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

2021-08-16, 19:38   #232
tuckerkao

"Tucker Kao"
Jan 2020

1F016 Posts

Quote:
 Originally Posted by Viliam Furik There is an output from the mfaktc program, you know? You just copy-paste it...
I don't know whether running mfaktc while playing the hard core video game will cause the program to miss a factor or not. I typically only open the internet browser or/and do other simple tasks during the meantime.

It took around 4~5 hours to TF from 2^78 to 2^79 for M168202123 on 3070 Ti. What's the highest bit that your GPU can complete the TF for this exponent and finish within 24 hours?

If I buy a new PC, I want its GPU to finish up 2^81 to 2^82 less than a full day.

Last fiddled with by tuckerkao on 2021-08-16 at 19:55

2021-08-16, 19:55   #233
Uncwilly
6809 > 6502

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

2·3·1,669 Posts

Quote:
 Originally Posted by tuckerkao It took around 4~5 hours to TF from 2^78 to 2^79 for M168202123 on 3070 Ti.
According to mersenne.ca only 78 bits is recommended for the 160-170 range. So, during that time you could have done another exponent from null to 78 bits. It is better to take all the exponents in a desired range to the recommended bit level before taking any beyond that level. You will find an appropriate number of factors and get the biggest bang for your money. Also, doing so will allow any newer hardware do more of the heavy lifting. This is part of the strategy that has been employed in some other ranges.

2021-08-16, 20:03   #234
tuckerkao

"Tucker Kao"
Jan 2020

7608 Posts

Quote:
 Originally Posted by Uncwilly According to mersenne.ca only 78 bits is recommended for the 160-170 range.
The recommended TF bit will rise to at least 2^81 for the exponents above M333M, I want the new GPU be able to do that on the daily basis, also the new CPU be able to finish P-1 before the last bit once per day.

If I get a chance to temporarily operate someone else's hardware, I'll hop on the bits higher than what my current PC be able to finish efficiently, otherwise I just use my current old PC.

Quote:
 Originally Posted by tuckerkao Only 109 days remain until AMD Threadripper 5970X launches, so I'll patiently wait for that, then I'll have the computing powers I need at latest by mid-Jan 2022. If these are the highly composite grounds, Z48,622,147 <- M168753223 Z48,62Ӿ,127 <- M168767023 Z48,65Ӿ,897 <- M168830323 Z48,673,Ɛ8Ɛ <- M168860123 Z48,6Ӿ6,20Ɛ <- M168926123 then the highly primitive ground should be at Z48,3Ɛ7,237 <- M168202123 I'll try do it all myself, see if I'm correct this time.
I surely picked a much more primitive exponent using this dozenal strategy, only wanted to fuel in more motives, so I could finish up more.

Last fiddled with by tuckerkao on 2021-08-16 at 20:14

 2021-08-16, 20:14 #235 Uncwilly 6809 > 6502     """"""""""""""""""" Aug 2003 101×103 Posts 1001410 Posts As is so often the case, you did not even address what I said. You quoted it, then went off on a tangent.
2021-08-16, 20:21   #236
tuckerkao

"Tucker Kao"
Jan 2020

1F016 Posts

Quote:
 Originally Posted by Uncwilly As is so often the case, you did not even address what I said. You quoted it, then went off on a tangent.
I've finished up to only 2^78 for many M168M exponents including M168412723 and M168424723, have no plans to go further. I thought it was fun to stack up a skyscraper like Drkirkby's M105211111 on only 1 exponent and called it as his primary representation, then he started to be able to test only the Cat 4 exponents. Chalsall landed on that and paid a huge rent for it.

I generally don't run TFs beyond the recommendation levels, only ViliamF enjoys doing that and tells me finding more factors will be really impressive, see example M103332391.

M120202123 reached the desired bit few minutes ago, how long will it take for someone to be willing to run a PRP of it? The supplies seem to be always way higher than the demands on the exponents I like.

