Patrice Mandin wrote:Not sure about the game tick problem, would have to test it myself to understand the issue. Lowering/changing game tick rate could cause problem with demos and/or network play.
Yes, demo recordings or network play with engines having different tick value (and related other values derived from game tick) are not compatible, but changing that was necessary. Being constantly behind the (35Hz) game tick increases Doom engine CPU overhead significantly.
Most of Douglas' changes to Doom code are behind defines so that one can do builds with original values/code and modified versions. This way optimizations can be easily switched on & off and tested against original version, both for functionality and performance.
Patrice Mandin wrote:I also did not look what Douglas did for sound propagation / monster wakeup.
Doom code for sound -> monster wakeup is recursive. There are some places in Doom maps where this can cause very noticeable delay / skip in frame updates as monsters further and further away are woken up and making noise. Douglas both optimized this code and put limit on recursion. I think this and game tick are only(?) changes to the game play, other optimizations don't have effect on it.
As result, when one enables all optimizations and speedup changes, BadMood is a bit easier than original PC Doom. Lower (4-12) FPS on Falcon compensates this though...
With games using same engine, demo recordings and network play are compatible, so that's not too much of a problem. One would anyway be badly disadvantaged when playing PC Doom against person on a PC with much higher FPS.