numpy.genfromtxt seems to ignore dtype
It sounds like you're asking whether it's possible to have a standard 2d array while also having named columns. It isn't. (At least not in the sense you seem to be asking.)
An "array with names" is a structured array -- it's an array of records (not really tuples), each of which has named fields. Think of it this way: the names aren't attached to the array, they're attached to the "tuples" -- the records. The fact that the data is of a homogenous type doesn't matter.