I have a static vertex buffer of say 1000 vertices. I am pumping this vertex buffer through the pipeline every frame. There is also an index buffer which is perfectly implemented and there is NO multiple transformations of vertices. If I updated my index buffer every frame to make it cull out the vertices that were behind me (not in the view frustum), would i get a performance increase? Before anyone says yes, remember that you are still pumping the same number of vertices through the pipeline. Also directX does culling of triangles automatically to a small extent. Still, would there be a performance increase?