Hmm. Does anyone know how Xavier NX compares to regular CPU or GPU tasks in ML?<p>I’m curious to know if it might be useful or interesting to have one of these as an adjunct to a SFF computer that I might have in my server closet, e.g., an Intel NUC or a Mac Mini.<p>Or is this a device that is meant to be pushed further to the edge, like attached to each of the cameras I have?