![]() ![]() |
Jun 4 2010, 04:49 AM
Post
#1
|
|
|
Newbie ![]() Group: Members Posts: 1 Joined: 23-March 09 Member No.: 2,402 |
Hello,
I'm evaluating CoDeveloper 3.6 as a coDSP builder for a larger LabVIEW FPGA 2009 project. I'm facing to a data corruption when the codeveloper process is longer than a cycle to produce an output. The LabVIEW documentation propose this code for HDL integration: CODE process( clk, reset ) begin if( reset = '1' ) then result <= (others=>'0'); enable_out <= '0'; elsif rising_edge(clk) then result <= x + y; -- result and enable_out follow input by 1 clock cycle if( enable_clr = '1' ) then enable_out <= '0'; elsif( enable_in = '1' ) then enable_out <= '1'; end if; end if; end process; The codeveloper process is supposed to replace the x+y addition. The description of enable_out is: "The HDL code receives enable_in, executes, makes sure outputs are valid and held constant, and asserts enable_out. " If I look on the wrapper code generated by codeveloper the enable chain controller is: CODE enable_chain_controller: process(clk, reset) begin if reset = '1' then enable_out <= '0'; elsif rising_edge(clk) then if enable_in = '1' then enable_out <= '1'; else enable_out <= '0'; end if; end if; end process enable_chain_controller; So enable_out is alway active when reset is not active. I was expected to see someting relative to fifo ready there... Any advice ? Thanks JF |
|
|
|
Jun 20 2010, 06:45 PM
Post
#2
|
|
|
Advanced Member ![]() ![]() ![]() Group: Impulse Staff Posts: 237 Joined: 20-April 06 Member No.: 594 |
Hi,
Actually, given the code for enable_chain_controller above, enable_out should be just a clock delay of enable_in which could explain what you are seeing in regards to an error if the result takes more than one cycle. It would help to see the CoDeveloper project being used, please send a complete set of files to support@impulsec.com. Best Regards, Ed -------------------- Ed Trexel
Impulse Accelerated Technologies, Inc. |
|
|
|
![]() ![]() |
| Lo-Fi Version | Time is now: 10th September 2010 - 10:54 AM |