How can I check for carry-out while using unsigned vector subtraction?

Question

I try to implement an architecture that has only an 8-bit adder-subtractor. But there is an issue I can't solve. When I use subtractor architecture, I need to calculate carry-out, but I couldn't.

library IEEE;
use IEEE.STD_LOGIC_1164.ALL;
use ieee.numeric_std.all;

entity sub is
            port ( a  : in std_logic_vector(7 downto 0);
                   b  : in std_logic_vector(7 downto 0);
                   o  : out std_logic_vector(7 downto 0)
                );
end sub;

architecture Behavioral of sub is
  signal a2,b2 : unsigned (7 downto 0);
begin
            a2<=unsigned(a);
            b2<=unsigned(b);

            o<=std_logic_vector(a2-b2);
 end Behavioral;

---

Edit: I'm talking about "c1" and "c5" signals in the picture.

Answer 1

You need to extend both operands by one bit, so the carry out is collected in the MSB of the result. The result is decomposed into carry bit c and subtraction result o.

library IEEE;
use     IEEE.STD_LOGIC_1164.ALL;
use     IEEE.numeric_std.all;

entity sub is
  port (
    a  : in  std_logic_vector(7 downto 0);
    b  : in  std_logic_vector(7 downto 0);
    o  : out std_logic_vector(7 downto 0);
    c  : out std_logic
  );
end entity;

architecture rtl of sub is
begin
  (c, o) <= std_logic_vector(unsigned('0' & a) - unsigned('0' & b));
end architecture;

---

Note: This is VHDL-2008 code.