benchies ... OK but quality comparison ??? LOL it is really useless to compare image which differ that much in location that you simply can't compare anything! I mean OK if you do not get the rotating healing portion at exactly the same place but when there is even weapon switching ??! this is really useless

Here's what I don't understand. When comparing 3.10's to 4.1's, in UT2003 there was barely a difference and yet all of a sudden a new set of tests are run and UT2003 ran far worse on the 4.10 set. Is this because you don't have a standard benchmark texhnique that you use every time? Or were these or the previous results incorrect? Oh, and as to your comment "Botmatch results are to much influenced by the processor speed", I agree to an extent but they are still a far greater indicator of actual performance than the flyby. This is especially important when one considers the fact that the Radeon cards are more CPU reliant than competing hardware.

Cool, thanks for the reply :) But it does kinda show that botmatch is probably a more 'realistic' reflection of ingame performance. Maybe in future comparisons is may be worthwhile posting both results?

"thus in that scenario, the fly-by is better" Sorry, but I don't agree as it could give a distorted picture. In pretty much every other benchmark, they 'tested' in close to real-gaming environments, whereas in 2K3 they did a flyby. That would be fine, except that it gives no clue about the CPU utilisation of the drivers, which is where the "real" horsepower is tested.... can it render terrain while also rendering bots, effects, etc at the same time and still allow the CPU enough cycles to look after the AI?

Go to bed. :P

hehe

Thank you :)