早些时候我使用NVCC编译代码的时候有些遇见一个很头疼的问题,Debug和Release模式下编译执行的结果竟然不同。下面详细叙述下我的问题和解决方案,以及给诸位同仁的建议。
先贴出我出错的代码:
//Prewitt
__global__ void kernelGrayPrewitt(uchar *src, uchar *dst,unsigned int width, unsigned int heigth, const uchar thresh)
{
int col = threadIdx.x + blockIdx.x*blockDim.x;
int row = threadIdx.y + blockIdx.y*blockDim.y;
if (col < 1 || row < 1 || row > heigth - 1 || col > width - 2)
{
return;
}
int offset = row*width + col;
int offsetUp = (row-1) * width + col;
int offsetDown = (row+1) * width + col;
uchar data[3][3] =
{ {src[offsetUp-1], src[offsetUp], src[offsetUp+1]},
{src[offset-1], src[offset], src[offset+1] },
{src[offsetDown-1], src[offsetDown], src[offsetDown+1]}};//prepare the data.
unsigned int po1 = data[0][2]+data[1][2]+data[