I was working on writing a big integer math implementation in C for doing elliptic curve math in Web Assembly.
There is no arithmetic with carry in wasm, and I'm trying to decide whether it would be more efficient to use 30 bit limbs for more efficient carry when squaring or tripling numbers.
It looks like it'll require an extra limb for multiply/square operations for 256/384 bit curves, and for all operations on 521 bit curves.