Hi
I have a problem when synthesizing my implementation. I am implementing simple controller for sorting two arrays X={1,3,5,7,9}, Y={2,4,6,8,10}.Two arrays are initially stored in dual port BRAM, RAMB16_S9_S9. My goal is to perform this in less than 15 cycles. To do this, "load-compare-store" for one element need to be done in one cycle. My intention was "read port from A" and "write to port B" in a parallel and pipelined way.
What I did was, as shown below Everything is synchronous, except the "address" setting. Behavioral simulation is okay, but mapper (ISE) complains (also shown below) that every logic is trimmed.
Does anyone design this way ? or anyone has this experience? or maybe problem is somewhere else.
thankyou in advance for comment and suggestion.
------------------------------------------------------------------------------------
--------------------------------
-- Asynchronous address setting Address_port_A