Well, look at what Sweeny told the Rev as recounted here:
http://www.beyond3d.com/forum/viewtopic.php?t=7006
Tim Sweeny said:
Anthony,
We're not in the business of publically critiquing the nuances of drivers or approving/disapproving them. We only get involved in the driver situation publically when there is a major driver bug that has a significant negative impact on people who play our games, and that hasn't happened in years. Things like mipmap bias factors and trilinear optimizations/shortcuts don't fall in this category; they are subjective and it's best for users and web critics to debate what performance/quality tradeoffs are acceptable.
We often have informal conversations with NVidia and ATI about performance/quality tradeoffs and shortcuts, but these are private discussions and not public endorsements or rejection of the tradeoffs they make.
Long term I would like to see the rendering API defined in terms of exactly reproducable results, so that there isn't room for debate on things like texture filtering optimizations or mipmap biasing. Then, a given set of triangles and render state passed down to 3D hardware is guaranteed to give the exact same results on all hardware and any software emulation or fallbacks. There are general and non-controversial definitions of how all such things should work, from IEEE 752 for floating point to Fundamentals of Computer Graphics for the definition of trilinear filtering and mipmap factor calculation.
-Tim
In the first paragraph here Sweeny makes it clear that Epic never discusses IQ issues with IHVs apart from major, detrimental driver bugs. That's pretty clear.
In the second paragraph he relates other "informal" discussions with IHVs which are "private" but do not equate to
endorsements or rejections of IQ "tradeoffs" discussed.
I find these paragraphs completely congruent. I would characterize them thusly relating to this situation with this hypothetical example:
nVidia: Tim, we are interested in in incorporating a performance trilinear mode in our drivers relative to UT2K3 which will be dependent on application recognition, as it isn't something we wish to do with other games. We want to tailor it to UT2K3. Any suggestions from your point of view on how we might do this? We're doing it to get better performance than we can get currently using standard full trilinear, so we thought we'd see if you have any suggestions..."
Sweeny: OK, I've got a couple of approaches I can think of you might wish to try. First....blah, blah, blah...
Basically, I think he's saying here that he doesn't care what an IHV does in this regard, as Epic considers this an issue between the IHV and the end user, not between Epic and the end user.
I would say that I agree with this position generally, except for one thing. As nVidia has a financial relationship with Epic/Atari which involves payment to Epic/Atari for nVidia logo display in the game itself, the appearance of a conflict of interest is at least suggested (notice I am not saying one exists.) However, this is something Epic should be very careful about. Frankly, as I've said since last year when UT2K3 shipped, although Epic intelligently made it easy to replace the nVidia logo in the game with one from the user's choice of IHV, I think such a policy is in very bad taste and the product of poor judgement on someone's part.
At best the TWIMTBP logo at the start of the game implies some sort of "special attention" Epic is including in its code for nVidia products, something which Tim flatly denies in the quote above and has denied in other quotes on other occasions. At worst, it makes it appear as if there is a conflict of interest in the development of the software (even if of course there is not.)
I think Epic/Atari in the future would be well-advised to sell whatever space it wants on its program CD to whatever IHV is willing to purchase it, and exclude all such promotional material from the content of the game itself. I would certainly hope Epic will do this.
