[MacDV] Re: G5 - is it really that fast?

[Michael Winter]

On Wednesday, June 25, 2003, at 09:09  PM, Video International wrote:

> Mike,
>
> I don't know if you had already read the article before your post. 
> Here it is if you hadn't:
>
> <http://www.haxial.com/spls-soapbox/apple-powermac-G5/>

Yes, I read it. He pretty much does what he accuses Apple of doing; 
showing the data but presenting misleading or inaccurate analysis (I'll 
assume its unintentional). He may have given it his best shot, but the 
bottom line is that benchmarking is hard to do and its impossible to do 
it in a way that their won't be people saying it should have been done 
a different way.

With regard to the inaccuracies the simplest explanation would be to 
read "Apple VP Defends Benchmarks" at 
<http://apple.slashdot.org/article.pl?sid=03/06/24/2154256>

Summary, the "cheats" (i.e. turning off hyperthreading) used to 
"cripple" the Dell actually improve performance on the Dell, and are 
things that Dell does too when running benchmarks. That and SSE2 was 
not disabled, the haxial article was simply wrong about that.
The "cheats" used to improve the performance of the Mac are options 
that either will or may be on by default in production Macs. The fact 
that Dell posts better SPEC numbers is a reflection of them using a 
different compiler, but Apple could do the same with a different 
compiler.

So, even if everyone is being completely honest and straightforward, 
when it comes to benchmarking there are still going to be disagreements 
and accusations of cheating. If you don't believe me, I challenge you 
to find a case where this isn't true.

-Mike