mersenneforum.org Getting others to do the work on exponents I like (was: Trial Factoring Progress)
 Register FAQ Search Today's Posts Mark Forums Read

 2021-04-28, 23:53 #100 Uncwilly 6809 > 6502     """"""""""""""""""" Aug 2003 101×103 Posts 225518 Posts Report your stats about every 330 runs (1 exponent & 1 bit level = 1 run) If you do that, I bet your data will start looking like what we expect. You should be seeing ~ 4 factors for each 330 runs. P-1 number of factors will depend on your bounds. But figure about 1 in 25 or 1 in 30.
2021-04-29, 00:25   #101
tuckerkao

Jan 2020

2·3·41 Posts

Quote:
 Originally Posted by Uncwilly Report your stats about every 330 runs (1 exponent & 1 bit level = 1 run) If you do that, I bet your data will start looking like what we expect. You should be seeing ~ 4 factors for each 330 runs. P-1 number of factors will depend on your bounds. But figure about 1 in 25 or 1 in 30.
I always use B1 = 1,000,000 and B2 = 40,000,000. I specifically looking for the combination of F 2^77 to 2^78 after NF-PM1 with the specified B1 and B2 for the exponents of M168,xxx,xxx. I realize that the chance of getting a factor is smaller in this case than directly performing the trial factoring from 2^77 to 2^78 without P-1. Of course the candidate will have no other factor below 2^77.

I guess it'll probably be only 2 to 3 factors for each 330 runs. Also how likely will there be a factor from 2^78 to 2^79 afterward? 330 tests from 2^77 to 2^78 will take quiet long time, then 2^78 to 2^79 will be double of that, bottom line is will Nvidia RTX 3090 be faster enough for the 330 runs per month?

Last fiddled with by tuckerkao on 2021-04-29 at 00:59

2021-04-29, 00:59   #102
Uncwilly
6809 > 6502

"""""""""""""""""""
Aug 2003
101×103 Posts

61·157 Posts

Quote:
 Originally Posted by tuckerkao I always use B1 = 1,000,000 and B2 = 40,000,000. I specifically looking for the combination of F 2^77 to 2^78 after NF-PM1 with the specified B1 and B2 for the exponents of M168,xxx,xxx.
Running P-1 before 79 will reduce your chances of finding a factor with TF.
Use https://www.mersenne.ca/prob.php to figure your chances finding a factor.

Quote:
 Also how likely will there be a factor from 2^78 to 2^79 afterward?
The normally expected number of factors is 1/(bit) (the bit level you are going to).

2021-04-29, 01:36   #103
tuckerkao

Jan 2020

F616 Posts

Quote:
 Originally Posted by Uncwilly Ok, let's teach you how to fish a little (learn how to estimate these numbers.) 19,960 Take those to 77 bits, 259 factors, add those at 77 (15) 19,716 Take those to 78 bits, 252 factors, add those at 78 (16) 19,480. Each of those P-1's will need about 30.8 Ghz-days.
Using the data chart you provided from an earlier post, it seemed like I've done quiet a bit of trial factoring and more during April 2021 -
https://www.mersenne.ca/status/tf/0/20210201/4/16800

How many trial factoring assignments will be considered as influential enough to make a big difference?

Nearly all of those exponents factored up to 77 and 78 have been my guesses. I also would like to have an estimate on what year will the mass PRP testing groups reach this region?

Last fiddled with by tuckerkao on 2021-04-29 at 01:43

 2021-04-30, 01:52 #104 tuckerkao   Jan 2020 2·3·41 Posts 1 more new factor found between 2^74 and 2^75, always this bit level for a decimal 3 ender and a dozenal 7 ender - https://www.mersenne.org/report_expo...8203323&full=1
2021-05-03, 11:03   #105
tuckerkao

Jan 2020

2×3×41 Posts

Quote:
 Originally Posted by Uncwilly If no prime is found before 104,000,000 and there is one found before 168,000,000, will you admit that you are wrong, your predications are foolish, and leave the forum in shame?
The only condition would be that many other people have run the PRPs for me in the M168,xxx,xxx range and nothing interesting returns.

