关键是更新tid
tid += blockDim.x * gridDim.x; (加上所有线程的数量，以便对其他数组元素接着更新)
__global__ void add(int *d_arr, int *d_brr, int *d_crr, int arrLength) { int tid = blockDim.x * blockIdx.x + threadIdx.x; if(tid <arrLength) { d_crr[tid] = d_arr[tid] + d_brr[tid]; tid += blockDim.x * gridDim.x; } }
转载请注明原文地址: https://www.6miu.com/read-39099.html
