library ieee;
use ieee.std_logic_1164.all;
use ieee.numeric_std.all;
use ieee.math_real.all; -- for floating point operations

use work.real_vector_pkg.all;

entity ddot_unit is --stands for double-precision dot product unit
  generic (
    N : integer := 2  -- number of vector elements (default 2 for a 2x2 matrix)
  );
  port (
    clk     : in std_logic;
    reset_n : in std_logic;                -- Active-low reset
    enable  : in std_logic;                -- Control signal to trigger dot product calculation
    x_vec   : in my_real_vector(0 to N-1); -- input row vector x
    y_vec   : in my_real_vector(0 to N-1); -- input column vector y as transposed for ease of calculation
    out_valid : out std_logic;             -- Output signal indicating result validity
    dot_out : out my_real                  -- output: dot product
  );
end entity;

architecture Behavioral of ddot_unit is
begin

  process(clk, reset_n) 
    -- Variables used for combinational operations within process
    variable partial_products : my_real_vector(0 to N-1); --Stores individual multiplication results of vector elements.
    variable temp_sum         : my_real;                  --Accumulates the total dot product
    variable temp_valid       : std_logic;                -- Temporarily holds the validity flag for the output
  begin
    if reset_n = '0' then
      dot_out   <= ZERO_REAL;
      out_valid <= '0';

      -- Also reset variables to known state (important for simulation correctness)
      partial_products := (others => ZERO_REAL);
      temp_sum   := ZERO_REAL;
      temp_valid := '0';

    elsif rising_edge(clk) then
      if enable = '1' then
        -- Step 1: Multiply corresponding elements
        for i in 0 to N-1 loop
          partial_products(i) := x_vec(i) * y_vec(i);
        end loop;

        -- Step 2: Sum up all the partial products
        temp_sum := ZERO_REAL;
        for i in 0 to N-1 loop
          temp_sum := temp_sum + partial_products(i);
        end loop;

        temp_valid := '1';

        -- Output assignments after processing
        dot_out    <= temp_sum;
        out_valid  <= temp_valid;
      else
        -- Disable case: reset valid output
        out_valid <= '0';
      end if;
    end if;
  end process;

end architecture;