Should I combine all my geometries into one super geometry?

OpenGL Development JOGL Development
#1

I’ve been reading through the NVidia docs on it’s VBO interface, and have been getting the impression that it expects to find the verticies, normals, etc, for multiple objects in the same vertex buffer. (I’m thinking this because it has a section on emphasizing optimizations by using buffer offsets, and says to use glVertexPointer() as seldom as possible).

Does it make sense for me to cram several unrelated objects into one massive vertex buffer? Would interleaving my data be a good idea? Would this be a performance improvment? Would it be good or common coding practice?

#2

I don’t think there is a good general answer to this question.
If you are making a general purpose 3d library then the answer is no since it is likely that all geometry won’t fit in the video memory space. You need a solution offering more flexibility and you will have to perform GPU memory management (i.e. VBO & other native resources) .
If you are making a game for which you perfectly know your needs (in terms of GPU memory, scene coherency,…) and you know your minimal spec, then it may be the optimal solution.

Vincent