Fixing CUDA memset value.

      dim3 blockSize, gridSize;

      blockSize. x = 512;

      int blocksNumber = ceil( ( double ) size / ( double ) blockSize. x );

      int gridsNumber = ceil( ( double ) blocksNumber / ( double ) maxCudaGridSize );

      int elementsPerThread = ceil( ( double ) blocksNumber / ( double ) maxCudaGridSize );

      setVectorValueCudaKernel<<< gridSize, blockSize >>>( data, size, value, elementsPerThread );

      cerr << "Grids number is " << gridsNumber << endl;

      for( int gridIdx = 0; gridIdx < gridsNumber; gridIdx ++ )

      {

         cout << "GridIdx = " << gridIdx << endl;

         if( gridIdx < gridsNumber - 1 )

            gridSize. x = maxCudaGridSize;

         else

            gridSize. x = blocksNumber - gridIdx * maxCudaGridSize;

         cout << "GridSize. x = " << gridSize. x << endl;

         setVectorValueCudaKernel<<< gridSize, blockSize >>>                                 

                                 ( &( data[ gridIdx * maxCudaGridSize * blockSize. x ] ),

                                   value );

      }

      return checkCudaDevice;

#else

      cerr << "I am sorry but CUDA support is missing on this system " << __FILE__ << " line " << __LINE__ << "." << endl;