mersenneforum.org  

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

Reply
 
Thread Tools
Old 2018-12-10, 14:30   #34
pinhodecarlos
 
pinhodecarlos's Avatar
 
"Carlos Pinho"
Oct 2011
Milton Keynes, UK

2·3·5·157 Posts
Default

Quote:
Originally Posted by axn View Post
What is the progress/ETA on the verification runs?

And, what is the plan on the press release? Pre-christmas or post new year?
I’ve verified on my TI-89, just awaiting for the slowest ones.
pinhodecarlos is offline   Reply With Quote
Old 2018-12-10, 14:48   #35
GP2
 
GP2's Avatar
 
Sep 2003

22·3·5·43 Posts
Default

Quote:
Originally Posted by henryzz View Post
Is it prime or prp currently? This could the first using the new testing methodology
It must have been found by LL test, because we know that Mlucas and CUDALucas are being used for the verification, not gpuOwL.

However, it might be worthwhile to retest every known Mersenne prime with PRP.

Currently our verification method is to use different software implementations of the same Lucas-Lehmer algorithm on different hardware (GPU vs CPU). PRP, on the other hand, is a different algorithm.

No, it can't prove primality. But it can prove compositeness. And at this point, with such an improbable surplus of Mersenne primes, maybe a drastic sanity check is needed for peace of mind, just in case software implementers exchanged ideas and somehow ended up all making some identical implementation error. It's a crazy notion, I know.
GP2 is offline   Reply With Quote
Old 2018-12-10, 15:07   #36
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

441710 Posts
Default

Quote:
Originally Posted by Prime95 View Post
On Dec 7th, a new Mersenne prime was reported. I emailed the discoverer, obtained the last save file, and reran the last 100,000 iterations. Sure enough, we have a new prime!

Aaron has started a full test using prime95, but that doesn't count as an official double-check. We need a volunteer to run CudaLucas and/or mlucas.
If it still matters, you have my standing offer from some time ago. Current fastest hardware is GTX1080Ti ready with CUDALucas v2.06 May 5 2017 beta, capable of knocking out a ~82M exponent in about 3 days, backed up by a GTX1080, and a 16-core e5-2670 system with prime95 V29.4b8. The GTX1080Ti at 84+GhzD/day for 85M is ~10% faster than a 2080 Ti for LL per https://www.mersenne.ca/cudalucas.php (14th fastest model in the list, and that includes combined throughputs of some duals and a quad; 10th fastest single). And no, that 82M or 85M is not a clue to the nature of Mp51; it's just what I checked the 1080Ti's run time on recently. It would take me some time to put up mlucas on the system.

Last fiddled with by kriesel on 2018-12-10 at 15:29
kriesel is online now   Reply With Quote
Old 2018-12-10, 15:19   #37
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

7×631 Posts
Smile

Quote:
Originally Posted by Prime95 View Post
As usual, the automatic email notification to myself and past Mersenne prime discoverers failed. Yes, the automatic email was debugged and tested after its last failure. Apparently bit rot set in.
Tradition continues!
kriesel is online now   Reply With Quote
Old 2018-12-10, 15:26   #38
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

22·7·307 Posts
Default

I've noticed some interesting names with some interesting exponent assignments that will be "home for the holidays" as it were.

MD5 hash guess of one that looks tasty (includes exponent assignee MM/DD expected completion date)
227a10ba5bbe1b4b033c2983ee417ee9

It does fit the 2 requirements posted by George and Alex. There also appears to be some funny business in the history on the assignment page.

Last fiddled with by Uncwilly on 2018-12-10 at 15:28
Uncwilly is offline   Reply With Quote
Old 2018-12-10, 15:50   #39
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

2×1,637 Posts
Default

The first verification run done by ATH confirms it: We have a new prime!

I'm not sure when George will do the press release - as was pointed out, the best way to predict when a prime is found is when George is on vacation.

