More efficient use of vertex buffers?

[lmcd]

I'm just looking through t3d_model_draw_object. I have a model that's been automatically split into multiple objects due to different materials. As an example, the objects could look like this:

• Object 1: 35 vertices
• Object 2: 10 vertices
• Object 3: 15 vertices

The objects combined still consume less than 70 vertices here, so it appears to me like these could be loaded all at once in t3d_vert_load, rather than separate calls increasing memory traffic.

Is there any technical limitation preventing this from happening?

[HailToDodongo]

It depends on the material, if it changes anything that the T&L part needs it needs to be split.
Other than that yes, it should at least load the vertices at once if possible.
There is soon going to be some refactoring needed for other material reasons, then that part will get handled too.