LIBRARY ieee;
USE ieee.std_logic_1164.ALL;
USE ieee.std_logic_signed.ALL;
USE ieee.numeric_std.ALL;
USE work.FIVEn_DFT_PKG.ALL;

ENTITY radix_2_cell_winograd IS
  GENERIC(
    w_in : natural
    );
  PORT(
    i_clk      :     std_logic;
    i_cos      :     smpl_cos_sin_wb;
    i_sin      :     smpl_cos_sin_wb;
    i_data1_re : IN  smpl_out_radix2;
    i_data1_im : IN  smpl_out_radix2;
    i_data2_re : IN  smpl_out_radix2;
    i_data2_im : IN  smpl_out_radix2;
    o_data1_re : OUT smpl_out_radix2;
    o_data1_im : OUT smpl_out_radix2;
    o_data2_re : OUT smpl_out_radix2;
    o_data2_im : OUT smpl_out_radix2
    );
END radix_2_cell_winograd;

ARCHITECTURE radix2 OF radix_2_cell_winograd IS

  TYPE vect_result_multiply IS ARRAY (0 TO 1) OF std_logic_vector(w_in + cst_w_precision_radix2_coeffs_5ndft-2 DOWNTO 0);
  SIGNAL signed_data_im                      : smpl_out_winograd5_signed_5ndft                                  := (OTHERS => '0');
  SIGNAL signed_data_re                      : smpl_out_winograd5_signed_5ndft                                  := (OTHERS => '0');
  SIGNAL multiply_by_2_power_cst_w_precision : std_logic_vector(cst_w_precision_radix2_coeffs_5ndft-3 DOWNTO 0) := (OTHERS => '0');
  SIGNAL data_matrix_im                      : matrix_radix2_cell                                               := (OTHERS => (OTHERS => (OTHERS => '0')));
  SIGNAL data_matrix_re                      : matrix_radix2_cell                                               := (OTHERS => (OTHERS => (OTHERS => '0')));
  SIGNAL data_vect_result_mult_re            : vect_result_multiply                                             := (OTHERS => (OTHERS => '0'));
  SIGNAL data_vect_result_mult_im            : vect_result_multiply                                             := (OTHERS => (OTHERS => '0'));

BEGIN


  -- assign block inputs and outputs their corresponding matrix columns
  data_matrix_re(0)(0) <= i_data1_re;
  data_matrix_re(1)(0) <= i_data2_re;
  data_matrix_im(0)(0) <= i_data1_im;
  data_matrix_im(1)(0) <= i_data2_im;
  o_data1_re           <= data_matrix_re(0)(3);
  o_data2_re           <= data_matrix_re(1)(3);
  o_data1_im           <= data_matrix_im(0)(3);
  o_data2_im           <= data_matrix_im(1)(3);

  

  --instanciating the MULT_BLK_5nDFT(Mult_Path)
  mult_inst1 : ENTITY work.MULT_BLK_5nDFT(Mult_Path)
    GENERIC MAP(
      w_in   => w_in,
      w_mult => cst_w_precision_radix2_coeffs_5ndft
      )
    PORT MAP(i_clk     => i_clk,
             i_data_re => data_matrix_re(1)(0),
             i_data_im => data_matrix_im(1)(0),
             i_cos     => i_cos,
             i_sin     => i_sin,
             o_data_re => data_matrix_re(1)(2),
             o_data_im => data_matrix_im(1)(2)
             );



  -- purpose: calculating the multiplication and the 2 additions/substractions.
  -- The multiplication per 1 in the top of the butterfly is replaced by a
  -- shift with 0s towards the MSBs
  -- inputs: data_matrix_im(x)(0), data_matrix_re(x)(0)
  -- outputs: data_matrix_im(x)(3), data_matrix_re(x)(3)
  radix2_structure : PROCESS(i_clk)
  BEGIN
    IF(rising_edge(i_clk)) THEN
      -- mult per 1 (top)
      data_matrix_im(0)(1)(w_in+cst_w_precision_radix2_coeffs_5ndft DOWNTO 0) <= data_matrix_im(0)(0)(w_in-1)&data_matrix_im(0)(0)(w_in-1)&data_matrix_im(0)(0)(w_in-1)&data_matrix_im(0)(0)(w_in-1 DOWNTO 0)&multiply_by_2_power_cst_w_precision;
      data_matrix_re(0)(1)(w_in+cst_w_precision_radix2_coeffs_5ndft DOWNTO 0) <= data_matrix_re(0)(0)(w_in-1)&data_matrix_re(0)(0)(w_in-1)&data_matrix_re(0)(0)(w_in-1)&data_matrix_re(0)(0)(w_in-1 DOWNTO 0)&multiply_by_2_power_cst_w_precision;
      data_matrix_im(0)(2) <= data_matrix_im(0)(1);
      data_matrix_re(0)(2) <= data_matrix_re(0)(1);

      -- mult (down) : see mult_blk instanciation


      
      --add
      data_matrix_re(0)(3)(w_in+cst_w_precision_radix2_coeffs_5ndft+1 DOWNTO 0) <= std_logic_vector(unsigned(signed(data_matrix_re(0)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft)&data_matrix_re(0)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft DOWNTO 0))+signed(data_matrix_re(1)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft)&data_matrix_re(1)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft DOWNTO 0))));
      data_matrix_re(1)(3)(w_in+cst_w_precision_radix2_coeffs_5ndft+1 DOWNTO 0) <= std_logic_vector(unsigned(signed(data_matrix_re(0)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft)&data_matrix_re(0)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft DOWNTO 0))-signed(data_matrix_re(1)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft)&data_matrix_re(1)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft DOWNTO 0))));
      
      data_matrix_im(0)(3)(w_in+cst_w_precision_radix2_coeffs_5ndft+1 DOWNTO 0) <= std_logic_vector(unsigned(signed(data_matrix_im(0)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft)&data_matrix_im(0)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft DOWNTO 0))+signed(data_matrix_im(1)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft)&data_matrix_im(1)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft DOWNTO 0))));
      data_matrix_im(1)(3)(w_in+cst_w_precision_radix2_coeffs_5ndft+1 DOWNTO 0) <= std_logic_vector(unsigned(signed(data_matrix_im(0)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft)&data_matrix_im(0)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft DOWNTO 0))-signed(data_matrix_im(1)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft)&data_matrix_im(1)(2)(w_in+cst_w_precision_radix2_coeffs_5ndft DOWNTO 0))));
    END IF;

  END PROCESS;

END radix2;