/data/development/ViennaCL/dev/viennacl/linalg/coordinate_matrix_operations.hpp

Go to the documentation of this file.

#ifndef VIENNACL_COORDINATE_MATRIX_OPERATIONS_HPP_
#define VIENNACL_COORDINATE_MATRIX_OPERATIONS_HPP_

/* =========================================================================
   Copyright (c) 2010-2011, Institute for Microelectronics,
                            Institute for Analysis and Scientific Computing,
                            TU Wien.

                            -----------------
                  ViennaCL - The Vienna Computing Library
                            -----------------

   Project Head:    Karl Rupp                   rupp@iue.tuwien.ac.at
               
   (A list of authors and contributors can be found in the PDF manual)

   License:         MIT (X11), see file LICENSE in the base directory
============================================================================= */

#include "viennacl/forwards.h"
#include "viennacl/ocl/device.hpp"
#include "viennacl/ocl/handle.hpp"
#include "viennacl/ocl/kernel.hpp"
#include "viennacl/scalar.hpp"
#include "viennacl/vector.hpp"
#include "viennacl/tools/tools.hpp"
#include "viennacl/linalg/kernels/coordinate_matrix_kernels.h"

namespace viennacl
{
  namespace linalg
  {
    
    
    // A * x
    template<class SCALARTYPE, unsigned int ALIGNMENT, unsigned int VECTOR_ALIGNMENT>
    vector_expression<const coordinate_matrix<SCALARTYPE, ALIGNMENT>,
                      const vector<SCALARTYPE, VECTOR_ALIGNMENT>, 
                      op_prod > prod_impl(const coordinate_matrix<SCALARTYPE, ALIGNMENT> & mat, 
                                     const vector<SCALARTYPE, VECTOR_ALIGNMENT> & vec)
    {
      return vector_expression<const coordinate_matrix<SCALARTYPE, ALIGNMENT>,
                               const vector<SCALARTYPE, VECTOR_ALIGNMENT>, 
                               op_prod >(mat, vec);
    }
    
    // A * x
    template<class SCALARTYPE, unsigned int ALIGNMENT, unsigned int VECTOR_ALIGNMENT>
    viennacl::vector_expression<const viennacl::coordinate_matrix<SCALARTYPE, ALIGNMENT>,
                                const viennacl::vector<SCALARTYPE, VECTOR_ALIGNMENT>, 
                                viennacl::op_prod > prod_impl(const viennacl::coordinate_matrix<SCALARTYPE, ALIGNMENT> & mat, 
                                                              const viennacl::vector<SCALARTYPE, VECTOR_ALIGNMENT> & vec, 
                                                              size_t NUM_THREADS)
    {
      return viennacl::vector_expression<const viennacl::coordinate_matrix<SCALARTYPE, ALIGNMENT>,
                               const viennacl::vector<SCALARTYPE, VECTOR_ALIGNMENT>, 
                               viennacl::op_prod >(mat, vec);
    }
    
    //namespace {
      template<class TYPE, unsigned int ALIGNMENT, unsigned int VECTOR_ALIGNMENT>
      void prod_impl(const viennacl::coordinate_matrix<TYPE, ALIGNMENT> & mat, 
                     const viennacl::vector<TYPE, VECTOR_ALIGNMENT> & vec,
                           viennacl::vector<TYPE, VECTOR_ALIGNMENT> & result)
      {
        assert(mat.size1() == result.size());
        assert(mat.size2() == vec.size());
        result.clear();
        
        //std::cout << "prod(coordinate_matrix" << ALIGNMENT << ", vector) called with internal_nnz=" << mat.internal_nnz() << std::endl;
        
        viennacl::ocl::kernel & k = viennacl::ocl::get_kernel(viennacl::linalg::kernels::coordinate_matrix<TYPE, ALIGNMENT>::program_name(), "vec_mul");
        unsigned int thread_num = 256; //k.local_work_size(0);
        
        k.local_work_size(0, thread_num);
        
        k.global_work_size(0, 64 * thread_num);  //64 work groups are hard-coded for now. Gives reasonable performance in most cases
        //k.global_work_size(0, thread_num);  //Only one work group
        viennacl::ocl::enqueue(k(mat.handle12(), mat, mat.handle3(),
                                 vec,
                                 result,
                                 viennacl::ocl::local_mem(sizeof(cl_uint)*thread_num),
                                 viennacl::ocl::local_mem(sizeof(TYPE)*thread_num)) );

