Well, if I remember right, there is a little goblin inside your PC which draws the image on your screen. This is a "incredible awesome" little guy, so it's called IA and if we invert this, its called AI.
Shitlord by trade and passion. Graphics programmer at Laminar Research. I write blog posts at feresignum.com
To be a bit serious I once made a floor that had 1024 pyramid meshes on it laid out in an even grid. When they were all in view it ran like crap on a decent computer. When only a few were in view it was fine. When I put them all in one mesh and had only 1 entity and not 1024 it ran just fine. As you see there is more to this then just poly-count.
If you include default.c, then you can press F11 and a Panel show you what make your fps down. It shows up the time that different things need (Functions, Entities, Panels, Particle, Script, Physic and so on)
most of the time there's one bottleneck, be it bandwith or vertex processing power. sometimes it's the scripts but then you've got either an incredibly time-consuming simulation or you're simply not good at it. there are hundreds of articles out there describing how to determine the bottleneck, F11 being one of them. there has also been some program (by nvidia?) out there which could measure performance of single draw calls; we once asked jcl to enable the engine for this tool (you have to compile the application with a specific driver), but i'm not sure if that has been done or not.
