Thank you, “Mom.” (I guess.) All of you are quite correct in guessing that the previous post was mine – that I did not bother to log in before responding. (But that I was making no particular attempt to hide myself ... why bother?)
Having both observed that a particular PRNG solution does not appear to provide 64 bits of entropy in an apparently 64-bit value, and having also expressed a concern with regards to this matter, it is quite reasonable to assume that the OP does require a more rigorous PRNG solution – perhaps, indeed, for some kind of cryptographic application. As I noted, “several such PRNGs are available in the Perl environment.”
I would also like to specifically call-out the proffered suggestion of “simply” combining two 32-bit values. From any cryptographic standpoint (or, from the standpoint of any other use-case requiring comparable rigor), this is n-o-t(!!) the same. The two halves of the resulting 64-bit value will in fact be joined to one another: one is the product of the PRNG at iteration (n), and the other is the product of that same PRNG at iteration (n+1), where the value of n is unpredictable but the relationship between the two parts is not. This is a potentially-deadly flaw. If you need n bits of entropy, you must use an algorithm that is designed to provide it. Such algorithms are available in Perl.