SPE vs. GPU Vertex shader

[ShootMyMonkey]

Its inevitable when you actually place decals on geometry( you dont want bulletholes on thin air)

Collision for placing decals is done in world space, though. So it's all happening before any transformations other than world-space transforms take place, and so you don't need to process a vertex shader of any kind. Also, world geometry typically has a perfect fit to collision geometry. It's just that the collision geometry will not have any extraneous tris that are inserted for texturing reasons or something similar. Also, a lot of these collison/raycast techniques will handle things more implicitly (i.e., it will not transform a whole model when you're hitting a single tri within it).

Even otherwise, you would never implement all collision aspects using the render geometry.

And just because its "typically" now, after PCs evolved into 2 seperate entities (CPU=Gameplay, GPU=Graphics, both have seperate assets ) aint meaning this is the better approach, nor that Next-Gen Consoles wont be capable of different and better approaches.

True enough. I just often find that putting faith in the advancement of anything is invariably a losing proposition. Okay, I feel old having said that... oh well... call me jaded. I'm sure developers would *like* to have collision that's voxel-accurate if it were feasibly possible. To a certain extent, it is possible on PCs right now, but doing it along with gameplay, AI, user input, networking, all at 30/60 fps... not happening.

Also, there's no guarantee that physics will always get the first pick to ride shotgun should there be that much power to spare.