Too late.

I don't buy nvidia anymore. I've lost all trust in them. I also don't like the fact that they're boycotting TR by not sending TR review units anymore. So I reciprcocate.

Much like AMD's PR rating is a joke, this means nothing w/o ATI's bilateral support. One has to ponder what is the measure of a graphics card if we can't come up with a baseline method where every card is measured on equal footing.

Three laws huh. Asimov should sue.

Sorry, but it sounds to me like a bull shit, an attemp to quiet angry users.
And the last point is really funny. Didn't they said it was a bug?
P.S:Sorry for may english.

What Nvidia could not do with hardware they accomplished with software.
Like 3dfx before them, the game is over and the fat lady is singing.
Nvidia has lost........bye, bye Nvidia.

My take on this whole thing is:

I think it's likely that some developers inside nVidia got a little greedy and under-handed. They started doing some really questionable things to make their product look better. Probably not with authority from the very top, but possibly with an affirmative nod from their immediate supervisor(s). Suddenly nVidia suffers a huge public black eye, and the guys running the whole show get very interested in what these developers have been doing. People get put on probation, some get fired. It's totally supposition, but it's a likely scenario.

Now, nVidia having worked through the ordeal, and having suitably reprimanded the responsible parties internally, comes up with a standard that will both placate their customers and give clear direction to the development teams which limits CAN and CANNOT be crossed. Management says to the guys inside, "don't make us look bad like that again or you're gone." Management says to the public, "look, here's what we're going to do to make sure this doesn't happen in the future."

Yeah they should have come a little cleaner, admitted they screwed up and apologized for it. But this is a good start, at least. And I think application-specific tuning is probably here to stay, anyway. Even ATI has dabbled in that (maybe more than we know).

I think this is a positive sign that nVidia realizes it screwed up, or at least people perceived it as having screwed up (whether they agree or not), and needs to do things a little differently. Hopefully they will honor their new rules, be a little more transparent (not likely), and raise their credibility. Only time will tell.

What makes a man turn neutral?

Where's BitBoys Oy to save us??!?!?!

I really don't give a crap what Nvidia says or does anymore. I supported their products (and drivers) for a solid two years until they started to go downhill in the fall of 2002. I got myself a Radeon 9700 Pro and haven't looked back. I also cannot support a company like Nvidia who has blackmailed consumers in general by striking an alliance with EA under a program called "The Way It's Meant To Be Played". Talk about trying to divide and destroy the PC gaming industry. I think any product should stand on its own merits and not receive unfair bias.

I have an Oak Technolgies SVGA ISA card with 512k of RAM. It's a pretty trick card with its yellow PCB.

I think it's better than anything Nvidia is putting out lately.

I hope I can run Doom 3 on it. It ran the original Doom just fine.

I see two problems with the guidelines. First, the whole process seems ridiculously labor-intensive. Maybe they'll try to follow it at the start, but then it will become too much of a hassle and start cutting corners. Second, since they allow the drivers to identify certain applications and apply various tweaks to the rendering and they don't address the issue of pixil shader precision, it seems like we just have to take NVIDIA's word that their future optimizations are legitimate. (Testing this stuff is going to be insanely time consuming.) I, for one, trust NVIDIA about as far as I can throw their building.

An optimization must accelerate more than just a benchmark case.

That does not necessarily mean a benchmark suite or application, Dis. It could even apply to Quake3 benchmarks or UTK3, or whatever. Of course, this would be an incredible change for nVidia, and I don't think it very likely, but one could and should certainly interpret that bullet in that way.

The good thing about this is that they will now be under scrutiny about following these guidelines. Even if their cards produce a different image that is a result of "wiggle room" instead of driver optimizations, they may have to answer the question about why it is different. This may just help to point out what other tricks they're using and maybe level the playing field somewhat.

liars

Isn't the "correct image," and therefore "proper" trilinear filtering defined as the one that is produced by the MS Direct3D reference drivers?

Assuming the driver team actually follows these guidelines, perhaps they can do what they should do - fix bugs (whilst not making new ones) and give us performance improvments that are actually worthwhile.

This is good news. Maybe this will force the driver development team to make something fast but provide good image quality. I wonder how bad this will affect current scores in UT2K3 and 3DMark2K3

like scott said in the mini-editorial... unless that "wiggle room" is tightened down, most companies are going to wiggle in whatever way will best suit them.

it's good that Nvidia is recognizing publicly what an optimization is, however that does not mean they will follow those guidelines (or that they are necessarily correct). as you've said, they've given themselves plenty of "wiggle room" here.

and we've all seen what Nvidia can do with a little wiggle.