mersenneforum.org  

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

Reply
 
Thread Tools
Old 2018-03-18, 08:21   #485
Dubslow
Basketry That Evening!
 
Dubslow's Avatar
 
"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

3·29·83 Posts
Default

Much like gapcoin, I've always thought a coin that finds zeros of the Riemann-Zeta function would be fun.
Dubslow is offline   Reply With Quote
Old 2018-03-18, 18:13   #486
CRGreathouse
 
CRGreathouse's Avatar
 
Aug 2006

10110111001002 Posts
Default

Quote:
Originally Posted by Dubslow View Post
Much like gapcoin, I've always thought a coin that finds zeros of the Riemann-Zeta function would be fun.
I could get in on that.
CRGreathouse is offline   Reply With Quote
Old 2018-04-03, 23:39   #487
ewmayer
2ω=0
 
ewmayer's Avatar
 
Sep 2002
República de California

101011101001002 Posts
Default

While poking around on the skywatching site EarthSky.org, happened across this 9. January piece:

http://earthsky.org/human-world/larg...ersenne-primes
ewmayer is offline   Reply With Quote
Old 2018-04-17, 21:53   #488
tServo
 
tServo's Avatar
 
"Marv"
May 2009
near the Tannhäuser Gate

1E716 Posts
Default

Quote:
Originally Posted by Dubslow View Post
Much like gapcoin, I've always thought a coin that finds zeros of the Riemann-Zeta function would be fun.
There was a D.C. project called ZetaGrid that operated around 2000 -> 2005 that did just that. It came to an untimely end when the host server crashed.
http://en.wikipedia.org/wiki/ZetaGrid
tServo is offline   Reply With Quote
Old 2018-05-17, 20:44   #489
GP2
 
GP2's Avatar
 
Sep 2003

257810 Posts
Default

M77232917 was confirmed by a bunch of programs, including gpuOwL. But Jacobi checking was mentioned, so that was an early version of the program, with LL testing rather than PRP.

It would be useful to run PRP tests on the largest known Mersenne primes, or perhaps even all of them, just to test whether the code and the server do the right thing when a discovery is made. Will the server send e-mails?

This seems like a useful sanity check and dry run. Maybe one PRP test with gpuOwL and one with mprime.

I could post worktodo lines for these.
GP2 is offline   Reply With Quote
Old 2018-12-14, 20:14   #490
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

EFD16 Posts
Default

Quote:
Originally Posted by GP2 View Post
M77232917 was confirmed by a bunch of programs, including gpuOwL. But Jacobi checking was mentioned, so that was an early version of the program, with LL testing rather than PRP.

It would be useful to run PRP tests on the largest known Mersenne primes, or perhaps even all of them, just to test whether the code and the server do the right thing when a discovery is made. Will the server send e-mails?

This seems like a useful sanity check and dry run. Maybe one PRP test with gpuOwL and one with mprime.

I could post worktodo lines for these.
It's not necessary to run the PRP tests to produce the requisite server notification test string. GpuOwL does not include the mprime/prime95 security code, so a GpuOwL PRP3 prime result record can be trivially faked. (Such a fake would of course fail the behind-the-scenes verification with multiple tests if verification was attempted.) I suggest such a server notification test attempt be done with a prior notification to the usual recipient list that a test is coming, and with the email notification mechanism unaltered, but the test string should come from a GIMPS user account created for the purpose, named something like "notification-test" to flag it as only a test. (The gpuOwL result record does not itself contain user id or computer id fields, so flagging there as a notification test is not available.) The faked exponent should be an actual prime exponent not associated with an already-known Mersenne prime, to pass certain screening tests the server applies, which include screening out primality reports for already-known Mersenne primes.

Actually running a large fraction of known Mersenne primes on the latest gpuOwL would provide both runtime scaling data, and some confidence regarding whether it is subject to false negatives. I'm in the process of doing that now. Results will appear in https://www.mersenneforum.org/showthread.php?t=23391

It's not possible to run gpuOwL PRP tests on _all_ known Mersenne primes, because gpuOwl through V5.0-9c13870 has a minimum fft length of 128K. The lowest it handles properly is M31, p=216091, with ~1.65 bits/word. (That Mersenne prime was originally found by a Cray supercomputer.)

