CPP11Sort is an efficient version of parallel quicksort algorithm written by using only the C++11 features without non-standard extensions (such as OpenMP) or third party libraries (such as Intel TBB).
Since CPP11sort is templated for the ability to provide generic sorting, the implementation consists of header files only. The API is defined in the include/cpp11sort.h header file. There are several sort functions presented in the cpp11sort namespace. They follow the practices from the C++ Standard and the C++ Stadnard Library for the iterator a comparator parameters and their requirements. The additional parameter allows users to control the number of threads used for parallel sorting. If not provided, the default value is set to the result of the std::thread::hardware_concurrency() call.
The sample program that utilizes CPP11sort may look like as follows:
#include <cassert>
#include <random>
#include <vector>
#include <cpp11sort.h>
int main()
{
long n = 1'000'000'000;
std::vector<int> v;
std::mt19937 eng(std::random_device{}());
std::uniform_int_distribution<int> dist;
while (n--)
v.push_back(dist(eng));
cpp11sort::sort(v.begin(), v.end());
for (long i = 1; i < v.size(); i++)
assert(v[i - 1] <= v[i]);
}Additionally, in the benchmark folder, there is a benchmark program that allows to compare CPP11sort with its various competitors for different type of initial data distributions. Its usage should be fully undestandable from inspecting its source code.
The code was initiali written by Klára Schovánková as a part or her Master’s thesis at the Faculty of Information Technology, Czech Technical University in Prague, Czechia. Then, it has been published by Daniel Langr in this repository in a partially rewritten form. The scientific article that describes the algorithm and the implementation in detail can be found here: https://doi.org/10.1002/cpe.6606.