比如在 iter=1 这一轮的时候，同时会写入和读取 sum_buf[1] naive_ker = SourceModule(""" __global__ void naive_prefix(double *vec, double *out) { __shared__ double sum_buf[1024]; int tid = threadIdx.x; sum_buf[tid] = vec[tid]; int iter = 1; for (int i=0; i < 10; i++) ...