__global__ void kernel(bool *a1_d, bool *a2_d,int N){
    int th_x=threadIdx.x+blockDim.x*blockIdx.x,
        th_y=threadIdx.y+blockDim.y*blockIdx.y,
        idx=th_y*N+th_x;
    
    int nboors=0, x_temp, y_temp;
    for(int x=th_x-1; x<=th_x+1; ++x){
        for(int y=th_y-1; y<=th_y+1; ++y){
            if(x==th_x && y==th_y) continue;

            if(x==-1) x_temp=N-1;
            else if(x==N) x_temp=0;
            else x_temp=x;

            if(y==-1) y_temp=N-1;
            else if(y==N) y_temp=0;
            else y_temp=y;

            if(a1_d[y_temp*N+x_temp]) ++nboors;
        }
    }
    
    if(a1_d[idx]){
        if(nboors==2 || nboors==3) a2_d[idx]=1;
        else a2_d[idx]=0;
    }else{
        if(nboors==3) a2_d[idx]=1;
    }
}