How large are arbitrarily large integers? (Was Re: Bitfield subtlety)

[paul zimmermann]

> I think one relevant question is on how large numbers we expect to be
> able to complete any computation with worse complexity than O(n) in
> reasonable time. But when we integrate fft multiplication with decent
> locality, the limit will be cpu power, not available RAM, I'd hope.

not sure. With GMP 17, we'll have multi-thread multiplication,
then memory will be the main limitation.

Paul Zimmermann