353
社区成员
发帖
与我相关
我的任务
分享
__global__ void testAdd(float* d_data, int width, int depth, float* d_ret)
{
float ret = 0;
int index = blockDim.x*blockDim.x + threadIdx.x;
if (index < width)
{
d_data += index;
d_ret += index;
for (int i =0; i < depth; i++)
{
ret += d_data[0];
d_data += width;
}
d_ret[0] = ret;
}
}
dim3 dimBlock(BLOCK_SIZE);
dim3 dimGrid(size+BLOCK_SIZE-1 / dimBlock.x);
testAdd<<<dimGrid,dimBlock>>>( d_data, width, depth, d_ret);