libxint - a C++ library for extended size unsigned integers

This is a headers-only C++20 library to emulate large fixed-size integers. It's licensed under the GPLv3+ license.

Installation

When building from a tarball, you can skip step 0.

0. ./bootstrap

1. ./configure

2. make

3. sudo make install

To uninstall:

sudo make uninstall

This is a standard Automake package; run ./configure --help or check INSTALL for more details.

Usage

First, include the header libxint/uint.hpp in your program. The type xint::uint<B> is now available; the template argument B is the number of desired bits. It can be used as any ordinary unsigned integer:

#include <iostream>
#include <libxint/uint.hpp>

using namespace std;
using namespace xint::literals;

int main()
{
    xint::uint<1024> x = 42;
    cout << x << endl;
    x <<= 1000;
    cout << x << endl;
    ++x;
    cout << x << endl;
    cout << popcount(x) << endl;
    auto y = 0xdeadbeef0000babecafe0000dec0de_uint;
    cout << hex << y << dec << " has " << y.num_bits << " bits." << endl;
}

Implementation details

This extended integer is implemented through multiple native integers, called limbs. The size of each limb can be controlled by defining the XINT_LIMB_SIZE macro. Valid sizes are 8, 16 and 32 (default is 32.) For instance:

g++ -std=c++20 -DXINT_LIMB_SIZE=16 my-program.cpp

The limbs can be accessed directly through the limbs() and limb(i) member functions.

NOTE: it's not possible to specify a bit size that is not a multiple of the limb size. No attempt is made to mask out the "excess" bits (for performance reasons) so the results of all operations will be inconsistent. A static_assert() exists to prevent invalid limb sizes.