> Because it's so so slow, even naively converting multiplication in a series > of modular additions is much faster:
dunno about this one - after fixing the insanely slow recursion, changing the implementation to a simple natural mul+divrem 100x:ed the performance - this is the textbook implementation and doesn't even try to be efficient (ie no special cases for different modulos, small numbers, montgomerys and whatnot. I'm sure this is trivial to both benchmark and implement better but the iteration approach sure doesn't look like a winner from this sample of one :) <https://github.com/status-im/nim-stint/pull/156>
