On Mon, 9 May 2011 19:15:28 +0200
ri...@happyleptic.org wrote:
> > Seriously though, take an x86 system with a processor built at the
> > same sized process (90nm?) and compare the total energy consumed to
> > perform a task.
>
> I do not have a 90nm x86 around, but I have a recent atom D510.
>
> So here are the result of the superscientific benchmark, the fight
> between a 45nm Intel Atom, 21W of horsepower, and its challenger from
> China, the 90nm(?) Loongson consuming 12W, and they are competing over
> this simple yet very acute and so representative CPU task :
>
> echo 'define f(x) { if (x == 1) return(1); return (x*f(x-1)); } f(10000)' |bc
>
> And the results are :
>
> Atom: 7.9s
> Loongson: 22.8s
>
> So, all in all, once corrected by the different applicable factors and
> all other things being equals, and taken the different feature sizes
> into account and so on, I'd say the results speak for themselves, don't
> they ?Well, actually they don't :) Yes, Loongson took almost 3x time, but even by your numbers, Intel Atom's "21W of horsepower" (according to Wikipedia, D510 consume 13W, so do you mean consumption of the complete system? vendor-rated, or measured personally by you at the outlet during that test?) is almost 2x of the Loongson's 12W. So in my opinion while Atom did show a better efficiency on this CPU benchmark, with these numbers it is not strikingly clear that one fared hands-down better than the other efficiency-wise. -- With respect, Roman
signature.asc
Description: PGP signature
