program declare_test !!!!$acc routine(kernel_sub) vector use global_data do i = 1, n A(I)=i b(i)=1 enddo ! copy a, b in the device memory !$acc update device (a,b) !$acc parallel loop gang do i = 1, n call kernel_sub(i) enddo !$acc end parallel ! copy back c from the device memory !$acc update host (c) do i = 1, 10 print *,c(i) enddo end program declare_test