All valid points, but the article is questioning why the benchmark produces different results when all that is changed is the name of the file. The game engine/game itself is completely unchanged.
Therefore, the speed increase is not thanks to clever optimisation for the engine or game, but due to changing settings in the options if it finds a certain filename.
An idea mentioned in the HardOCP article relating to the screenshots they did is that ATI are doing something like dropping the colour depth for textures. If this is true, then they are not improving the performance of the game through optimisations in the driver, but merely forcing it to drop quality for speed in a aprticular benchmark.
I'd be interested to see if someone could find out what it's up to (the driver), and if it is lowering a setting such as texture colour depth, then to match that/those setting/s in the rival cards to compare fairly.
'Of course I can speak, I'm Minister for Overseas Development'
Slashdot Mirror
Therefore, the speed increase is not thanks to clever optimisation for the engine or game, but due to changing settings in the options if it finds a certain filename.
An idea mentioned in the HardOCP article relating to the screenshots they did is that ATI are doing something like dropping the colour depth for textures. If this is true, then they are not improving the performance of the game through optimisations in the driver, but merely forcing it to drop quality for speed in a aprticular benchmark.
I'd be interested to see if someone could find out what it's up to (the driver), and if it is lowering a setting such as texture colour depth, then to match that/those setting/s in the rival cards to compare fairly.
'Of course I can speak, I'm Minister for Overseas Development'