Hi, I wrote a program for change GpuMat by CUDA in c++. But my CUDA code can not find GpuMat cells correctly and it works wrong. please say how I can access to cells address correctly ?
Main :
GMat_s.upload(MyMat_s);
Read_Stats << <gridS, blockS >> > ((int*)GMat_s.data, MyMat_s.rows, MyMat_s.cols, GMat_s.step); GMat_s.download(NewMat_s);
Device (GPU)
__global__ void Read_Stats(int* mt, int h, int w, size_t step)
{
int x = blockIdx.x * blockDim.x + threadIdx.x;
int y = blockIdx.y * blockDim.y + threadIdx.y;
int index = ((w*row) + col);
mt[index] = 0;
if (index >= h*w)
return;
}