[RC5] Speed.

Peter A. DeNitto denitto at llamas.net
Fri Mar 20 16:31:41 EST 1998


On Fri, 20 Mar 1998, Justin Chant wrote:

> For a certain class of problems FLOP is a good measure of performance.
> This class of problems (floating point intensive) includes the type of
> problems that the super computer was designed for.  If you are designing
> a super computer or DSP FLOPs is everything.  If you are designing a
> desktop PC then you should look to a different benchmark (SpecInt
> maybe).
> 
> As for the original inquiry I think that there have been references made
> in the past to the equivalent computing power of dnet.  (I think the
> number was something on the *order* of 30,000 PIIs)
> 
> Justin Chant
> -------------------
> > >I was reading this article in byte, and they said that the fastest 
> > >computer broke 1 teraflops per second.  It is made from 9000 Pentiums
> > 
> > >(don't know exactly which ones, think P II 233).
> > 
> > 	Measuring FLOPS has got to be the most useless performance
> > indicator I've ever heard of.  The only reason I can see them bragging
> > about it is that the MIPS were laughable.

Well, using the current speed stats and a file Nugget gave me out of the
stats server, I did some calculations from extrapolating data from our RC5
speed.  

What I came up with was that right now d.net encompasses the work of
89k P100's or 31k 604e 200's... Which is a more correct representation
that just taking current speed/speed for the machine.  This one factors in 
SpecINT numbers as well.

Remember this is a real ROUGH approximation.

--Pete

--
Help Crack Government Encryption - Join the Bovine Project: www.distributed.net
The cows are tromping all over America. Cows can replicate. They're
being born all over the world. There's plenty of beef available!
		-- US Representative Zoe Lofgren (D-Ca)
--
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