User:Realazthat/Projects wishlist/LLVM/Arbitrary integer reduction

From Rosetta Code

A pass that will reduce arbitrary integer math to a predefined set of allowed integers. So for example, it would reduce an i19 to an i32 or three i8's with some addition logic to achieve functional equivalence.

Should be a matter of using and linking an arbitrary precision integer library that allows fixed bit-width integers to be emulated.