VBO's : One big buffer or several small ones??

[dopeflow]

Sorry if this has been asked before, but I was wondering what would be the better solution : to create one big vbo and then have my own memory manager for it or just save myself the trouble and generate a new buffer for each model??

It would be much easier to have one buffer per model, but with possibly 100 models would it still be effective? ( i say 100, because there will be a lot of minor models, like trees, rocks, etc..)

[PanzerSchreck]

AFAIK VBO have been designed with both cases in mind.This means that many small VBOs won't be much slower than one big one, if they have at last some hundred tris.
The only negative effect would be the memory-overhead when using many smaller VBOs.But in times of 512 MByte main memory and gfx-cards with 128-256 MByte VRAM that it won't be a big drawback.

[dopeflow]

Right, I was hoping this was the case http://www.opengl.org/discussion_boards/ubb/smile.gif

Maybe I will assign one buffer for all the environnement models because they will be pretty low poly, like around 50 each so...

Thanks for your help!