![]() |
|
|
#1 |
|
Oct 2005
Italy
3·113 Posts |
I 've seen that Mersenne primes are checked at least 3 times:
1 by the discoverer 1 by Tony Reix 1 by Guillermo Ballester No other checks are done ? How long does the fastest supercomputer (136800 GigaFlops) take to verify the largest Mersenne known (with prime95) ? Thanks |
|
|
|
|
|
#2 |
|
Jul 2004
Mid Calder, Scotland
5·37 Posts |
I don't know the answer to your question. But I noticed that the blue supercomper you're referring to has DOUBLED in capacity!!! Check it out
http://www.localtechwire.com/article.cfm?u=12630 |
|
|
|
|
|
#3 |
|
Oct 2005
Italy
3·113 Posts |
[QUOTE=Bundu]I don't know the answer to your question. But I noticed that the blue supercomper you're referring to has DOUBLED in capacity!!! Check it out
Yes , now it should be 287,334 GFlops/s (they didn't build the entire computer before ... ) If we know how many flop a 10 million digit prime requires (totally) we can compute 287334 / 10-million-prime-flops = how many seconds Blue Gene L requires Why don't GIMPS or somebody borrows this supercomputer for one day and check as many exponents as possible ? GIMPS current speed: 19686 Gflop/s Blue Gene L speed: 287334 GFlop/s Blue Gene L = 14.595 * GIMPS
|
|
|
|
|
|
#4 |
|
Aug 2002
26×5 Posts |
It doesn't work like that. Using Gigaflops to compare performance between totally different architectures is for the most part meaningless.
|
|
|
|
|
|
#5 |
|
Dec 2003
Hopefully Near M48
2×3×293 Posts |
Also, I think it is very unlikely that Prime95 has been optimized to work well on Blue Gene.
|
|
|
|