They seem to be confused on what constitutes a 256bit memory bus, they seem to have followed nVidia's PR mischief in twisting around numbers to imply the GF2 GTS and all cards since have a 256bit memory bus, which is quite incorrect.
Their specs for minimum and maximum theoretical DRAM buffer is incorrect straight across the board.
A, but about their NV30 claims...
The listed maximum resolution seems extremely dubious to say the least.
A refresh rate of 300Hz at 2048x1536x32bpp?
I'm sorry., but if it uses a 400MHz RAMDAC as they claim then that is not even theoretically possible, hell it just about quarupled the theoretical bandwidth capabilites of a 400MHz RAMDAC. Besides that, what in the hell kind of filters are these cards going to use to allow that kinda refresh rate even if the RAMDAC was capable of it?!
The interference from filters that powerful would be ridiculous.
Hmm, they claim it supports 64bit color, but only 32bit Z-Buffering? Seems like an extremely unusual decision to limit the z-buffering to 32bits if they allow a 64bit color depth.
The funniest thing I've seen this week though....
Look at what they list for the number of hardware accelerated lighting sources? 256!
All I have to say is BULLSHIT!
I don't even want to think about the complexity that would require. Christ, the WildCat II 5110 only accelerates 24, the FireGL4 which is an absolute wizard at calculating lighting support 16. And they are telling us a card intended for the CONSUMER market will accelerate 256??!!!
Not to mention I'd love to know what in the hell kinda scenes would have 256 simultaneous light sources, and I doubt they'd increase chip complexity unnecessarily.
I think it's 100% obvious beyond any possible doubt that they just picked numbers at random and paid no attention to what is possible, or what makes any sort of logical sense.
I'll give you one thing Gog... that page was just hilarious the claims they make
