You can convert polygonal geometry to a subdivision surface by using an AttributeSet node to set the "type" to subdmesh.  You may want to do this if your geometry looks faceted as you move the camera towards the geometry.  If you have a low-poly model, you will see differences in the rendered surface.  If your surface already has a large number of polygons, the difference will be less noticeable.

Converting your model to a subdivision surface has a some advantages, though it will cost more memory and time to render.  With a subdivision surface, displacements can be made in a more robust manner, and you will be guaranteed a smooth render.  See the RenderMan documentation for Subdivision Surfaces for more information.