[RC5] dnet processing power?

Sat Apr 27 17:09:44 EDT 2002

top500.org has a widely used algorithm for
comparing across platforms, but again, it is
based on high-level language compilers 
translating mathematic operations into processor
ops--very different on each machine.

A crude comparison is available at
stats.html.  This puts d.net at
about the 20 TFLOP mark.

All of Dennis' comments regarding the accuracy of 
this figure apply.

Bruce Ford

----- Original Message -----
From: "Dennis Lubert" 
Sent: Monday, April 22, 2002 5:35 PM
Subject: Re: [RC5] dnet processing power?

> At 20:14 21.04.02 -0700, you wrote:
> >A new computer in Japan is cabable of 35,600 
gigaflops, now how would
> >dnet compare to this?
> Well, its hard to compare dnet to it. gigaflops 
means "billions of
> POINT operations per second" but dnetc uses 
integer arithmetics (like
> bitshifting/rotating for rc5). Another had 
thing to compare is different
> architecture. On some processors one rc5 cycle 
might need 200 processor
> cycles, while on another it needs 600 or even 
more. So one can't really
> calculate the "mips" from rc5 keyrates. rc5 is 
generally a bad benchmark
> value.

To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest

More information about the rc5 mailing list