如 NVidia 论坛上的此线程中所述，看来即使 Fermi 支持非内联函数，nvcc 仍然需要在编译期间拥有所有可用的函数，即在同一个源文件中:没有链接器(是的，这很可惜......).

As explained on this thread on the NVidia forums, it appears that even though Fermi supports non-inlined functions, nvcc still needs to have all the functions available during compilation, i.e. in the same source file: there is no linker (yep, that's a pity...).