It is possible to run PRP from p=3 up using prime95/mprime for known Mersenne primes, since prime95 has fft lengths down to 32 (words, not 32K). I've done it recently up to p~3M to generate a quick runtime scaling data set. It spins through the first 25 Mersennes on one core of an old slow i3 in a couple seconds total. See the attachment at https://www.mersenneforum.org/showpo...78&postcount=2

Last fiddled with by kriesel on 2018-12-14 at 20:17
kriesel is offline   Reply With Quote
Old 2018-12-14, 20:45   #491
GP2
 
GP2's Avatar
 
Sep 2003

2·1,289 Posts
Default

Quote:
Originally Posted by kriesel View Post
Actually running a large fraction of known Mersenne primes on the latest gpuOwL would provide both runtime scaling data, and some confidence regarding whether it is subject to false negatives. I'm in the process of doing that now. Results will appear in https://www.mersenneforum.org/showthread.php?t=23391

It's not possible to run gpuOwL PRP tests on _all_ known Mersenne primes, because gpuOwl through V5.0-9c13870 has a minimum fft length of 128K. The lowest it handles properly is M31, p=216091, with ~1.65 bits/word. (That Mersenne prime was originally found by a Cray supercomputer.)

It is possible to run PRP from p=3 up using prime95/mprime for known Mersenne primes, since prime95 has fft lengths down to 32 (words, not 32K). I've done it recently up to p~3M to generate a quick runtime scaling data set. It spins through the first 25 Mersennes on one core of an old slow i3 in a couple seconds total. See the attachment at https://www.mersenneforum.org/showpo...78&postcount=2
With the recent discovery of M51 on top of M50, I decided to just go ahead and PRP test all the Mersenne primes with exponent larger than 1M, using mprime 29.5.

For exponents smaller than 1M, the tests would take less than a minute with mprime, mostly far less. I think at one point I tested those small Mersenne primes with a very simple C implementation of Lucas-Lehmer and also PRP with a simple GMP modular exponentiation function call, but didn't properly note and record the results.

So, anyways, if you're PRP testing the non-small Mersenne primes on gpuOwL and I'm doing it with mprime, then the two efforts will complement each other. There is no doubt what the result will be, but it bugs me that this simple sanity check has apparently never been done. Or if it has, there is no record of it.
GP2 is offline   Reply With Quote
Old 2018-12-14, 21:11   #492
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

3×2,281 Posts
Default

Quote:
Originally Posted by kriesel View Post
It's not necessary to run the PRP tests to produce the requisite server notification test string.
Potential new Mersenne primes reported via the manual results web page are quietly emailed to Aaron, James, and myself. The bulk have been from CudaLucas 2.05 which can have build and/or driver issues.

That email process has been working quite well.

Last fiddled with by Prime95 on 2018-12-14 at 21:11
Prime95 is offline   Reply With Quote
Old 2019-12-21, 21:17   #493
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

3·1,279 Posts
Default

What, 4 days to Christmas, and no encore with Mp52?
kriesel is offline   Reply With Quote
Old 2019-12-22, 04:37   #494
Uncwilly
6809 > 6502
 
Uncwilly's Avatar
 
"""""""""""""""""""
Aug 2003
101×103 Posts

3×7×383 Posts
Default

Quote:
Originally Posted by kriesel View Post
What, 4 days to Christmas, and no encore with Mp52?
Curtis and Ben have determined to announce the "discovery" of the next prime on Jan 6th.
Uncwilly is offline   Reply With Quote
Old 2019-12-23, 20:05   #495
ewmayer
2ω=0
 
ewmayer's Avatar
 
Sep 2002
República de California

256448 Posts
Default

Perhaps all those Defense Department supercomputers which are normally secretly repurposed to searching for the next holiday prime around this time of year are unavailable this year, thanks to an unfortunate typo:

NORAD Tracking Satan for the Holidays due to Typo | Duffel Blog
ewmayer is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Merry Christmas everyone ... petrw1 Lounge 26 2013-01-02 23:01
Merry Christmas Everyone petrw1 Lounge 1 2010-12-25 07:54
Merry Christmas Everyone.... petrw1 Lounge 8 2009-12-25 16:09
Merry Christmas! Prime95 Lounge 17 2006-12-26 19:31
Merry Christmas ltd Prime Sierpinski Project 3 2004-12-31 07:31

All times are UTC. The time now is 08:31.

Thu Jun 4 08:31:43 UTC 2020 up 71 days, 6:04, 0 users, load averages: 1.35, 1.39, 1.35

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.