ap_math

simple compile-time abitrary-sized artihmetic types.

Ap_int<>

Behaves like native types as much as humanly possible.

• Arithmetic
• std::numeric_limits<> specialization.
• std::ostream support.

The only real difference with native types is that numbers greater than the largest uint64_t cannot be initialized using integre literals, so you’ll have to use the string-based constructor (which is constexpr too!).

Example:

#include "vecpp/ap_math/ap_int.h"
#include "vecpp/ap_math/limits.h"
#include <iostream>
int main() {
  vecpp::Ap_int<80> a{"1234567890123456789012345"};
  vecpp::Ap_int<80> b = 12;
  a = a * b;
  std::cout << a << "\n";
  // ap_math also support very small ints for completeness
  vecpp::Ap_int<3> a = 1;
  vecpp::Ap_int<3> b = 2;
  for(int i = 0 ; i < 20; ++i) {
    a = a + b;
    std::cout << a << "\n";
  }
  // std::numeric_limits<>
  std::cout << "the largest possible 265 bits int: " << std::numeric_limits<vecpp::Ap_int<265>>::max() << "\n"; 
  return 0;
}

Why?

In order to hit precision objectives in cste_math,
we need to be able to compute values of sizes that are not available with the core types.

Acknowledgements

The implementation details are heavily inspired from llvm::APInt and llvm::APFloat.