ATI get game optimization wrong way round ;)

[dangel]

http://www.elitebastards.com/#news12559

"Try this for size

If you rename the FEAR.exe file to something else and then run the built-in benchmark @ 1,600 x 1,200 x4 x8 (max details except no soft shadows) you get 40FPS for the X1800 XT (A64 FX-55, NF4 etc).

Run the benchmark with the exe file name unchanged and you get 29FPS.

Now, if it was the other way round the reason would be obvious (ie driver recognising fear.exe and switching on ops, etc), but this is certainly odd. Any thoughts anyone?"

[unreal]

Same thing happens with Battlefield 2. Rename bf2.exe or whatever it is to whatever you want, like omglolwtf.exe, run it, and depending on the set up it can increase quite well. I was reading up on it, and apparently it turns on DX8.1 instead of 9, or something like that

[smtkr]

Interesting bug. I think driver coders, if they are going to try 'clever' stunts like this, should check md5 sums instead of filenames. It would at least leave some consistency and keep people from running these ridiculous tests.