Not hard to tell who the fan boys are, out 'n force as usual.
But here's the subject '.....then this week we find NVIDIA reclaiming their position as a mainstream leader'
Now actually read it, and put it in context. 'Mainstream' is the US$200 price card. The article is specificaly comparing FX5700ultra to 9600XT. The 5700 scores slightly higher in most tests allowing the reviewer to come to his conclusion. Nothing special here.
'...Using the Gunmetal "Benchmark" is Just plain WRONG. It is so clearly Nvidia dare i say it.. um.. biased...'
Yes, it is biased because it was optimised specifically for nV hardware. Just like other games are specifically optimised for Ati hardware. The reviewer just didn't have a copy of HL2 Demo to bench for an opposite perspective.
It is actually really easy to optmise for each card and make the other look bad. I can write a shader that uses the strengths or weaknesses of FX or 9x series. Hey, I can make the Ati run slow and even crash just by accessing one more textures than the limit in PS. I can make the FX run slow even easier by using full precision or writing short programs using different registers instead of longer ones reusing the same registers. I would normally write a shader that performs well on both, or write different ones specific for each by checking the device Id, unless I was specificaly give another directive by a sponsor, which has happened multiple times during my career.