Posted by Dave on November 09, 19103 at 19:31:51:
Just wondering what you think of this.
While 'discussing' the game engine in ITC208 with Jim, Dan was pushing the idea that every cycle we should be performing the render tasks first, then taking input and doing other processing and updating the game state. In this way, you get some measure of parallelism as the render data is sent to the graphics card and is already being computed while other game processing is going on. Therefore, it's faster.
I thought this might be of interest for us, since I know the minigames labour a little for framerates on the kind of hardware our client has (a least I know it's a bit choppy on my P200MHz, but that was a debug version running in a window). I don't know to what extent the old hardware would work in this way, but it sounds like a good optimisation. Have you got something like this implemented already?