This is incorrect. There are a number of eerors, as others have pointed out, but for me the most central one is not that almost all reals are uncomputable numbers, but that the product of two 64-bit integers is 128 bits, as anyone who has done arbitrary-precision rational math has noticed.

I think it's great to experiment with improving fundamental algorithms, but not to make misleading claims about your results.