MPIR is a highly optimised library for bignum arithmetic forked from the GMP bignum library. It is written in assembly language and C. It is currently maintained by William Hart.
The features of MPIR include:
- MSVC solution files for 32 and 64 bit Windows
- 64 bit inputs to all high level (_si and _ui) functions on Windows 64
- Superoptimised assembly sequences for x86_64 hardware
- Very fast Schoenhage-Strassen implementation for large integer products
- Improved divide-and-conquer division
- Many other algorithmic and coding tricks
MPIR is licensed LGPL v3+.
MPIR is written in ANSI C and assembly language and runs on many platforms (including Linux, Mac OS X and Windows on common hardware configurations), but is currently mostly optimised for x86 and x86-64 CPUs. It is designed to be threadsafe.