[RC5] RC5-72 on Blue Gene/L?

Jim C. Nasby decibel at distributed.net
Thu Nov 11 22:28:25 EST 2004


The last time I looked at it (which admittedly was like 5+ years ago)
there was a pretty clear distinction between CPUs that were good at OGR
and CPUs that were good at RC5. Looking at the numbers I came up with
2,000 nodes = 1 key. Even if the exact number isn't true anymore it's
still generally pretty easy to see which CPUs are good at which project.

On Thu, Nov 11, 2004 at 07:24:52PM +0000, Elektron wrote:
> 
> On 11 Nov, 2004, at 15:38, Jim C. Nasby wrote:
> 
> >On Thu, Nov 11, 2004 at 11:14:37AM +0100, Didier Levet wrote:
> >>17,250,000 x 2 x 1100 = 37,950,000,000 keys/s
> >>obtained with a sub-optimal RC5 core.
> >
> >Of course, since the G5 is better at OGR, we'd certainly have it work 
> >on
> >that instead ;P
> 
> "better at" carries little meaning in this context. If you want to do 
> both OGR-25 and RC5-72, you're better off running OGR on an Athlon, and 
> RC5 on a G5. Until you say "this many keys is worth this many nodes", 
> you can't compare speeds between the projects.
> 
> And really, OGR-25 is a lot more feasible than RC5-72, if you say 
> "completing OGR is worth as much as completing RC5". And I don't really 
> see the worth of completing RC5.
> 
> - Purr
> 
> _______________________________________________
> rc5 mailing list
> rc5 at lists.distributed.net
> http://lists.distributed.net/mailman/listinfo/rc5
> 

-- 
Jim C. Nasby, Database Consultant           decibel at distributed.net
Give your computer some brain candy! www.distributed.net Team #1828

Windows: "Where do you want to go today?"
Linux: "Where do you want to go tomorrow?"
FreeBSD: "Are you guys coming, or what?"


More information about the rc5 mailing list