I wonder why 68k JIT is faster then WarpOS wrapper in this Voxelrace.
It isn't. It is almost exactly the same speed. You notice this better in high resolutions, where the drawing effort for the frame is much higher than the time spend in the OS to double-buffer etc.