SIMDProcessing.h

Go to the documentation of this file.

/*
Copyright (C) 2024 Scott Grauer-Gray
 
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
 
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.
 
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
*/
 
#ifndef SIMD_PROCESSING_H_
#define SIMD_PROCESSING_H_
 
#include <math.h>
#include <omp.h>
#include <algorithm>
#include <iostream>
#include "RunEval/RunTypeConstraints.h"
#include "RunImpCPU/RunCPUSettings.h"
 
namespace simd_processing
{
  template<RunData_t T, RunDataVect_t U>
  U LoadPackedDataAligned(unsigned int x, unsigned int y, unsigned int current_disparity,
    const beliefprop::BpLevelProperties& current_bp_level, unsigned int numDispVals,
    const T* inData)
  {
    std::cout << "Data type not supported for loading aligned data" << std::endl;
  }
 
  template<RunData_t T, RunDataVect_t U>
  U LoadPackedDataUnaligned(unsigned int x, unsigned int y, unsigned int current_disparity,
    const beliefprop::BpLevelProperties& current_bp_level, unsigned int numDispVals,
    const T* inData)
  {
    std::cout << "Data type not supported for loading unaligned data" << std::endl;
  }
 
  template<RunDataVect_t T>
  T createSIMDVectorSameData(float data) {
    std::cout << "Data type not supported for creating simd vector" << std::endl;
  }
 
  //
  template<RunDataSingOrVect_t T, RunDataSingOrVect_t U, RunDataSingOrVect_t V>
  V AddVals(const T& val1, const U& val2) { return (val1 + val2); }
 
  template<RunDataSingOrVect_t T, RunDataSingOrVect_t U, RunDataSingOrVect_t V>
  V SubtractVals(const T& val1, const U& val2) { return (val1 - val2); }
 
  template<RunDataSingOrVect_t T, RunDataSingOrVect_t U, RunDataSingOrVect_t V>
  V divideVals(const T& val1, const U& val2) { return (val1 / val2); }
 
  template<RunDataSingOrVect_t T, RunDataSingOrVect_t V>
  T ConvertValToDatatype(V val) { return (T)val; }
 
  template<RunDataSingOrVect_t T>
  T GetMinByElement(const T& val1, const T& val2) { return std::min(val1, val2); }
 
  template<RunData_t T, RunDataVectProcess_t U>
  void StorePackedDataAligned(unsigned int indexDataStore, T* locationDataStore, const U& dataToStore) {
    locationDataStore[indexDataStore] = dataToStore;
  }
 
  template<RunData_t T, RunDataVectProcess_t U>
  void StorePackedDataUnaligned(unsigned int indexDataStore, T* locationDataStore, const U& dataToStore) {
    locationDataStore[indexDataStore] = dataToStore;
  }
};
 
//headers to include differ depending on architecture and CPU vectorization setting
#if defined(COMPILING_FOR_ARM)
#include "ARMTemplateSpFuncts.h"
 
#if (CPU_VECTORIZATION_DEFINE == NEON_DEFINE)
#include "NEONTemplateSpFuncts.h"
#endif //CPU_VECTORIZATION_DEFINE == NEON_DEFINE
 
#else
//needed so that template specializations are used when available
#include "AVXTemplateSpFuncts.h"
 
#if (CPU_VECTORIZATION_DEFINE == AVX_256_DEFINE)
#include "AVX256TemplateSpFuncts.h"
#elif (CPU_VECTORIZATION_DEFINE == AVX_512_DEFINE)
#include "AVX256TemplateSpFuncts.h"
#include "AVX512TemplateSpFuncts.h"
#endif
 
#endif //COMPILING_FOR_ARM
 
#endif //VECT_PROCESSING_FUNCTS_H_