mersenneforum.org Primo
 Register FAQ Search Today's Posts Mark Forums Read

 2022-01-02, 01:39 #166 Jayder     Dec 2012 32×31 Posts Is anybody currently working on the PRPs <= 3000? I may do a little or a lot of work, and I don't want to run into anyone.
 2022-01-02, 18:01 #167 chris2be8     Sep 2009 2·11·101 Posts I'm working on them from the bottom up. If you keep away from the lowest few hundred PRPs you should be OK (anything over 1500 digits should be safe for at least a month).
 2022-01-20, 19:30 #168 kruoli     "Oliver" Sep 2017 Porta Westfalica, DE 33·31 Posts Currently, I am running a certification on Sm(2445)*10^8677+Smr(2446). While the system I am running it on has a throughput of slightly above one 10k digit number per day, it now stands at 46806/57633 bits in phase 1 after more than a week (slightly above eight days). This is with the same number of threads as with the 10k candidates. Even if I assume that Primo operates in $$\mathcal{O}(\log(n)^{5+\varepsilon})$$ for small $$\varepsilon$$ instead of $$\mathcal{O}(\log(n)^{4+\varepsilon})$$, this seems way slower than it should be expected. Is my expectation flawed (maybe I computed the ETA wrong) or is there something else that could slow it done? I know that ECPP is non-deterministic algorithm and I might got an extreme sample here. Can somebody chime in if this might be the case?
2022-01-20, 19:39   #169
paulunderwood

Sep 2002
Database er0rr

76148 Posts

Quote:
 Originally Posted by kruoli Currently, I am running a certification on Sm(2445)*10^8677+Smr(2446). While the system I am running it on has a throughput of slightly above one 10k digit number per day, it now stands at 46806/57633 bits in phase 1 after more than a week (slightly above eight days). This is with the same number of threads as with the 10k candidates. Even if I assume that Primo operates in $$\mathcal{O}(\log(n)^{5+\varepsilon})$$ for small $$\varepsilon$$ instead of $$\mathcal{O}(\log(n)^{4+\varepsilon})$$, this seems way slower than it should be expected. Is my expectation flawed (maybe I computed the ETA wrong) or is there something else that could slow it done? I know that ECPP is non-deterministic algorithm and I might got an extreme sample here. Can somebody chime in if this might be the case?
I found log^4 is a good rule of thumb. Have you selected max settings on the certification page? 11k dd? (?). Things will speed up! I know it can be disheartening to watch it backtrack, A watched kettle never boils!

 2022-01-20, 19:41 #170 kruoli     "Oliver" Sep 2017 Porta Westfalica, DE 33·31 Posts Yes, I used 11k digits and the maximum setting in the other field. Watching it is only fun when it has sped up considerably towards the end of phase 1.

 Similar Threads Thread Thread Starter Forum Replies Last Post PawnProver44 Information & Answers 14 2016-04-09 05:49 WraithX Software 15 2013-09-10 07:24 wblipp FactorDB 1 2012-05-28 03:16 Cybertronic Five or Bust - The Dual Sierpinski Problem 17 2009-08-13 20:42 fivemack Math 35 2009-04-28 15:03

All times are UTC. The time now is 00:24.

Fri Jan 21 00:24:28 UTC 2022 up 181 days, 18:53, 0 users, load averages: 1.50, 1.59, 1.53