i have a quad matching the size of my screen (640x480 fullscreen) with texture mapping,
and it is extremely slow on non-accel’d machines. is there any way to make this a
little faster on software implementions? i could make the same thing in dos run n^x times faster!
Don’t expect wonders from the SW implementation.
Don’t know about the filter modes, but no mipmaps and GL_NEAREST for minification and magnification filter gives the best performance in SW.
Though I still doubt it’s really fast then.