In my OpenGL program have a huge vertex buffer with data (normals,position,texcoords) for 2048x2048 points. In each frame i reduce my indexed buffer with a LOD algorithm and bind GL_ELEMENT_ARRAY_BUFFER again.
I wonder if it makes sense to additionally reduce the vertex buffer as well, so that it only contains the used vertices from the index buffer. So the question is, if there is a performance gain even with rebuilding and rebinding the vertex array per frame.