Drivers, benchmarks and new products

This is becoming weird. A few years ago, if a graphics card manufacturer brought a version of a graphic driver specially prepared for a specific software, everybody cried wolf and it was a scandal, a benchmark cheat and unwelcome optimisation to favor results on a specific measurement. Today, all this evolved significantly. Did you notice haw many version of the graphics drivers are officially targeted to some precise video game (and published just after its launch)?

In the spotlight:

This seems to become a rush for getting the most information from the publisher. nVidia or AMD? An upstream partnership then allows to have a driver optimized for this game or that one (hopefully for this demo of a video game, even before the game itself is ready).

Comparison becomes more and more difficult. Think about it: How can you be sure to buy the right graphics card for your set of PC video games? You must be sure that there was at least two months between the game(s) launch and the tests published in the press (and check that the reviewer actually used the latest version of the driver package). This was epitomized by the recent delivery of an AMD driver, several months after the launch of STALKER Shadow of Chernobyl (AMD does it again with a new driver for Quake Wars : Ennemy Territory), to push the performance up by 10% to 20% on this game. This is no a little detail if you notice that it is common to see a $100 price difference between two graphic cards that have this difference in performance.

Is it the sign of a matured market? I doubt it. I believe that it is more a rush for performance where the driver developers are loosing ground against the hardware (graphic chipset) developers and the game developers. Time has come for chipset manufacturers and game developers to cooperate more (is it possible?) or to ease the work of the driver designers (but the Windows Vista drivers do not bode well).