Subsurf modifier vs mesh object.

My question is regarding the choice of either keeping meshes with the subsurf modifier or simply converting the model to a higher poly. Has anyone ever set down and tested on which would be more efficent for rendering, skinning, uv-mapping, etc. I have done a little bit of testing with small file sizes and haven’t really noticed a difference. I would like to know would there be a difference with larger file sizes and renders.

I’m sure the file size is larger once you convert the meshes from it’s modifier to a real mesh, but would it take more memory for the computer to “compute” how a model would look like virutally with the modifier than it would to read to file and draw what it is written.

It really makes no difference I think, the modifier is basically the same as using “subdivide smooth” with proportional editing on, except that you can undo it all at any time you want whilst keeping the position.

Basically what I am asking is if I have a model that is smoothed and refine with the subsuf modifier, could I just leave the modifier on or would I get better results through rendering (memory and time wise) if I just go ahead and convert the modified object to mesh with increase of polys.