128 bit integer accelerated by x64-avx2 instruction set.
There are only four operations: addition, subtraction, conversion from floating point and conversion to floating point.
The operations are highly optimized by a lot of majikal operations.
The converted floating-point only supports ieee754-binary64 at present.
Header-only library.