Q: problem about dsplib real fft

Started by jamesztyou August 13, 2002
I had searched the former msgs, but can not find answer. so I post a
help message again.
follow is the test code where FFTSZ is a const integer 256, noscale
defined as 0, scale defined as 1; and rfftws is an Uns array header,
pointed to a 256 elements' Uns array, rfftws equals 0x1200(cording to
myout.map), data are in Q15 format.

cbrev( rfftws, rfftws, FFTSZ/2 );
rfft ( rfftws, FFTSZ, noscale );

//do actual processing.

cbrev( rfftws, rfftws, FFTSZ/2 );
rifft( rfftws, FFTSZ, scale );

and here is the question, the result is quite different from the
former datas... help..