Would have been nice if the Inquirer had shown the data this investigation has evidently compiled.
As is we're left with no verifiable evidence this is true, no testing of any GF FX product that can definitively prove this.
They give precious little besides a few comments from someone they claim is an nVidia Technical Marketing Manager.... whom this person is we have no idea, and nVidia's clearly not changed their official stance.
Frankly, I'd like to see something to back up their claims.
Dave Baumann at Beyond3D has been espousing a
similar theory over the past 10days, and at least he's brought forth some degree of data to back up his claims, and sought others that can further verify the accuracy pof his theory.
Right now it certainly sounds possible that the GF FX has some sort of hybrid 8pixel pipe architecture that is capable of outputting 8 pixel/ clock is certain cases. The evidence for 2 fixed TMU/pixel pipeline still seems to be lacking however.
It seems like X-updates and Stencil Updates can be output with 8pixel/clock but rendering to the framebuffer is may only be 4pixel/clock.
More evidence is definitely needed before anyone can claim to make any definitive statement however.