Originally posted by Ruined
So I assume you'd agree that every review with the R9800PRO vs FX5900Ultra should include a disclaimer that the ATI's scores only are done at FP24, possibly inflating the score over the FX5900Ultra which can do FP32, which is more work, and therefore ATI may have an inflated score?
When the differences between two cards are in hardware, and thus hard-wired into them, then that kind of information should already be in the technical specifications that are listed at the beginning of both reviews.
The problem comes when changes are made in drivers, particularly when these changes are forced regardless of what settings the user has requested. This is the big issue here as far as I'm concerned - Not the differences in image quality so much (although it's not a good precedent to set), but the fact that even if a user explicitly requests trilinear filtering, he doesn't get it in this game. There's no point giving users a control panel full of options if you are going to choose for them anyway.
As for the image quality differences not being discernable is concerned, that seems to be a very subjective filtering. While some people don't seem to notice at all, others have verified that 'once you've noticed it once, you'll spot it all the time'. It's unfair for me to give my own opinion as to how noticeable it is, seeing as I don't actually have a GeForceFX card to hand (although I am working on it
), but from Beyond3D's screenshots (even without coloured mip-maps) it seems pretty obvious.
I also would ask people not to confuse the AMDMB.com article with the discussion at [H] too much - To my mind they are looking at different issues. [H] are looking at bi/trilinear filtering issues, and AMDMB are looking at an aniso (or lack of) issue.