      }
    //};

  } //namespace linalg



    template <typename SCALARTYPE, unsigned int ALIGNMENT>
    template <unsigned int MAT_ALIGNMENT>
    viennacl::vector<SCALARTYPE, ALIGNMENT> & 
    viennacl::vector<SCALARTYPE, ALIGNMENT>::operator=(const viennacl::vector_expression< const coordinate_matrix<SCALARTYPE, MAT_ALIGNMENT>,
                                                                                          const viennacl::vector<SCALARTYPE, ALIGNMENT>,
                                                                                          viennacl::op_prod> & proxy) 
    {
      // check for the special case x = A * x
      if (proxy.rhs().handle() == this->handle())
      {
        viennacl::vector<SCALARTYPE, ALIGNMENT> result(proxy.rhs().size());
        viennacl::linalg::prod_impl(proxy.lhs(), proxy.rhs(), result);
        *this = result;
        return *this;
      }
      else
      {
        viennacl::linalg::prod_impl(proxy.lhs(), proxy.rhs(), *this);
        return *this;
      }
      return *this;
    }

    //v += A * x
    template <typename SCALARTYPE, unsigned int ALIGNMENT>
    template <unsigned int MAT_ALIGNMENT>
    viennacl::vector<SCALARTYPE, ALIGNMENT> & 
    viennacl::vector<SCALARTYPE, ALIGNMENT>::operator+=(const vector_expression< const coordinate_matrix<SCALARTYPE, MAT_ALIGNMENT>,
                                                                                 const vector<SCALARTYPE, ALIGNMENT>,
                                                                                 op_prod> & proxy) 
    {
      vector<SCALARTYPE, ALIGNMENT> result(proxy.lhs().size1());
      viennacl::linalg::prod_impl(proxy.lhs(), proxy.rhs(), result);
      *this += result;
      return *this;
    }

    template <typename SCALARTYPE, unsigned int ALIGNMENT>
    template <unsigned int MAT_ALIGNMENT>
    viennacl::vector<SCALARTYPE, ALIGNMENT> & 
    viennacl::vector<SCALARTYPE, ALIGNMENT>::operator-=(const vector_expression< const coordinate_matrix<SCALARTYPE, MAT_ALIGNMENT>,
                                                                                 const vector<SCALARTYPE, ALIGNMENT>,
                                                                                 op_prod> & proxy) 
    {
      vector<SCALARTYPE, ALIGNMENT> result(proxy.get_lhs().size1());
      viennacl::linalg::prod_impl(proxy.lhs(), proxy.rhs(), result);
      *this -= result;
      return *this;
    }
    
    
    //free functions:
    template <typename SCALARTYPE, unsigned int ALIGNMENT>
    template <unsigned int MAT_ALIGNMENT>
    viennacl::vector<SCALARTYPE, ALIGNMENT> 
    viennacl::vector<SCALARTYPE, ALIGNMENT>::operator+(const vector_expression< const coordinate_matrix<SCALARTYPE, MAT_ALIGNMENT>,
                                                                                const vector<SCALARTYPE, ALIGNMENT>,
                                                                                op_prod> & proxy) 
    {
      assert(proxy.get_lhs().size1() == size());
      vector<SCALARTYPE, ALIGNMENT> result(size());
      viennacl::linalg::prod_impl(proxy.lhs(), proxy.rhs(), result);
      result += *this;
      return result;
    }

    template <typename SCALARTYPE, unsigned int ALIGNMENT>
    template <unsigned int MAT_ALIGNMENT>
    viennacl::vector<SCALARTYPE, ALIGNMENT> 
    viennacl::vector<SCALARTYPE, ALIGNMENT>::operator-(const vector_expression< const coordinate_matrix<SCALARTYPE, MAT_ALIGNMENT>,
                                                                                const vector<SCALARTYPE, ALIGNMENT>,
                                                                                op_prod> & proxy) 
    {
      assert(proxy.get_lhs().size1() == size());
      vector<SCALARTYPE, ALIGNMENT> result(size());
      viennacl::linalg::prod_impl(proxy.lhs(), proxy.rhs(), result);
      result = *this - result;
      return result;
    }

} //namespace viennacl


#endif