Well obviously they're different processes, but they have the very same adaptive settings, so (I suppose aside from differences in what the output of the cache looks like vs the render pass affecting the 'contrast') the same settings will add more sampling in the very same places.
Mind you, in the one tutorial I've seen about using the render cache, they kept it at a constant setting and used the retrace or auto-occlusion options to get more detail where needed instead.