Thinking out loud about getting under 20M unfactored exponents - Page 23

Prime95 · 2020-12-18, 23:45

Linux 64:

https://www.dropbox.com/s/9ksfnkwtx5...64.tar.gz?dl=0

masser · 2020-12-19, 02:03

If a factor is not found, does the new code report the requested B2 or the computed B2? In my initial results, ~~it appears to report the requested B2, which seems inaccurate.~~

These are the lines I tried:

Quote:

Pminus1=1,2,21150827,-1,6133,596857,66
Pminus1=1,2,31919773,-1,1901,84737,66

NEVERMIND: I'm wrong - misread the outputs. Code reported the B2 it used, which was not adequate in this case... I'll try to see how much memory to allocate to get the desired factor.

Prime95 · 2020-12-19, 02:34

A present for your project -- I was comparing P-1 speeds for 30.3 vs. 30.4 and found this:

processing: P-1 factor 74415394148849438918449121 for M16963013 (B1=500,000, B2=20,500,000)

petrw1 · 2020-12-19, 04:28

Quote:

Originally Posted by Prime95

A present for your project

And....inquiring minds want to know...how do they compare please?

Prime95 · 2020-12-19, 06:19

Quote:

Originally Posted by petrw1

And....inquiring minds want to know...how do they compare please?

Was not a good test as stage 2 was limited to 300MB memory. Stage 1 was same speed.
30.3 did 17.3% of stage 2 to B2=10M in 470 seconds.
30.4 did 43.4% of stage 2 to B2=20.5M in 1940 seconds.

My extrapolation is 30.3 would have taken 2480 seconds to do what 30.4 did.

axn · 2020-12-19, 06:52

@George: Can you share the B2 selection logic (source code)? I want to understand/predict how memory allocation will affect this. Does the logic take into account MaxHighMemWorkers / worker-specific memory settings?

petrw1 · 2020-12-19, 17:35

Are the save files compatible. I'm currently at 29.8

Prime95 · 2020-12-19, 20:06

Quote:

Originally Posted by axn

@George: Can you share the B2 selection logic (source code)? I want to understand/predict how memory allocation will affect this. Does the logic take into account MaxHighMemWorkers / worker-specific memory settings?

B2 is selected when stage 2 starts. It uses the available memory according to Day/Night MaxHighMemWorkers/etc settings at the time stage 2 begins.

The source code will be made available, but emulating the B2 selection logic will not be easy.

Quote:

Originally Posted by petrw1

Are the save files compatible. I'm currently at 29.8

See post #242

kruoli · 2020-12-19, 20:13

Quote:

Originally Posted by axn

@George: [...] Does the logic take into account MaxHighMemWorkers / worker-specific memory settings?

That would be specifically interesting if e. g. an extra 100 MB would cause the algorithm to select a higher B2 with relatively little higher ETA.

An ECM got: M630901, B1 = 250k, B2 =155 * B1 @ 6 GB availible. Stage 1 took around 0.5h, stage 2 took 0.25h.

Edit: @George, thanks for a rough clarifications.

masser · 2020-12-20, 00:28

Could someone teach me how to find the sigma that was used to find an ECM factor from the mersenne.org website?

I'm trying to test that the new ECM in version 30.4 recovers known factors without running hundreds of curves.

axn · 2020-12-20, 02:40

Potential P-1 (display) bug:

Stage 2 % resets to zero once a batch of relative primes is processed.

At least, that's what I think is happening -- haven't completed a full P-1 run yet.

2020-12-20, 00:28	#252
masser Jul 2003 wear a mask 1,553 Posts	Could someone teach me how to find the sigma that was used to find an ECM factor from the mersenne.org website? I'm trying to test that the new ECM in version 30.4 recovers known factors without running hundreds of curves. Last fiddled with by masser on 2020-12-20 at 00:29

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Thinking of Joining GPU to 72	jschwar313	GPU to 72	3	2016-01-31 00:50
Thinking about lasieve5	Batalov	Factoring	6	2011-12-27 22:40
Thinking about buying a panda	jasong	jasong	1	2008-11-11 09:43
Loud thinking on irregular primes	devarajkandadai	Math	4	2007-07-25 03:01
Question on unfactored numbers...	WraithX	GMP-ECM	1	2006-03-19 22:16

2020-12-18, 23:45	#243
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 1110010111101₂ Posts	Linux 64: https://www.dropbox.com/s/9ksfnkwtx5...64.tar.gz?dl=0

2020-12-19, 02:34	#245
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 7×1,051 Posts	A present for your project -- I was comparing P-1 speeds for 30.3 vs. 30.4 and found this: processing: P-1 factor 74415394148849438918449121 for M16963013 (B1=500,000, B2=20,500,000)

2020-12-19, 06:52	#248
axn Jun 2003 2³×607 Posts	@George: Can you share the B2 selection logic (source code)? I want to understand/predict how memory allocation will affect this. Does the logic take into account MaxHighMemWorkers / worker-specific memory settings?

2020-12-19, 17:35	#249
petrw1 1976 Toyota Corona years forever! "Wayne" Nov 2006 Saskatchewan, Canada 2·3²·11·23 Posts	Are the save files compatible. I'm currently at 29.8

2020-12-20, 02:40	#253
axn Jun 2003 2³·607 Posts	Potential P-1 (display) bug: Stage 2 % resets to zero once a batch of relative primes is processed. At least, that's what I think is happening -- haven't completed a full P-1 run yet.