other mersenne numbers may be divisible by that factor, iirc.

2012-04-07, 13:33	#7
Chuck May 2011 Orange Park, FL 923₁₀ Posts	After I joined GPU272, I changed the setting to stop when a factor is found. When I was doing those very high 600M TFs it only took a short time to check one, but now that I am working in the 59M range it takes 5 1/2 hours for each test, so it seems smarter to stop when a factor is found.