extern "C" __constant__ int my_constant = 314; extern "C" __global__ void sum(const float* x, const float* y, float* out, int count) { for (int i = blockIdx.x * blockDim.x + threadIdx.x; i < count; i += blockDim.x * gridDim.x) { out[i] = x[i] + y[i]; } }