I plan to have a 2d software buffer which will be 640x480. I want to generate an image to that buffer -every frame- and draw it with openGL. From what i remember textures are required to be 2^n (being no more then 512pixels?)

From what i can tell is after i generate the openGL texture i apply the texture data with glTexImage2D. The problem is there is no pitch, so i would have to make a 2nd software, blt the chunk i want to it and then use glTexImage2D on that. Since i am doing it every frame i would like to minimize overhead. Is there a way i can use a pitch? or maybe allocate vram data for textures and directly blt/dma/write to that from my software image buffer?

I hope i was not confusing. Generally my terminology sucks.

if you use the GL_TEXTURE_RECTANGLE_ARB target, your textures do not have to be power-of-two, though there are other restrictions.

http://oss.sgi.com/projects/ogl-sample/r...tangle.txt

if you texture from a pixel buffer object, you can get an asynchronous DMA to VRAM, too:

http://oss.sgi.com/projects/ogl-sample/r...object.txt