I'll say a few things:
  • It was found by LL, not PRP. Although as noted, I think if we found one by PRP we'd do a pair of LL tests to confirm.
  • It's more than 80M and less than 90M
  • The discoverer has not found any previous primes
Madpoo is offline   Reply With Quote
Old 2018-12-10, 16:14   #40
irowiki
 
Sep 2018

3·23 Posts
Default

So just to be sure, I set the lion's share of my machines to PRP. Should they be doing LL instead, or is PRP fine?
irowiki is offline   Reply With Quote
Old 2018-12-10, 16:23   #41
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

7,151 Posts
Default

Quote:
Originally Posted by irowiki View Post
So just to be sure, I set the lion's share of my machines to PRP. Should they be doing LL instead, or is PRP fine?
PRP is a good way to go. I've moved half of my CPUs to PRP, eventually all will PRP.
Prime95 is offline   Reply With Quote
Old 2018-12-10, 17:31   #42
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

CCA16 Posts
Default

Quote:
Originally Posted by irowiki View Post
So just to be sure, I set the lion's share of my machines to PRP. Should they be doing LL instead, or is PRP fine?
Smarter folks than I can weigh in, like George did in mentioning that PRP is the future...

Here's my understanding: PRP is probabilistic and LL is deterministic. So although a PRP can say "M###" is probably prime, an LL test can say "M###" is prime.

PRP tests have the benefit of the Gerbicz error checking code that may make double-checking the tests unnecessary.

As to just how probable the PRP tests are, I understand that when PRP says it's probably prime, that probability is VERY high. And when I say the PRP error checking makes double-checks obsolete, I think that's generally true although I foresee cases where spot-checks make sense, but overall that's where I think it's headed. If I'm wrong in my assumptions, someone will correct me I'm sure.

When it comes to finding a new Mersenne Prime, I guess the question about whether a probable finding is good enough to be accepted, or would the LL test be required before it's confirmed. Maybe that's similar to how we already verify a prime using different software - it's a way of ensuring there aren't any code issues or other obscure things.
Madpoo is offline   Reply With Quote
Old 2018-12-10, 17:55   #43
Madpoo
Serpentine Vermin Jar
 
Madpoo's Avatar
 
Jul 2014

2·1,637 Posts
Default

The mersenne.org homepage has been updated with a little blurb about the discovery, linking to this thread. More of a pre-release release with a promise of details to follow.
Madpoo is offline   Reply With Quote
Old 2018-12-10, 18:28   #44
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

7×631 Posts
Default PRP and other runs on already-known mersenne primes

Quote:
Originally Posted by GP2 View Post
It must have been found by LL test, because we know that Mlucas and CUDALucas are being used for the verification, not gpuOwL.

However, it might be worthwhile to retest every known Mersenne prime with PRP.
Looking back through mersenne.org, I see some exponents determined prime also have (anonymous) incorrect composite LL indications; https://www.mersenne.org/report_expo...exp_hi=&full=1
https://www.mersenne.org/report_expo...exp_hi=&full=1
https://www.mersenne.org/report_expo...exp_hi=&full=1
https://www.mersenne.org/report_expo...exp_hi=&full=1

Quite a few Mersenne primes also have recent failed TF or P-1 attempts also.

These following PRP runs on already known mersenne primes have not been reported yet to mersenne.org, but could be if it were useful. I may assemble them into my gpuowl reference data instead or also. Running known mersenne prime PRP checks was done as QA on gpuowl new versions/features.

gpuowl v5.0-df2bdf2 build for Win7 x64 PRP with P-1, p= 216091, 756839, 859433 by kriesel
https://www.mersenneforum.org/showpo...&postcount=867
Per Preda, for a prime, residue = PRP-1 base, see
https://www.mersenneforum.org/showpo...&postcount=826
https://www.mersenneforum.org/showpo...&postcount=823

gpuowl v3.8-91c52fa p=1257787
https://www.mersenneforum.org/showpo...&postcount=827

(p=756839, example of PRP GEC in https://www.mersenneforum.org/showpo...&postcount=815)

gpuowl v5.0-f604bb1 p=1257787, 1398269, 2976221, 3021377 by kriesel
https://www.mersenneforum.org/showpo...&postcount=818

Also, unposted(?), gpuowl v5.0-9c13870 p=6972593

Last fiddled with by kriesel on 2018-12-10 at 18:48
kriesel is online now   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Lucky number gaps robert44444uk Prime Gap Searches 23 2019-08-23 20:45
Fun with the Lucky Numbers of Euler ewmayer Probability & Probabilistic Number Theory 0 2015-10-18 01:37
Extremely lucky assignments apocalypse GPU to 72 6 2015-04-07 04:41
Lucky ECM hit Dubslow Factoring 3 2014-10-19 19:10
Lucky gmp-ecm curve... WraithX GMP-ECM 4 2009-01-12 16:29

All times are UTC. The time now is 01:32.

Tue Sep 22 01:32:41 UTC 2020 up 11 days, 22:43, 0 users, load averages: 1.55, 1.52, 1.56

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.