ecm with Fermat numbers

ET_ · 2016-07-17, 14:36

Question: with regard to both mprime and gmp-ecm, is it faster running the Fermat number "as-is", or divided by its known factors?

IIRC, mprime was faster using the standard notation and providing the known factors either on the worktodo.txt line or on the lowp.txt file, while gmp-ecn was faster as the number to test grew smaller.

Is it still correct?

Luigi

GP2 · 2016-08-02, 19:40

As an experiment, I decided to run mprime twice on the same machine, once with the known factors and once without.

This was using the latest version, 28.9

In other words:

Code:

ECM2=1,2,4096,1,800000000,80000000000,1,"114689,26017793,63766529,190274191361,1256132134125569,568630647535356955169033410940867804839360742060818433"

versus

Code:

ECM2=1,2,4096,1,800000000,80000000000,1

Stage 1:
The two versions ran absolutely identically. The second version was maybe 0.1% faster.

Stage 2:
The first version completed stage 2 in 42% of the time that stage 1 took.

The second version "discovered" a composite factor after stage 1 and stopped without doing stage 2.

PS, when I run ECM exponents I always add:

Code:

ScaleOutputFrequency=1

to prime.txt for convenience, to avoid excessively rapid output for small exponents. See undoc.txt for explanations.

You can also add

Code:

ContinueECM=1

if you like.

2016-07-17, 14:36	#1
ET_ Banned "Luigi" Aug 2002 Team Italia 3×1,619 Posts	ecm with Fermat numbers Question: with regard to both mprime and gmp-ecm, is it faster running the Fermat number "as-is", or divided by its known factors? IIRC, mprime was faster using the standard notation and providing the known factors either on the worktodo.txt line or on the lowp.txt file, while gmp-ecn was faster as the number to test grew smaller. Is it still correct? Luigi Last fiddled with by ET_ on 2016-07-17 at 14:37

2016-08-02, 19:40	#2
GP2 Sep 2003 5035₈ Posts	As an experiment, I decided to run mprime twice on the same machine, once with the known factors and once without. This was using the latest version, 28.9 In other words: Code: ECM2=1,2,4096,1,800000000,80000000000,1,"114689,26017793,63766529,190274191361,1256132134125569,568630647535356955169033410940867804839360742060818433" versus Code: ECM2=1,2,4096,1,800000000,80000000000,1 Stage 1: The two versions ran absolutely identically. The second version was maybe 0.1% faster. Stage 2: The first version completed stage 2 in 42% of the time that stage 1 took. The second version "discovered" a composite factor after stage 1 and stopped without doing stage 2. PS, when I run ECM exponents I always add: Code: ScaleOutputFrequency=1 to prime.txt for convenience, to avoid excessively rapid output for small exponents. See undoc.txt for explanations. You can also add Code: ContinueECM=1 if you like. Last fiddled with by GP2 on 2016-08-02 at 19:46

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
P-1/P+1 on Fermat numbers	ATH	Operazione Doppi Mersennes	2	2015-01-25 06:27
Generalized Fermat Numbers	ET_	Programming	4	2008-06-23 07:59
Are there any Fermat numbers that might be prime?	jasong	Math	39	2007-10-27 23:11
LLT numbers, linkd with Mersenne and Fermat numbers	T.Rex	Math	4	2005-05-07 08:25
Fermat Numbers	devarajkandadai	Math	8	2004-07-27 12:27