Added atomic operations.

/***************************************************************************

                          AtomicOperations.h  -  description

                             -------------------

    begin                : Feb 26, 2020

    copyright            : (C) 2020 by Tomas Oberhuber et al.

    email                : tomas.oberhuber@fjfi.cvut.cz

 ***************************************************************************/

/* See Copyright Notice in tnl/Copyright */

// Implemented by: Tomas Oberhuber, Jakub Klinkovsky

#pragma once

#include <TNL/Devices/Sequential.h>

#include <TNL/Devices/Host.h>

#include <TNL/Devices/Cuda.h>

namespace TNL {

namespace Algorithms {

template< typename Device >

struct AtomicOperations{};

template<>

struct AtomicOperations< Devices::Host >

{

   template< typename Value >

   static void add( Value& v, const Value& a )

   {

#pragma omp atomic update

      v += a;

   }

};

template<>

struct AtomicOperations< Devices::Cuda >

{

   template< typename Value >

   __cuda_callable__

   static void add( Value& v, const Value& a )

   {

#ifdef HAVE_CUDA

#if __CUDA_ARCH__ < 600

      if( std::is_same< Value, double >::value )

      {

         unsigned long long int* v_as_ull = ( unsigned long long int* ) &v;

         unsigned long long int old = *v_as_ull, assumed;

         do

         {

            assumed = old;

            old = atomicCAS( v_as_ull,

                             assumed,

                             __double_as_longlong( s + __longlong_as_double( assumed ) ) ) ;

         // Note: uses integer comparison to avoid hang in case of NaN (since NaN != NaN)

         }

         while( assumed != old );

         return;

      }

#endif

      atomicAdd( &v, a );

#endif

   }

};

} //namespace Algorithms

} //namespace TNL

 No newline at end of file