Last fiddled with by tuckerkao on 2021-08-16 at 20:43

2021-08-16, 22:50   #237
Viliam Furik

"Viliam Furík"
Jul 2018
Martin, Slovakia

10101010102 Posts

Quote:
 Originally Posted by tuckerkao I generally don't run TFs beyond the recommendation levels, only ViliamF enjoys doing that and tells me finding more factors will be really impressive, see example M103332391.
I think that was to factor that specific exponent.

2021-08-18, 06:59   #238
tuckerkao

"Tucker Kao"
Jan 2020

24·31 Posts

Quote:
 Originally Posted by Uncwilly It is better to take all the exponents in a desired range to the recommended bit level before taking any beyond that level.
I take the precaution that before I formally claim M168202123 as my Headquarter or Center of Operation, I factor it up a bit more, so ViliamF won't come back and topple it by finding a possible factor that I may barely miss myself, then laugh endlessly to a heart attack. M168202123 seems to be very factor-resistant, has already passed the stage 1 of P+1 factoring(see screenshot below).

My theme has been consistent during the recent days: Search M168,***,*23 starting year 2021, so M168202123 will be a logical location for me to establish my primary asset. The professional Mersenne Prime search operation is indeed many years long, thus an excellent starting location is a must-have, call it a place of rest and hopefully a billionaire like Ben Delo will tour my avenue someday later this year.

It's easier for me to own my unique range and a central area with lots of missions and/or tasks accomplished. Search of Mersenne Prime will be somehow more convenient. I'm used to the way that there's a "Town Hall" in the "Forge of Empires" or "Kingdom Reborn". The random seeds of the map generator contain 9 numerical digits, thus perfect to type the Mersenne exponents in. "Command Center" is typically the very first building available to the players in the Skirmish game in the Command & Conquer series. Since there are tons of available exponents in the higher ranges, every Mersenne user will be able to find his/her own desired region(s) to settle down.

Quote:
 Originally Posted by Viliam Furik Stop it or I'll get a heart attack from laughing too hard... Even though I'm 18. (that is a bad joke, don't take it seriously) If you somehow get a P-PRP result, "GIMPS prime discovery protocols" will come into action and a bunch of undisclosed trusted individuals will be tasked with running an LL test to verify the discovery.
According to the numerical arrangements in the dozenal base, M168202123 should survive the trial factoring up to the 2^82, but I still need the real decimal evidence that it's true. The only way is to borrow the Geforce 3070 Ti of 1 of my friends on certain Sundays. It'll be the best if the property value of my headquarter eventually worth more than Drkirkby's M105211111.

I enjoy to play Monopoly with my friends, owning an entire row of properties is a good way to win the championship. With the Mersenne numbers, owning a chain of exponents of M168,202,*** will show the works that I have done in a clearer manner.

If the trial factoring of the Mersenne exponents can display the possible tasks into several categories with the enhanced levels of the gaming graphics such as tech trees -> "2^73 to 2^74", "2^74 to 2^75", "2^75 to 2^76", "P-1 with the recommended bounds", grayed-out choices indicate that the prerequisite haven't been fulfilled yet but the Mersenne users can still perform them.
Attached Thumbnails

Last fiddled with by tuckerkao on 2021-08-18 at 07:47

 2021-08-18, 12:28 #239 jnml   Feb 2012 Prague, Czech Republ 2648 Posts More than once you're talking about some kind of ownership of some set of numbers. I don't understand. What concept are you talking about?
2021-08-18, 14:24   #240
drkirkby

