PocketMatrix

You could try using the Intel Graphics Performance Primitives.

See .

Dan East

I'd say pocketGL wouldn't really help you - the textured triagle draw implementation is in C, and is pretty much the same as any other efficient texture mapper out there. OK it's a reasonable amount of code, but it's not that complex - a good GX book or website should be able to help you.
Additionally, PocketGL has an unusal but clever optimisation for its z buffer which would probably get in your way.

Intel GPP looks like it uses scanline functions, which is not that i'm looking for. I can make use of the scanline functions, but don't think it will give me that much speed increase.

I know exactly how to do texture mapping, and have a C routine working. But want to know if ARM assembler is going to boost the speed or not.

How much triangles ( Plainly texture mapped, no perspective correction, no Zbuffer ) should I try for in a second ?
Lets say the size of a triangle is about 70 x 70
pixels on the screen.

1000 ? 5000 ? 10000 ?