sort512.hpp

//////////////////////////////////////////////////////////
/// Code to sort an array of integer or double
/// using avx 512 (targeting intel KNL/SKL).
/// By berenger.bramas@mpcdf.mpg.de 2017.
/// Licence is MIT.
/// Comes without any warranty.
///
///
/// Functions to call:
/// Sort512::Sort(); to sort an array
/// Sort512::SortOmp(); to sort in parallel
/// Sort512::Partition512(); to partition
/// Sort512::SmallSort16V(); to sort a small array
/// (should be less than 16 AVX512 vectors)
///
/// To compile such flags can be used to enable avx 512 and openmp:
/// - KNL
/// Gcc : -mavx512f -mavx512pf -mavx512er -mavx512cd -fopenmp
/// Intel : -xCOMMON-AVX512 -xMIC-AVX512 -qopenmp
/// - SKL
/// Gcc : -mavx512f -mavx512cd -mavx512vl -mavx512bw -mavx512dq -fopenmp
/// Intel : -xCOMMON-AVX512 -xCORE-AVX512 -qopenmp
///
/// Or use "-march=native -mtune=native" if you are already on the right platform ("native can be replaced by "knl" or "skylake")
//////////////////////////////////////////////////////////
#ifndef SORT512_HPP
#define SORT512_HPP

#include <immintrin.h>
#include <climits>
#include <cfloat>
#include <algorithm>
#include <cassert>

#if defined(_OPENMP)
#include <omp.h>
#include "parallelInplace.hpp"
#endif