Optimization

[omegadox]

If you want optimization, then implement stream caching, ZTP slows down because of the geometry payload that occurs basically on every frame. Dolphin spends most of its time loading all the vertex loading, gamecube can use over 8 vertex loaders and they all do their work every second basically. This technique is fast on hardware because the hardware spends time just doing that and nothing else, else for software, this isn't a good approach to mimic something that only good in performance on dedicated hardware, so we need to make a vertex array hash table and cache them, they way it saves time on dolphin from loading too much every second and same can go for Display Lists, they could be cached and then point to them stored in ram, this is how Gamecube does it, I am not sure if Dolphin does this already.