Well, it is a matter of scale, I know ...
Looking at the node networks in the posts above ... they do the same ... and they work, which is the most important.
One thing to notice is, that the simple Perlin noise produces too noticable patterns ... or the nodes get very complicated. In fact I like patterns ... like inside clouds ... different and the same at once ...
Another thing is, you cannot scale the Get position .... which inconvenient for most cases, where you want to use one function for more than one occasion. And you want to scale it for the pattern purpose mentioned above ,-)
A PF handles both: Patterns in form of fractal details and scalability with a transform shader.
Another point is: why setting up, two copies of Parts-shader, when one is enough!?!
P.S.: What I am awaiting now is to be proofed wrong ... as many time before ,-)