6.7.1 : The C++ kernel
Here, we only have to extract the kernel from the main_intrinsics.cpp file.
The full hadamard_intrinsics_pitch.h file :
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 |
/*************************************** Auteur : Pierre Aubert Mail : pierre.aubert@lapp.in2p3.fr Licence : CeCILL-C ****************************************/ #ifndef __HADAMARD_INTRINSICS_PITCH_H__ #define __HADAMARD_INTRINSICS_PITCH_H__ void hadamard_product(float* tabResult, const float* tabX, const float* tabY, long unsigned int nbElement); #endif |
You can download it here. The full hadamard_intrinsics_pitch.cpp file :
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 |
/*************************************** Auteur : Pierre Aubert Mail : pierre.aubert@lapp.in2p3.fr Licence : CeCILL-C ****************************************/ #include <immintrin.h> #include <string.h> #include "hadamard_intrinsics_pitch.h" ///Do the Hadamard product /** @param[out] tabResult : table of results of tabX*tabY * @param tabX : input table * @param tabY : input table * @param nbElement : number of elements in the tables */ void hadamard_product(float* tabResult, const float* tabX, const float* tabY, long unsigned int nbElement){ long unsigned int vecSize(VECTOR_ALIGNEMENT/sizeof(float)); long unsigned int nbVec(nbElement/vecSize); for(long unsigned int i(0lu); i < nbVec; ++i){ __m256 vecX = _mm256_load_ps(tabX + i*vecSize); __m256 vecY = _mm256_load_ps(tabY + i*vecSize); __m256 vecRes = _mm256_mul_ps(vecX, vecY); _mm256_store_ps(tabResult + i*vecSize, vecRes); } } |
You can download it here.