As of today, I'm sure there will be no Mersenne Prime between M103M to M104M. Less likely the chance to have a hidden Mersenne Prime in between M104M to M168M since more PRPs have been conducted since our previous conversation.

At least none of my PRPs have met this type of fate, so I'm so far happy about the exponent guessing performance -
https://www.mersenne.org/report_expo...5582541&full=1

That "Architects Cubed" did the same thing to me, but unsuccessful -
https://www.mersenne.org/report_expo...exp_hi=&full=1

I only need several months of time, so I'm able to factor up all of my guesses in the M168,xxx,xxx ranges, hopefully I can get a new PC without the cooling problems for both CPU and GPU, AMD maybe the better solutions since they install more fans than the competitors.

Last fiddled with by tuckerkao on 2021-05-03 at 11:18

2021-05-03, 17:16   #106
S485122

Sep 2006
Brussels, Belgium

2·5·167 Posts

Quote:
 Originally Posted by tuckerkao The only condition would be that many other people have run the PRPs for me in the M168,xxx,xxx range and nothing interesting returns. ...
Nobody will run those PRP tests for YOU. They will just run PRP tests on exponents handed out by PrimeNet.
Those exponents, once you finished working on them or fall foul of the assignment rules, are nobody's until assigned by PrimeNet.
If one of those would correspond to Mersenne prime, only the person or group having done the first known PRP (or LL) test will be celebrated as the ”discoverer”. People having done TF, PM1 and whatever on that exponent will be included in the announcement, but it will be in the ”et all” together with all those who worked on other numbers.
You seem to be unable to understand that we are all working together. Nobody is working for you.

Jacob

2021-05-05, 01:00   #107
tuckerkao

Jan 2020

F616 Posts

Quote:
 Originally Posted by S485122 Nobody will run those PRP tests for YOU. They will just run PRP tests on exponents handed out by PrimeNet. Those exponents, once you finished working on them or fall foul of the assignment rules, are nobody's until assigned by PrimeNet. Jacob
If I finish the trial factoring up to 2^78 and P-1 with the recommended bounds for some of the M168,xxx,xxx exponents, PrimeNet will assign them to other users for PRP tests once the exponents hop to Category 1 while GPU72 reserves all the remaining unfinished exponents which means unless other users try to poach them at the specific future time, my guesses will get the service earlier under the automated system if not by myself already. I will run the PRPs on some of them beforehand, but probably not be able to finish all within the time frame. I was glad Ben Delo completed so many of them in the M103,xxx,xxx range with the warp speed operations because they all turned out to be composite. At least Ben Delo acted like a robot, so he wouldn't be upset at me.

I understand the prize reward will only be given to the discoverer who run the first PRP of the new Mersenne Prime. All I did was to try to figure out the patterns of GPU72 and worked around them to get PrimeNet assigning out my exponent guesses to other users quicker.

For example M168.314,323, GPU72 will never reserve an assignment on this exponent because I've done all the corresponding tasks already -
https://www.mersenne.org/report_expo...exp_hi=&full=1

Last fiddled with by tuckerkao on 2021-05-05 at 01:26

 Similar Threads Thread Thread Starter Forum Replies Last Post LaurV GPU to 72 81 2020-12-02 05:17 jfamestad PrimeNet 3 2016-11-06 20:32 lidocorc PrimeNet 4 2008-11-06 18:48 jocelynl 15k Search 0 2003-07-11 14:23 ThomRuley Software 5 2003-05-30 20:34

All times are UTC. The time now is 03:21.

Sat May 15 03:21:09 UTC 2021 up 36 days, 22:02, 0 users, load averages: 2.40, 2.02, 2.13