And no primes... LOL

More seriously, mdettweiler, see post #230 of this thread, where the figures are correct. Philmoore´s post should be even better, in terms of the approximation taken, if it weren´t for the missing zeroes.

Oops! But with numbers that big, what difference is one more or one less zero anyway?

[QUOTE=philmoore;176765]Oops! But with numbers that big, what difference is one more or one less zero anyway?[/QUOTE]

yeah, right. Particularly when it sits at the far right of the number... :smile:

This guy seems convinced:
[URL]http://isthe.com/chongo/tech/math/prime/mersenne.html[/URL]

edit:
An interesting bio:
[URL]http://en.wikipedia.org/wiki/Landon_Curt_Noll[/URL]

[quote=Flatlander;176774]This guy seems convinced:
[URL]http://isthe.com/chongo/tech/math/prime/mersenne.html[/URL]

edit:
An interesting bio:
[URL]http://en.wikipedia.org/wiki/Landon_Curt_Noll[/URL][/quote]
Well spotted. He is also responsible for the remarkable observation
that in a sequence of random numbers, there may be clusters
(which he dubbed "islands") closer than expected on average.
Freakish as GMPS last 8 discoveries may seem, I still don't see
any need to invoke divine intervention:smile:

[QUOTE=davieddy;176783]Well spotted. He is also responsible for the remarkable observation
that in a sequence of random numbers, there may be clusters
(which he dubbed "islands") closer than expected on average.
Freakish as GMPS last 8 discoveries may seem, I still don't see
any need to invoke divine intervention:smile:[/QUOTE]

Has anyone crunched the numbers on just how likely this cluster is, assuming our models are right? If we view the region searched so far as a (scaled) Poisson test, could we reject it as nonrandom at the 5% level?

[QUOTE=davieddy;176783]Well spotted. He is also responsible for the remarkable observation
that in a sequence of random numbers, there may be clusters
(which he dubbed "islands") closer than expected on average.
Freakish as GMPS last 8 discoveries may seem, I still don't see
any need to invoke divine intervention:smile:[/QUOTE]

I hope you're being sarcastic about the "remarkable" part.

Kevin, what kind of hardware are you running your (hoped-for) DC on, and what kind of throughput are you getting?

I'll let Kevin respond with the particulars, but I believe it was a quad 9550 running Prime95 or mprime. What I find exciting is that his run on 4 cores is progressing almost as fast as Tony's 16-core Itanium run. Granted, the Itanium is an older machine, but 9.5 days is a good time either way. I need a quad core too, now!

[QUOTE=ewmayer;176786]I hope you're being sarcastic about the "remarkable" part.

Kevin, what kind of hardware are you running your (hoped-for) DC on, and what kind of throughput are you getting?[/QUOTE]

All 4 cores of a Q9550 at stock speed (2.83ghz) on 64-bit Ubuntu, and I'm getting around .018 seconds per iteration.

[QUOTE=Kevin;176788]All 4 cores of a Q9550 at stock speed (2.83ghz) on 64-bit Ubuntu, and I'm getting around .018 seconds per iteration.[/QUOTE]

Nice - that's about the same throughput Rob Giltrap's 16-core Sparc VII verify is getting, although the latter is using a 4096K FFT and compiled C code, no SSE2.

Once I finish the ongoing 64-bit linux port of my Mlucas SSE2 code and make the enhancements needed for it to run multithreaded I'll be interested to see how things scale ... Getting 0.011 seconds/iter for exponents the size of the DC one running on one core of my 2GHz Core2Duo Macbook, and 0.012 sec/iter if I run a separate LL test on each of the 2 cores. All other things being equal (which they probably aren't - I have slow ddr2 memory), at 2.83 GHz that would mean .07-.08 sec/iter/core, so if one got very good || scalability up to 4 cores, there's your 0.02 sec timing range. Amazing, this modern science. ;)

Very nice... I wish my computer would run that fast. Is 11pm Central Standard Time on Saturday still predicted to be when the first of the double checks will be completed?