pea
developer
well, its definately a lot faster for me. The top-end speed is a few fps faster, but the original memcopy suffered from huge slowdown on non-alignment, whereas the new one is pretty much constant.
Using the old one (values at 40MHz, drawing a fullscreen sprite):
aligned 38fps, misaligned >11fps
Using the new one
aligned 42 fps, misaligned >38fps
Using the old one (values at 40MHz, drawing a fullscreen sprite):
aligned 38fps, misaligned >11fps
Using the new one
aligned 42 fps, misaligned >38fps