Original Post
I need some input on the different ways of vertexbuffer management: Some people prefer 1 vertexbuffer per "renderable object", some prefer several vertexbuffers per "renderable object" (multiple streams) and some ( nVidia and ATI ?) prefers several large ( 1-4Mb+ ) vertexbuffers. I was planning to use the ATI and nVidia recommended system - however that makes is harder to remove unused vertexbuffer data, harder to implement non-hardware instancing and probably some more. So.. give me input (pro's and con's) for each (and perhaps some new) management system