Hrm - there's quite a few bits in those benchmarks I don't like & would have done differently.
In the end, benchmarking is an art which is difficult to get right - esp. when you're trying to match up different micro-architechtures (i.e.: P6 vs P7 a.k.a "NetBurst"

as well as the usual manufacturer slug-out.
These benchmarks to be taken with a pinch of salt & some more care

.
Would love to speak italian so I could make some more sense about the article & have a few suggestive words with the author

.
Edit:
In case anyone wonders what I am talking about - let me point at a few things:
* SysMark 2000 - that's still running at 16-bit! Using anything other than SysMark 2001
will bring the P4 systems' scores down by quite a bit.
* Winstone 99 - Shouldn't have been used - should be retired.
Outdated benchmarks - that's one thing. I'm sure the software-stack for all the systems (and yes, I'm talking about both the Intel and the AMD systems) could have been better. The problem is in part that many such benchmarkers don't really know how to "properly" configure their machines.
Something the industry has to live with. I assume, of course, that the author merely did this due to ignorance or just could not afford/get ahold of the information/benchmarks. I tend to prefer this option to thinking of people as "forging" benchmarks to prove that X is better than Y.
