Well, Pando sugested UTLRP.SQL, not UTLIRP.SQL.

UTLRP.SQL only recompiles object which are invalid. So no overhead here.

UTLIRP.SQL in turn first calls UTLIP.SQL which invalidates *all* object in the database, then it calls UTLRP.SQL which will recompile *all* database objects, which has just be invalidated by previous script. So yes, there is an overhead with this combination, but it has its purpose: it is ment to use after database migration where all objects should be recompiled.