also wonder if you know why D3DCREATE_SOFTWARE_VERTEXPROCESSING would be better framerate then D3DCREATE_HARDWARE_VERTEXPROCESSING

with new engine i get now 2300-2600fps on software_vertex proc but with hardware proc i get 350-400fps
Hmm.. that's odd. Hardware should be faster than software. Maybe you should update your drivers :? I guess that's a hardware problem or driver problem. Testing your app on different hardware machines might give you a good idea of what's going on. Can you upload it again?

Can you give me the code you use to set up your quad? Then i can modify it for you to render a part of a texture. (So you can render both sub-images seperately) It isn't that hard. Didn't you understand my explanation about the texture coordinates? :?