The packge circl/ecc/p384 contains optimized code for AMD64 and ARM64, but lacks of a generic implementation using math/bits package.

Some of the header files of LEDA (#186) have functions in them for the purposes of inlining. It doesn't look very clean to me, but on the other hand that means losing the inlining (if I understand things correctly).

Do we want to take a position on this?