Wikipedia contains an article and code too. And another wikipedia article shows an algorithm (even for roots greater than 2), which can be easily implemented in binary format (such as the operation on bits).
If you want to stick to only real bitwise operators, you need to implement + using Ands, Ors, Xors, Nots ... If you want to do this on floats according to IEEE, you will need more work (the first wikipedia code can be used on the mantissa “directly”, probably under a certain restriction, and adjust the indicator “accordingly” ... you need to figure out how, however!)
Shintakezou
source share