Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
17 years ago

megacore fft 8.0 problem

Hi everyone,

i face a strange problem with megacore fft.

The operation mode i used is

burst mode, Quad output, 1024 points, 2 engines

and the device is Cyclone II

Quartus II v80sp1

i have do a lots of testing for the fft megacore...

first, i would like to read data from fifos which also the megacore produce from the wizard to be the fft inputs (include the real part and image part).

i write control module (in Verilog) to handle the control signals between fifos and fft.

in my design, the fft only accept one frame(1024 data samples) one time, so i need to use burst mode. i read user guide of FFT Megacore Function and understand that there is a small fifo buffer at the sink of the block.

i write finite state machine to control sink_sop, sink_eop, sink_valid carefully, and watch the simulation result.

it goes wrong!

when the sink_ready is asserted, i asserted rdreq(fifo read enable signal) to read data from fifo, and then asserted sink_valid.

sink_sop asserted only at the first data, sink_eop asserted only at the last data, and sink_valid asserted from sink_sop to sink_eop.

there is no any source_error occurs in the process, but the source_real and source_image are wrong. :((the source_sop, source_eop and source_valid are OK, source_ready is asserted) <= the source_real and source_image come out, but the value is incorrect.

more information i got, when i remove fifos and control module. only the fft megacore, and input the sink data from simulation(Vector waveform file). it works correctly!!! Why?:confused:

anyone can do me a favor, please~

ps. attach files are the simulation results

2 Replies

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    Hello again,

    i'm having some problems with FFT megacore, because source_sop is never asserted and sink_ready is asserted just for a few clock's and then comes to low.

    i'm using a 30ns clock.

    I attach my simulator waveforms.

    Thanks.