I would be exceedingly surprised if your simple TRNG actually generates statistically robust random number sequences.

Not to mention that it's going to be fairly expensive in terms of time to G them, what with each N, however R, costing a million-count loop!