1 __global__ void kernel_convoGene8r( unsigned char *output, int j_dim, int r)
7 // absolute coordinates of base point
8 int j = __umul24( blockIdx.x, blockDim.x ) + threadIdx.x ;
9 int i = __umul24( blockIdx.y, blockDim.y) + threadIdx.y ;
11 // convolution computation
12 for (ic=-r ; ic<=r ; ic++)
13 for(jc=-r ; jc<=r ; jc++)
14 outval0 += masque[ __umul24(ic,L)+jc+r ]
15 *tex2D(tex_img_inc, j+jc, i+ic) ;
17 output[ __umul24(i, j_dim) + j ] = outval0 ;