Since good bignum libraries already use tricks like this, and many others besides (like using FFT to handle very large bignum multiplication), it doesn't seem t

Since good bignum libraries already use tricks like this, and many others besides (like using FFT to handle very large bignum multiplication), it doesn't seem that the hardware needs to change.