Additional Benchmark comparison

[apmccartney]

The mp++ library is considered by many to be the state of the art for performant multiprecision integer arithmetic and underlies the symengine library behind the popular sympy python symbolic mathematics library.

A performance comparison to ctbugnum would be very interesting

[niekbouman]

Dear Austin, @apmccartney

Thank you for your comment. I am aware of the existence of mp++, but haven't used it myself. With ctbignum, I am focusing on compile-time computations and run-time computations with integers whose limb size are known at compile time, and modular arithmetic with a modulus known at compile-time. I specifically target on operands with a few hundred bits (say, two to four 64-bit limbs), whereas mp++ seems to focus on implementing a "small-vector optimization" for big integers of arbitrary size.

What is your use case if I may ask? General yet fair performance comparisons are always a bit hard to make, a benchmark targeted to a specific use case could be more insightful.