If Valve ships Half Life 2 and the performance trend lines today match up with the performance trend lines a year ago, then ATI can at least say that they didn’t mis-represent the performance advantage the 9800 / 9600 series held over the GF FX series. The release of the X800 and 6800 product lines may have made such comparisons spurious, but it isn’t ATI’s fault that Valve lied about their own release date.
If, however, the “real” Half Life 2 engine can’t duplicate the results demonstrated at Shader Days almost a year ago, ATI and Valve both come away looking like first-class liars. This kind of scenario could (and probably would) leave a lot of consumers who bought ATI cards FOR the express purpose of Half Life 2 quite unhappy, especially if they bought the many cards that do NOT have a free coupon for the game.
Could this happen? Yes. Remember at the beginning of the article where I said that driver revisions don’t tend to increase performance 30-40% unless there’s a serious problem? That’s still true—but in Half Life’s case, its now clear that the engine wasn’t finished either, and that can make a tremendous difference in the final game. (Ironically, the thieves who stole the game reported the engine wasn't finished, but were shot down as being "thieves and liars").
Worst-case scenario for ATI would be the GF FX series being able to come within much-closer range of equivalently-positioned Radeon 9600 / 9800 hardware while the 6800 assumed the lead by a Doom-3-esque margin.
Even if this doesn’t happen, even if ATI pounds NVIDIA into the ground in Half Life 2, its still not a good situation for consumers as a whole.
Its All About the Money:
What changed my mind about the advisability of game developers working hand-in-glove with hardware designers was the supposedly-demonstrated bifurcation between ATI + Half Life 2 and NVIDIA + Doom 3. Yes, this lends itself to a simple solution “Buy ATI for HL2, buy NVIDIA for D3”, but what if you’re a consumer that wants to play both at high quality? Buying two video cards is not an option for the vast majority of people even when the cards are designed to work with each other SLI-style via PCI-Express.
I believe John Carmack when he says he didn’t optimize unfairly for one architecture over the other, but John Carmack is the exception in the gaming industry, not the rule. 99.9% of game developers (and game development firms) are not rich, well-established, and powerful, they do not have long, proven track-records, they are not regarded as near-Messiah’s on 3D gaming design, and they do not have huge personal cash reserves.
The line between “We built the best game possible that happens to run better on NVIDIA (ATI) hardware” and “We built the best game possible to run better on NVIDIA hardware” is razor-thin, especially for a hypothetical company struggling to survive long enough to push their product out the door. While obviously there must be communication between hardware designers, software developers, and standard developers, such communication should not rise to the level of hand-in-hand marketing and development.
We’ve already seen what can happen when a manufacturer is faced with sub-standard performance in a flagship product line (NV3X); imagine that effect across the entire industry as a rule-of-business. Do we want to see ATI constantly optimizing for NVIDIA games or NVIDIA digging around for driver tweaks in ATI games, each with the sole goal of gaining a few FPS? We complain about that situation now. Imagine how much worse it could be if picking up "the next big game" became a direct race. Corporate “sponsorship” could end up stifling the industry if partnership with one company or the other is seen as essential to building a successful product.
Did NVIDIA gain from its Doom 3 partnership? Yes. Did it gain more from Doom 3 than it lost on Half Life 2? Certainly not yet—and final verdict will doubtlessly depend on how good or bad Half Life 2 looks.
Did ATI gain from its Valve partnership? No. Will it, in the long run? Only if final performance figures vindicate the purchasing decisions of people who bought HL2-enabled cards as long as 18 months ago. If they don’t, ATI and Valve both end up looking like liars, if they do, we end up with a bifurcated system where you buy NVIDIA for Doom, ATI for Half Life, and suffer if you wanted both.
In this case its not even Doom 3 or Half Life 2 themselves that's the problem so much as what'll happen if the game industry picks up on the benefit of this type of partnership. Too much focus on one manufacturer (or simply walking too close to that line) can lead to problems for both. Given the delays, the coupon embarassments, and the further delays, I'm betting some at ATI wish they hadn't touched HL2 with a ten foot pole. But then again, with NVIDIA backing Doom 3, ATi needed SOMETHING. Wanna take bets on who gets Quake 4 or Unreal Tournament 2005? Hopefully no one. This type of fragmentation doesn’t help consumers, the designers, or the industry. In the long run, it might not even help NVIDIA or ATI.