I am a student working on a simple MPEG-2 encoder and want to use the Xilinx DCT in Application note 610. The DCT has 8 bit input and 12 bit output, but if I connect a DCT and IDCT in series to test them the result is wrong. I get about the right result for the 7 lowest bits, but I miss the highest bit. If i compere the result of the DCT and Matlab I also get different results.
Have any of you used this reference design?
Anders Rustad Student, NTNU Trondheim, Norway