PocketMatrix

One other thing you could try: There's a doc called 'fatmap.txt' floating around on the web (easy to find using google) that describes a technique using 8x8 pixel tiles to speed up texture mapping. The idea is that once you read the first pixel of a 8x4 tile, all other pixels are in the cache. It thus doesn't matter whether you're scanning from left to right, right to left or up-down. The inner loop of such a mapper is remarkably simple in x86 assembler. I'm not sure if the ideas port well to C(++) or arm assembly though.

Hey... I been trying to rotate my bit maps too..
er... Maybe this is a dumb one, but HOW to do it man?
I need some directions. How do you rotate a BIT MAP!????

Oh well, that's the easiest answer I could come with. It seems pretty straightforward to me... I started with this in PocketFrog to make sure everything worked properly. I am now in the process of optimizing it...