Our rendering is a little lag behind the game. We lock a render target for CPU reading to force all previous rendering to be finished. This method is introduced by NVIDIA (http://developer.nvidia.com/object/General_FAQ.html#G4)?Is this method still good for ATI cards?