Thanks for this review.

I would like to point out that having at least one source engine game included in the test would be great.

Interesting review. Did you guys use their new version of avivo? There was one for cat 9-6. Found in the link below:
http://game.amd.com/us-en...?p=vista64/common-vista64

The others are for 32-bit XP and Vista. I wonder if there is a difference?? The files are larger on Vista then in XP.
http://game.amd.com/us-en...?p=vista32/common-vista32
http://game.amd.com/us-en...aspx?p=xp/radeonxavivo-xp

In my view, 9.6 is the best release for quite a while. The bug whereby SetDeviceGammaRamp and ICM profiles did not work under 64-bit XP has been fixed, although it wasn't mentioned in the release notes as such. Also, DXVA stream decode now mostly works properly, with a realistic amount of CPU usage. The remaining problem with it is slight image corruption in H.264 streams, although VC-1 plays back normally.

It's good that you guys review newly released drivers. How about using them for your GPU reviews?

Just a quick comment on video encoding benches: when hardware acceleration is enabled, how much difference does the processor make. For most of us, I would think that the i7 965 isn't a realistic comparison class because it's not in a realistic price segment? If the CPU makes little difference while producing hardware accelerated encoding, these benches say a lot about the ability to produce video encodes faster (if not at a high quality.)

Same as Nvida, huh - EPIC FAIL. They already both got their 10% optimizations out of their drivers way back when the 260/280 and 4850/70 hit. I saw them both pump near everything up like 10-15% - of course some quality is likely lost they just make sure the observer cannot readily tell - keep taking down anything that doesn't appear on screen...
Well, there's only so much blood in a turnip.
I suppose in the corporate world the wipes above and all around keep pushing for that 10-15% gain again and again and again. The coders can't produce that, but marketing keeps shooting off their mouths.
I feel sorry for the programmers. There just is no way they can meet the hype and marketing goals (lies).
Even the reviewer gave 'em a big break after they claimed 30% increase (max) and got at best 14% on just one game - and said they made their goal.
So fudge is the name of the game. Can't blame anyone really, their driver teams red and green are in a fierce battle and one half one percent in this or that game must be party time at both houses if you know what I mean.
I saw the Nvidia driver review and it was the same thing, a big zero compared to their market bragging hype. Maybe a gain here or there, but all the blood is drained.
They'ell have to start cutting down visual quality soon or maybe they already have - just to squeeze that extra fps - probably code it for firefight areas based upon the bloating of draw info and let it kick in then so it is not noticeable.
Thanks for the comparison and the nvidia article on their new drivers too.
The other chart I think in the other recent review shows the gains and losses going from vista sp1 to W7, and that was great as well.
This site has some VERY GOOD stuff on it, I'm surprised I don't see a lot more comments.
Thumbs up to the reviewers who have some very unique methods.