I'll try the PPC c2p later on, anyone know what kind of speedup it will give?
In all, using PPC for c2p is rather stupid, at least if the main application is 68k and the 68k c2p routine is any good. It only really pays off if you compare against slow 040 and poor c2p routine.
Compared to 040, some speedup, assuming the 040 c2p routine you compare against is decent.
Compared to 060, probably no speedup at all. Any decent 060 c2p is copyspeed (as fast as writing chipmem can get) anyway, thus PPC can't do any better.
If you could make the PPC c2p run asyncronous then you would get some real benefits (that is PPC would be running c2p for the frame while 68k would be calculating the next frame gfx). however, this would be really hairy to implement with all the cache coherency issues.