PrimeGrid
Please visit donation page to help the project cover running costs for this month

Toggle Menu

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise
1) Message boards : Number crunching : Credit Milestones (Message 152413)
Posted 4 days ago by Profile GellyProject donor
2 billion
2) Message boards : Number crunching : Crunching on paper? (Message 151224)
Posted 102 days ago by Profile GellyProject donor
Hello everyone.
I'm joking about paper, of course,


Oh beans.


but I have a little experience in programming microcontrollers and have an idea to make a computing unit using a handful of logic or PLD. Something like an ASIC, but for a primegrid task.
Yes, I know what you think, but I need your help.
I have experience in programming, but little experience in mathematics. Which project is the easiest to describe as the movement of numbers on paper? Yes, every project has a forum page and wiki page, but ... Little math experience, yes.


This has come up before in discussion on the Discord, and if you're willing to scour through source code a little bit for things like LLR2 by Atnashev, you'll find it's a lot of FFT and such that can, as far as I'm aware, be turned into an ASIC.

The problem has almost always been, even if these were possible to make, the cost would be prohibitive
3) Message boards : Number crunching : Crunching on paper? (Message 151223)
Posted 102 days ago by Profile GellyProject donor
I have been tempted to do something similar with PRP tests and Proth tests on paper, but for much smaller primes. The current record for the largest calculated prime not on computer is likely 2^148+1 / 17, which is 44 digits.

Doing similar for a 50 digit Proth prime, one would first want to do a Fermat PRP-3 test on whatever q = k*2^n+1 you pick. The way I would try to do this is to first calculate the number into decimal digits, then into binary. Next, you do the squaring-multiplication method of exponentiation for 3^(q-1), taking the modulo of your candidate q at each relevant step. For 50 digits, this is about 170 steps. At the end of all that, if the result is 1, then it's PRP.

For a Proth test, which would be more conclusive, you first want to find a base a such that the Jacobi symbol of a over q is -1. The Jacobi symbol can be calculated by hand "relatively easily", and so you go through small numbers like 2, 3, 5 etc. until you find one. Once the base is found, it's simply a matter of calculating a^((q-1)/2) mod q, with the same squaring-multiplication method as before (the binary still will come in handy for this as well!). If you get q-1 at the end of all _that_, then it's prime.

There is a reason nobody has really considered doing this, as it's a lot of squaring 50-digit numbers 300 ish times and then dividing those squared numbers by q until you get a remainder for each step as well. If you really are interested, it may be practical to start with smaller digit numbers before going for a record like this I guess.

For actual primegrid candidates, since you have millions of digits, you may as well give up while you're ahead.
4) Message boards : Number crunching : Credit Milestones (Message 149560)
Posted 256 days ago by Profile GellyProject donor
1 billion. hilarious.
5) Message boards : The Riesel Problem : K=2293 reported in top 5k page (Message 148973)
Posted 281 days ago by Profile GellyProject donor
Hmm, US privacy laws are liberal. With 5 minutes of google I found out his age, which party he is a registered voter for, the house he lives in and which sector he's probably working in and a rough estimate of the income.

:)


I sure hope I am misinterpreting this as a "jokey" threat, because if it was intended that way it was in poor taste. At the end of the day, Ryan Propper did all his own sieving work and found the prime with no assistance from Primegrid. It may be a bit tasteless from our perspective, but considering he's helped the conjecture with just his own work, he is a valuable resource to the prime finding community and should be respected as such.

Primegrid is meant to be collaborative, not competitive, and if we fail to see that in terms of this situation, we become an enemy of prime finders everywhere for no other reason than sour grapes. Don't do this.
6) Message boards : Number crunching : Credit Milestones (Message 147650)
Posted 317 days ago by Profile GellyProject donor
Obviously posting Credit Milestones and acquired badges is a job for the PG web server. It could be a front page item like the newly reported primes, under the list of Top Crunchers. Maybe with links to the latest Credit Milestones & Badges forums. What constitutes a milestone could be a matter of preference. Comments anyone?


pick whatever numbers you like, honestly. some people have grander objectives than others - i'll be waiting to post 1B credit instead of doing updates on 800m, 900m, etc. i like people being able to celebrate whatever arbitrary lines they like and not having it be a server announcement. it's more human that way, whatever that means.
7) Message boards : Antarctic Crunchers message board (Message 147591)
Posted 320 days ago by Profile GellyProject donor
wooooo, spring break
8) Message boards : Number crunching : Good Riddance 2020! Challenge (Message 147564)
Posted 321 days ago by Profile GellyProject donor

Yawn...... these challenges have just descended into who can hire the most pc's to crunch with. I don't even bother looking at the ranks any more.


Original post was pretty bitter. All I'll say is this:

It's always been "pay-to-win". The difference is I don't have to keep the expensive hardware in my apartment.
9) Message boards : Number crunching : Credit Milestones (Message 147304)
Posted 333 days ago by Profile GellyProject donor
500M. Blasted past it while running some TSC instances against WW 1.05
10) Message boards : General discussion : Extended generalized Fermat prime? (Message 146710)
Posted 350 days ago by Profile GellyProject donor
ECPP is achingly slow. There is a reason the official version of primo only goes up to 50k decimal digits. The current record, 40k digits, was set by Paul Underwood on 48 thread hardware that took more than a year and was nothing to sneeze at. My 3970x threadripper, a 32 core, 64 thread beast, takes two weeks to months to prove candidates in the 20k digit range.

As a rule of thumb, when you double the amount of digits in the number to be tested, the time it takes goes up 16 to 32 times, depending on how lucky you are. It will likely be years before Paul proves the next big step in ECPP (the smallest unproven repunit, 49k digits), and I'm sure he either got a 3990x (64 cores!) or maybe even doshed out on EPYC.

100k digits, on current hardware with current methods, would take decades.


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2021 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 1.26, 1.33, 1.55
Generated 1 Dec 2021 | 22:44:19 UTC