Lars, the question is whether they are purposeful optimizations (read cheats) or not. Issues and poor performance (which is what HL2 w/ FX was/is) are didfferent than the issues from this summer, when anyone but nV's lawyers knew that something smelt fishy. How can anyone show that these ATI issues are either genuine purposeful FLOPTIMIZATIONS, or just errors. With your opening sentence looking like the one I quote, you in effect equate the two as being equal cheats of equal magnitude:
<i>Earlier this year, NVIDIA drew a lot of flak over a number of questionable optimizations in their drivers for 3DMark 2003 - optimizations which could definitely be called cheats.</i>
Calling these cheats before getting ATI's reaction is still simply wrong. There are anomalies obviously, but that's not new for EITHER company. The questions was were they intentional in the same way that the nV ones are?
If they are cheats then we and many will crucify them like everyone before (not just nV but SIS too), but that leading senctence pretty much judges them before all the info is in.
Now if it's a limitations of the drivers, and it isn't fixed or addressed saying it can't be fixed, after a few releases, then it would be logical to say that ATI was 'up to something' or 'lacking in something'. If it's the way things look under 24vs32 precision (mainly for the overdraw I would think) then that would be another issue that isn't there as a cheat but a natural byproduct of design, as negative as it may be.
As for gamma, did anyone try adjusting it before making a big deal, or do only we gamers get to do that on a regular basis from game to game?
The thing that bothers me the most, is a level of difference between what is perceived as an ATI 'cheat' and an nVidia one.
The most glaring and STILL CURRENT example is nV turning off trilinear AF at ALL times. Now while UT2K3 does do a similar effect on the ATI's it can be initialized by the .ini file, however nothing will switch it on in the nV line, yet that isn't mentioned in the UT2K3 segment NOR in the segment about ATI's issues. WHY?
Can we expect equal time/space for ATI's response as <A HREF="http://www.tomshardware.com/graphic/20030912/half-life-01.html" target="_new">nV's response on HL2</A>?
MY issue isn't so much with your review I liked it and the issues made me go 'hmmm', however labelling them as being ANYWHERE near the same thing as the issues in A3 (why weren't the ATI issues found during that review?), or the Issue of this past summer isn't fair.
I also wonder what use it is to include Randy Pitchford's comments when he won't back them up? Proof would be nice. At least Gabe suplied us with that, and even let YOU do tests. The quality/veracity of these two are NOT equal.
It's also interesting that there were no issues in HALO, yet there are issues elsewhere.
Taken from <A HREF="http://www.digit-life.com/articles2/gffx/nv38-36.html/nv38-36.html" target="_new">THIS Digit-Life review of the FX5700/5950</A>, the FX's aren't rendering Lights in the ceilings and on the walls. Check these two Stills for comparison <A HREF="http://www.ixbt.com/video2/images/asus-ati/r9800-38-halo3.jpg" target="_new">ATI no AA/AF</A> and <A HREF="http://www.ixbt.com/video2/images/asus-ati/fx5900-5214-halo3.jpg" target="_new">nV no AA/AF</A>. It's small and easily missed I guess, but it is there none the less.
Now are the nV issues worse or better simply because no one pointed people in the right direction to notice them unlike the ATI 'issues'? Everyone got a nice subtle nudge from nV right before the release of their latest product. The issues that the FX line has with PS2.0 are still there, and are being glossed over by the nice 'new' run-time compiler that is needed to make standard HLSL run well on the FXs. And as evidenced by many recent reviews of MAX PAYNE, this means that nV's products will come out with performance behind ATI's until they can 'optimize' their drivers to re-compile on the fly for each game. Pretty big issue IMO, that wasn't given much mention; it's an even bigger issue in light of nV's statements about moving to fewer driver releases (wait almost a year for your card to perform properly?) that doesn't sound good at all.
I guess everyone is waiting with baited breath for ATI's statements in reply to all this, at which time we will have a better picture of what's happeneing (hopefully).
I'm also waiting for a revamped VGA buyer's guide to correct the glaring differences between THEN and NOW.
And could you PLEASE add a Parhelia to the tests, then we can see GOOD IQ to compare to! Even if it's framerates are comparable to the R9200/FX5200non-ultra.
Once again it's announce, hoopla, then wait.
Oh yes, and I almost forgot: <b>The Catalyst 3.8 drivers are causing World Hunger!</b>
More on this later, just take my word for it, for now.
- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <A HREF="http://www.redgreen.com" target="_new"><font color=green>RED</font color=green> <font color=red>GREEN</font color=red></A> GA to SK