Is the overhead of calling a library function actually significantly higher than the serialization overhead that RDTSC already incurs? It's paying that price a

Is the overhead of calling a library function actually significantly higher than the serialization overhead that RDTSC already incurs? It's paying that price anyway, and if that shadows the cycles spent on the call and its overhead, it would be preferable to just call a library function...