I can run Mlucas on my SGI. ..sigh.. too bad only that the verification will finish in 2013.
(no fault of Mlucas, of course. It's just that my SGI is 9 (!!) years old.)

[QUOTE=Uncwilly;140116]The verification run [U][COLOR="Orange"]has to be run on a different hardware [B]platform[/B][/COLOR][/U]. Not just a different physical machine.
Jeff and Tony are being given time on machines by their workplaces (IIRC).
A dual CPU, quadcore, (total of 8 cores) machine of the x86 class (Core 2 Duo or what ever) can't do a true 'verification'. That would same software and same hardware.
Running M/GLucas on a different platform (like an Itanium, 68xxx, or a Cray) is a real verification. Tony's system is a 'cluster' not a box. It is being shared and the time is free, so we have to take what we are given, unless someone wants to pay to run a verification.[/QUOTE]So that discounts the latest Apple boxes also. I wonder what happens when, in a few years, all the Itanium, 68xxx, etc. CPU lines are retired and all the new systems are x86 based (even the top supercomputers are starting to move toward x86 systems).

FWIW: Can AMD be considered different enough from Intel to serve as an independent platform? Since one box can be NUMA and the other not. Finding different software should not be a problem.

[QUOTE=retina;140121]
FWIW: Can AMD be considered different enough from Intel to serve as an independent platform? Since one box can be NUMA and the other not. Finding different software should not be a problem.[/QUOTE]

I played around with GLucas on a 32-core (8 quadcore cpu) Opteron system. Using more than 8 cores slows down the computation, probably because of NUMA memory bandwidth between processors. Testing an exponent would take about 2-3 weeks, so about the same as the verifications running.

Greg

We've been slash-dotted
 

[url]http://science.slashdot.org/science/08/08/27/2220226.shtml[/url]

At the time of this writing, top of the front page, too.

Jester

I was wondering how long that would take.

As usual they make up facts and call it a "world record prime number".

It isn't yet necessarily a world record. Nobody who should know has indicated that it is.

[quote=ADBjester;140136]We've been slash-dotted.[/quote]

Actually, we've already been wikipedia'd: [URL="http://http://en.wikipedia.org/wiki/Mersenne_prime"]http://en.wikipedia.org/wiki/Mersenne_prime[/URL] (below the table)

[QUOTE=ckdo;140149]Actually, we've already been wikipedia'd: [URL="http://http://en.wikipedia.org/wiki/Mersenne_prime"]http://en.wikipedia.org/wiki/Mersenne_prime[/URL] (below the table)[/QUOTE]

Heh, I was the first one that added that information!

[QUOTE=Wikipedia]According to GIMPS protocol, the identity of the number is not officially disclosed unless it has been verified to be a prime.[/QUOTE]

That doesn't sound right. If the verification runs show that the number is in fact not prime, isn't the number disclosed anyway?

I'm pretty sure it's not disclosed. I think it's to prevent embarrassment to the "discoverer" in the event of a false positive, which is what happened several years back.

See this thread for more information: [url]http://mersenneforum.org/showthread.php?t=672[/url]

[QUOTE=Prime95;140115]There are two verifications running. Current estimated completion dates are September 8 and 12. These may change as more or less cores become available, parallel tuning, etc.[/QUOTE]

What machines are being used for the verifications? That isn't too much of a hint, is it? :wink:

[QUOTE=jinydu;140160]What machines are being used for the verifications? That isn't too much of a hint, is it? :wink:[/QUOTE]
I can tell you that one of the verifications is being performed on a Unix server with at least one cpu core. Sorry, but you'll have to wait for more information :smile: