mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > PrimeNet

Reply
 
Thread Tools
Old 2019-07-03, 22:14   #144
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

3·312 Posts
Default

As of right now there's still about 4 million in the queue, but they're almost all above 4G, just over 100k left in the 3M range and they should be processed within a couple hours. The bulk factors page may still show a large queue count for a few more days but the impact on everyone's submitted work should be just about past. At least, submitted results should now always be processed within 2-3 hours (probably faster, but I don't want to make too many promises I can't monitor).
James Heinrich is online now   Reply With Quote
Old 2019-07-04, 17:44   #145
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

3×312 Posts
Default

Sub-4G factor queue remains empty, the remaining backlog is just above 4G where it won't affect anyone. And I have just cleared out the nonfactor backlog so everything should be pretty much back to normal now. I'll devise a separate queue system for the really-bulk factors once I get back online. Which I'm currently not. After 7.5 days. Amazing.
James Heinrich is online now   Reply With Quote
Old 2019-07-06, 00:54   #146
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

3×312 Posts
Default

Wow. I'm back online?!

Of course, now that I can see what's going on, I see there are some things that desperately need fixing, so I've taken mersenne.ca down temporarily (less than an hour I hope) while I urgently fix something I broke 2 weeks ago but didn't discover till now, that's causing major slowdowns and broken pages (probably also contributed to the slower-than-it-should-be processing of the factor queue).

I'll put it back online just as soon as I can.

edit: I've put the site back online, just be aware that TF history will be largely missing from site reports for the next day or so as I work around my oops. The data is there, I just need to work on it a bit before you can see it again.

Last fiddled with by James Heinrich on 2019-07-06 at 01:06
James Heinrich is online now   Reply With Quote
Old 2019-07-06, 20:29   #147
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

288310 Posts
Default

All the major chaos that I'm aware of should now be resolved.

The factor queue is holding steady at under 1 million. I will reduce that if possible, but even if it stays there (or grows) don't worry, I've made adjustments so that larger factors are processed first, and since the mass influx of new factors are all smaller than 55-bit, that means all the "normal" TF results (which are 67-bit or larger) will be processed first. You should always see your results processed within 30 minutes (probably less) from time of submission now.

Thanks for everyone's patience.
James Heinrich is online now   Reply With Quote
Old 2019-07-06, 20:57   #148
hansl
 
hansl's Avatar
 
Apr 2019

5×41 Posts
Default

Hello everyone, I figured I may as well reveal myself as the factoring maniac that's been hopefully not causing too much headache lately. I've been coordinating with James mostly in PM since I started doing this bulk work. As he's mentioned already, my goal is to TF the full range 1-10G to 55bits, without stopping for found factors. And yes, TJAOI's efforts have been a source of inspiration for this project idea, but also to serve as a general double check which has proven useful already.

I recently downloaded the exports of all factors from mersenne.ca and implemented a filter on my side so that I will no longer submit factors which are already in the db. This cuts down my submissions to the queue by about 5x, or 20% of what they were just days ago. Between that and recent improvements from James, the queue is rapidly returning to 0 again.

In case anyone is curious about my setup, I have 5 different computers working on various ranges, using a specialized PARI script for most of the work. My biggest beasty which I recently upgraded is a dual socket Xeon E5-2697 v2 with 24C/48T and all threads devoted to trial factoring. The other computers are various 4-6 year old boxes which are mostly doing PrimeNet DC tasks, and using the spare hyperthreads on this TF project. I have a few bash scripts to tie it all together, and I wrote the known-factor-filtering program using C++/GMP.

It looks like I've found about 7-8 million new factors so far, with a few thousand of those actually being first time factors on an exponent which were missed one way or another the first time around. A rough estimate is that it should take me another 5-6 months to fully complete this project at my current rates. I also estimate that there should be about 40 million new factors by the end of it
hansl is offline   Reply With Quote
Old 2019-07-07, 14:23   #149
alpertron
 
alpertron's Avatar
 
Aug 2002
Buenos Aires, Argentina

1,319 Posts
Default

It appears that the page https://www.mersenne.ca/manyfactors.php does not show the number of Mersenne numbers with 4 factors known.
alpertron is offline   Reply With Quote
Old 2019-07-07, 14:34   #150
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

3·312 Posts
Default

Quote:
Originally Posted by alpertron View Post
It appears that the page https://www.mersenne.ca/manyfactors.php does not show the number of Mersenne numbers with 4 factors known.
No, I removed the 4-factor list a few weeks ago, with all these new factors piling in recently the list was becoming very large. Exponents with 5 known factors are already more than 100k, you can extrapolate that progression to how many 4-factor exponents there would be.
James Heinrich is online now   Reply With Quote
Old 2019-07-07, 17:33   #151
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

3·312 Posts
Default

Quote:
Originally Posted by hansl View Post
Between that and recent improvements from James, the queue is rapidly returning to 0 again.
It's been a real fight with your "big beasty" vs my little server trying to record the factors faster than you find them, but I think I'm finally victorious! The queue is down to zero! Well, approximately zero; inevitably more factors come in during a processing run, but the queue size should be hovering somewhere around 10k now.
James Heinrich is online now   Reply With Quote
Old 2019-07-07, 18:09   #152
hansl
 
hansl's Avatar
 
Apr 2019

5·41 Posts
Default

Yeah I had it working on 2G range, but after you asked me to work on >4G, I switched it to 5G range. One side effect of that is that it goes through them ~2.5x faster! (each 2*k*p+1 covers proportionally more of the bit range). Anyways I should be done with the 5G range in a few more days, after which I'll reassign it to lower ranges again, which should slow it down a bit. I'll keep my slower machines on the highest ranges.
hansl is offline   Reply With Quote
Old 2019-07-08, 13:32   #153
nomead
 
nomead's Avatar
 
"Sam Laur"
Dec 2018
Turku, Finland

4748 Posts
Default

Just a rough estimate for comparison... I measured run times here and there between 1G and 4.2G, and getting all the factors up to 64 bits (StopAfterFactor=0) should take about a year on mfaktc and a single RTX 2080. But of course this would still leave everything above that 32-bit limit untouched.
nomead is offline   Reply With Quote
Old 2019-07-08, 13:49   #154
James Heinrich
 
James Heinrich's Avatar
 
"James Heinrich"
May 2004
ex-Northern Ontario

55038 Posts
Default

Quote:
Originally Posted by hansl View Post
It looks like I've found about 7-8 million new factors so far, with a few thousand of those actually being first time factors on an exponent which were missed one way or another the first time around. I also estimate that there should be about 40 million new factors by the end of it
You've now passed 10 million new factors, out of the ~42 million I expect you to find. And you found factors for 3260 exponents that previously didn't have any recorded (mostly due to some data errors I've since fixed). Of course, the higher ranges will complete quicker and so your overall rate will likely slow down a bit (I'm not complaining ), but not bad for a month's work!
James Heinrich is online now   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Small inconsistencies between mersenne.org and mersenne.ca factor databases GP2 Data 44 2016-06-19 19:29
mersenne.ca (ex mersenne-aries.sili.net) LaurV Information & Answers 8 2013-11-25 21:01
Gaussian-Mersenne & Eisenstein-Mersenne primes siegert81 Math 2 2011-09-19 17:36
Mersenne Wiki: Improving the mersenne primes web site by FOSS methods optim PrimeNet 13 2004-07-09 13:51

All times are UTC. The time now is 17:54.

Thu Jul 9 17:54:49 UTC 2020 up 106 days, 15:27, 1 user, load averages: 1.61, 1.78, 1.78

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.