"David Kirkby"
Jan 2021
Althorne, Essex, UK

3×149 Posts

Quote:
 Originally Posted by tuckerkao Drkirkby built a skyscraper on M105211111, but many of the surrounding properties were already owned by Ben Delo, thus Drkirkby will have to find another area in the upcoming months. He picked this exponent probably because of the repunit digits pattern from the right-end, so it's easier to remember and he can come back to visit this place often.
I don't know what "built a skyscraper" was supposed to mean. All I wanted to do was benchmark testing exponents on my computer, using different amounts of RAM (8, 16 , 32, 64, 128, 256 and 320 GB), and different amounts of "saved tests" (1.0, 1.05 and 2) then see what gave the maximum performance, based on maximising the equation:

probably_of_finding_a_factor_using_P-1 * time_for_PRP_test - time_for_P-1_factoring

I preferred to do it on the same exponent, so I knew any changes in time were not due to change of exponent. No "surrounding properties were already owned by Ben Delo" Many people, including Ben and myself were working around 105 million, as it is the most sensible place to look.

If you want to find a prime number, there seems to me only two sensible places worth consideration.

1. The smallest exponents you can get, that have not had a LL or PRP test, as those have both the highest probability of being producing a Mersenne prime, and also take the shortest period of time to test.
2. Exponents somewhat over 332,000,000 which will yield at least 100,000,000 decimal digits, so nett one $50,000. Having tested one of them, M332646233, I concluded it was too much work. But at least I can see a point in why people do that. Your testing exponents around 168 million has several disadvantages. 1. If the Mersenne number happens to be prime, the number of decimal digits is insufficient to get you the$50,000 prize. All you will get is \$3,000.
2. The exponent is unlikely to be trial-factored sufficiently high, so you need to do the trial-factoring, so it takes extra time.
3. The exponent is unlikely to have any P-1 factoring, so you need to do that, so it takes extra time.
4. The exponent is less likely to yield a Mersenne prime than an exponent around 105.5 million, as it's generally believed the spacing between Mersenne primes increases with the exponent.
5. The exponent is going to take around (168/105.5)2.1=2.54x as long to test
To me at least, there's no reason to test around 168 million now.
Quote:
 Originally Posted by tuckerkao I've finished up to only 2^78 for many M168M exponents including M168412723 and M168424723, have no plans to go further. I thought it was fun to stack up a skyscraper like Drkirkby's M105211111 on only 1 exponent and called it as his primary representation, then he started to be able to test only the Cat 4 exponents. Chalsall landed on that and paid a huge rent for it.
I never called it my "primary representation". I have never come across such a term. Please, if you are going to write "drkirkby said xyz...", make sure I actually wrote it - POST A LINK. You seem to be inventing things you believe I wrote, when I never wrote them at all. You also seem to think you can read my mind, but you are not doing so.

There are one of 3 possible reasons I starting getting category 4 exponents, and none have anything to do with chalsall, nor my category 4 issue had any impact on chalsall.

2021-08-18, 14:47   #241
Uncwilly
6809 > 6502

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

2×3×1,669 Posts

Quote:
 Originally Posted by drkirkby The smallest exponents you can get, that have not had a LL or PRP test, as those have both the highest probability of being producing a Mersenne prime, and also take the shortest period of time to test.
A minor quibble. exponents in the 105M range vs the 110M or 115M range have basically no difference in probability of being a prime. While one might be able to calculate a theoretical difference, the difference is too slight to have a practical consideration, and so far the primes are so scattered and random that we have little real clue when one might be 'due'.
Quote:
His kit, his joy.

2021-08-18, 15:11   #242
axn

Jun 2003

2×52×103 Posts

Quote:
 Originally Posted by Uncwilly A minor quibble. exponents in the 105M range vs the 110M or 115M range have basically no difference in probability of being a prime.
For the same amount of TF/P-1 done, a 105M exponent has a roughly 10% higher (a priori) probability of being prime compared to a 115M exponent. Just sayin'.
EDIT:-
Quote:
 Originally Posted by Uncwilly His kit, his joy.

Last fiddled with by axn on 2021-08-18 at 15:12

 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 07:40.

Sun Oct 24 07:40:22 UTC 2021 up 93 days, 2:09, 0 users, load averages: 0.86, 1.08, 1.08