Scaling problem of TI FFT library

Dear Sir..
I'm trying to implement a fft program with TI 320f2812.
I'm using Ti FFT library.

I can't get the correct amplitude of result.

1. If the input value is small than 100000, the output  is "0", Is
this right?
2. In my opinion, Question 1 has relation to this question.
    some sample code use bit shift command(<<16) to input data,
    Should is I use this command too?
    and that case, What should I do with the amplitude of fft result.
3. The amplitude of fft result seems to be 1/1000 of expected
amplitude.

any answer will be appreciated.
Thanks.