How to fix the the problem that CUDA Code only print C Code's result? [duplicate]

The code is a simple hello world code:

#include <stdio.h>
__global__ void helloFromGPU(void){
        printf("Hello World From GPU!\n");
}

int main(void){
        printf("Hello World From CPU!\n");
        helloFromGPU <<<1, 10>>>();
        cudaDeviceReset();
        return 0;
}

Then I use the following code: nvcc -arch sm_52 hello.cu -o hello

Run ./hello, The output is:

Hello World From CPU!

Why does the device code not work? My OS is Ubuntu 22.04, and CUDA Toolkit version is 12.1, thanks!