Reserved for MF - Sequence 4788 - Page 143

yoyo · 2012-03-26, 18:11

Quote:

Originally Posted by Batalov

Did anyone accidentally throw 17800 11e7 curves at the c156? Don't be shy! ;-)

Nearly done and nothing found :(
http://www.rechenkraft.net/yoyo/y_status_ecm.php
There are more curves needed?
yoyo

debrouxl · 2012-03-26, 19:17

Nope, the t55 you ran on yoyo@home has raised the bar significantly above the usual level for a C156

GPU-based ECM stage 1 is slowly making the "2/7 of GNFS difficulty, 2/9 of SNFS difficulty" rules of thumb for ECM work obsolete, but you're not using that (yet) on yoyo@home.

Batalov · 2012-03-29, 01:44

Postprocessing in on, ETA for LA is 25 hours.

Batalov · 2012-03-30, 06:48

The good news is the sequence is still alive.
The bad news is none so far! ...Ooops. Spoke too soon. There was a tiny earthquake, @3.3, tiny, but right here. Well. Almost..
Iteration 2953, c127. Will do in 6 hrs.

schickel · 2012-03-30, 06:57

Quote:

Originally Posted by Batalov

The good news is the sequence is still alive.
The bad news is ....

Urk....you scared me for a second there; I thought maybe you broke the downdriver again.....

Batalov · 2012-03-30, 17:28

c149 in i2954. (t35 only scanned)
I am switching back to Lucas/Fib "regular scheduled programming".

c153 now

jrk · 2012-03-30, 20:24

Quote:

Originally Posted by Batalov

c153 now

I have started poly selection, just in case.

yoyo · 2012-03-30, 20:29

Quote:

Originally Posted by jrk

I have started poly selection, just in case.

I started ecm curves for 43M and afterwards for 110M.

yoyo · 2012-03-30, 21:01

Quote:

Originally Posted by yoyo

I started ecm curves for 43M and afterwards for 110M.

And found a factor ;)

Dubslow · 2012-03-30, 21:26

Out of a lack of better places to put this...

Can some explain (or link me to an explanation of) how much ECM to throw at a given composite before NFS/QS? Meaning, what exactly does the "txx" mean, and how do I decide what B1 to use based on size of n and current t-depth?

I tried Googling it, but didn't find much. (I do understand that ECM finds factors probabilistically, and that txx says something about the probability of having found a yy-digit factor, but I have no idea of size-of-factor relates to bounds or to txx.)

akruppa · 2012-03-30, 22:34

The conventional wisdom for GNFS numbers is: do the expected number of curves to find a factor of 1/3 the size of the number. For a 150-digit input, that meant do enough curves for a 50-digit factor, which people abbreviate as "t50" or "p50 test."

Note that the 1/3 rule is not a result of an analysis of the algorithms; in fact, when you look at the complexity functions of ECM and NFS, there can't be a simple such constant. It's just a rule of thumb that, for numbers of the size we usually factor, is usually "close enough."

2012-03-30, 17:28	#1568
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 9,497 Posts	c149 in i2954. (t35 only scanned) I am switching back to Lucas/Fib "regular scheduled programming". c153 now Last fiddled with by Batalov on 2012-03-30 at 18:18

2012-03-30, 21:26	#1572
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 3×29×83 Posts	Out of a lack of better places to put this... Can some explain (or link me to an explanation of) how much ECM to throw at a given composite before NFS/QS? Meaning, what exactly does the "txx" mean, and how do I decide what B1 to use based on size of n and current t-depth? I tried Googling it, but didn't find much. (I do understand that ECM finds factors probabilistically, and that txx says something about the probability of having found a yy-digit factor, but I have no idea of size-of-factor relates to bounds or to txx.) Last fiddled with by Dubslow on 2012-03-30 at 21:27

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Reserved for MF - Sequence 3366	RichD	Aliquot Sequences	470	2021-04-22 02:17
Reserved for MF - Sequence 3408	RichD	Aliquot Sequences	474	2021-03-07 20:28
Reserved for MF - Sequence 276	kar_bon	Aliquot Sequences	127	2020-12-17 10:05
Assignments are reserved but not showing up	prism019	GPU to 72	6	2020-09-21 22:11
80M to 64 bits ... but not really reserved	petrw1	Lone Mersenne Hunters	82	2010-01-11 01:57

2012-03-26, 19:17	#1564
debrouxl Sep 2009 1111010001₂ Posts	Nope, the t55 you ran on yoyo@home has raised the bar significantly above the usual level for a C156 GPU-based ECM stage 1 is slowly making the "2/7 of GNFS difficulty, 2/9 of SNFS difficulty" rules of thumb for ECM work obsolete, but you're not using that (yet) on yoyo@home.

2012-03-29, 01:44	#1565
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 9,497 Posts	Postprocessing in on, ETA for LA is 25 hours.

2012-03-30, 06:48	#1566
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 9497₁₀ Posts	The good news is the sequence is still alive. The bad news is none so far! ...Ooops. Spoke too soon. There was a tiny earthquake, @3.3, tiny, but right here. Well. Almost.. Iteration 2953, c127. Will do in 6 hrs.

2012-03-30, 22:34	#1573
akruppa "Nancy" Aug 2002 Alexandria 2,467 Posts	The conventional wisdom for GNFS numbers is: do the expected number of curves to find a factor of 1/3 the size of the number. For a 150-digit input, that meant do enough curves for a 50-digit factor, which people abbreviate as "t50" or "p50 test." Note that the 1/3 rule is not a result of an analysis of the algorithms; in fact, when you look at the complexity functions of ECM and NFS, there can't be a simple such constant. It's just a rule of thumb that, for numbers of the size we usually factor, is usually "close enough."