URL
https://opencores.org/ocsvn/fpu100/fpu100/trunk
Subversion Repositories fpu100
[/] [fpu100/] [tags/] [arelease/] [post_norm_mul.vhd] - Rev 21
Compare with Previous | Blame | View Log
------------------------------------------------------------------------------- -- -- Project: <Floating Point Unit Core> -- -- Description: post-normalization entity for the multiplication unit ------------------------------------------------------------------------------- -- -- 100101011010011100100 -- 110000111011100100000 -- 100000111011000101101 -- 100010111100101111001 -- 110000111011101101001 -- 010000001011101001010 -- 110100111001001100001 -- 110111010000001100111 -- 110110111110001011101 -- 101110110010111101000 -- 100000010111000000000 -- -- Author: Jidan Al-eryani -- E-mail: jidan@gmx.net -- -- Copyright (C) 2006 -- -- This source file may be used and distributed without -- restriction provided that this copyright statement is not -- removed from the file and that any derivative work contains -- the original copyright notice and the associated disclaimer. -- -- THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY -- EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED -- TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS -- FOR A PARTICULAR PURPOSE. IN NO EVENT SHALL THE AUTHOR -- OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, -- INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES -- (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE -- GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR -- BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF -- LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT -- (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT -- OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE -- POSSIBILITY OF SUCH DAMAGE. -- library ieee ; use ieee.std_logic_1164.all; use ieee.std_logic_unsigned.all; use ieee.std_logic_misc.all; library work; use work.fpupack.all; entity post_norm_mul is port( clk_i : in std_logic; opa_i : in std_logic_vector(FP_WIDTH-1 downto 0); opb_i : in std_logic_vector(FP_WIDTH-1 downto 0); exp_10_i : in std_logic_vector(EXP_WIDTH+1 downto 0); fract_48_i : in std_logic_vector(2*FRAC_WIDTH+1 downto 0); -- hidden(1) & fraction(23) sign_i : in std_logic; rmode_i : in std_logic_vector(1 downto 0); output_o : out std_logic_vector(FP_WIDTH-1 downto 0); ine_o : out std_logic ); end post_norm_mul; architecture rtl of post_norm_mul is signal s_expa, s_expb : std_logic_vector(EXP_WIDTH-1 downto 0); signal s_exp_10_i : std_logic_vector(EXP_WIDTH+1 downto 0); signal s_fract_48_i : std_logic_vector(2*FRAC_WIDTH+1 downto 0); signal s_sign_i : std_logic; signal s_output_o : std_logic_vector(FP_WIDTH-1 downto 0); signal s_ine_o, s_overflow : std_logic; signal s_opa_i, s_opb_i : std_logic_vector(FP_WIDTH-1 downto 0); signal s_rmode_i : std_logic_vector(1 downto 0); signal s_zeros : std_logic_vector(5 downto 0); signal s_carry : std_logic; signal s_shr2, s_shl2 : std_logic_vector(5 downto 0); signal s_expo1, s_expo2b : std_logic_vector(8 downto 0); signal s_exp_10a, s_exp_10b : std_logic_vector(9 downto 0); signal s_frac2a : std_logic_vector(47 downto 0); signal s_sticky, s_guard, s_round : std_logic; signal s_roundup : std_logic; signal s_frac_rnd, s_frac3 : std_logic_vector(24 downto 0); signal s_shr3 : std_logic; signal s_r_zeros : std_logic_vector(5 downto 0); signal s_lost : std_logic; signal s_op_0 : std_logic; signal s_expo3 : std_logic_vector(8 downto 0); signal s_infa, s_infb : std_logic; signal s_nan_in, s_nan_op, s_nan_a, s_nan_b : std_logic; begin -- Input Register process(clk_i) begin if rising_edge(clk_i) then s_opa_i <= opa_i; s_opb_i <= opb_i; s_expa <= opa_i(30 downto 23); s_expb <= opb_i(30 downto 23); s_exp_10_i <= exp_10_i; s_fract_48_i <= fract_48_i; s_sign_i <= sign_i; s_rmode_i <= rmode_i; end if; end process; -- Output Register process(clk_i) begin if rising_edge(clk_i) then output_o <= s_output_o; ine_o <= s_ine_o; end if; end process; --*** Stage 1 **** -- figure out the exponent and howmuch the fraction has to be shiftd right/left s_carry <= s_fract_48_i(47); process(clk_i) begin if rising_edge(clk_i) then if s_fract_48_i(47)='0' then s_zeros <= count_l_zeros(s_fract_48_i(46 downto 1)); else s_zeros <= "000000"; end if; s_r_zeros <= count_r_zeros(s_fract_48_i); end if; end process; s_exp_10a <= s_exp_10_i + ("000000000"&s_carry); s_exp_10b <= s_exp_10a - ("0000"&s_zeros); process(clk_i) variable v_shr1, v_shl1 : std_logic_vector(9 downto 0); begin if rising_edge(clk_i) then if s_exp_10a(9)='1' or s_exp_10a="0000000000" then v_shr1 := "0000000001" - s_exp_10a + ("000000000"&s_carry); v_shl1 := (others =>'0'); s_expo1 <= "000000001"; else if s_exp_10b(9)='1' or s_exp_10b="0000000000" then v_shr1 := (others =>'0'); v_shl1 := ("0000"&s_zeros) - s_exp_10a; s_expo1 <= "000000001"; elsif s_exp_10b(8)='1' then v_shr1 := (others =>'0'); v_shl1 := (others =>'0'); s_expo1 <= "011111111"; else v_shr1 := ("000000000"&s_carry); v_shl1 := ("0000"&s_zeros); s_expo1 <= s_exp_10b(8 downto 0); end if; end if; if v_shr1(6)='1' then --"110000" = 48; maximal shift-right postions s_shr2 <= "111111"; else s_shr2 <= v_shr1(5 downto 0); end if; s_shl2 <= v_shl1(5 downto 0); end if; end process; -- *** Stage 2 *** -- Shifting the fraction and rounding -- shift the fraction process(clk_i) begin if rising_edge(clk_i) then if s_shr2 /= "000000" then s_frac2a <= shr(s_fract_48_i, s_shr2); else s_frac2a <= shl(s_fract_48_i, s_shl2); end if; end if; end process; s_expo2b <= s_expo1 - "000000001" when s_frac2a(46)='0' else s_expo1; -- signals if precision was last during the right-shift above s_lost <= '1' when (s_shr2+("00000"&s_shr3)) > s_r_zeros else '0'; -- ***Stage 3*** -- Rounding -- 23 -- | -- xx00000000000000000000000grsxxxxxxxxxxxxxxxxxxxx -- guard bit: s_frac2a(23) (LSB of output) -- round bit: s_frac2a(22) s_guard <= s_frac2a(22); s_round <= s_frac2a(21); s_sticky <= or_reduce(s_frac2a(20 downto 0)) or s_lost; s_roundup <= s_guard and ((s_round or s_sticky)or s_frac2a(23)) when s_rmode_i="00" else -- round to nearset even ( s_guard or s_round or s_sticky) and (not s_sign_i) when s_rmode_i="10" else -- round up ( s_guard or s_round or s_sticky) and (s_sign_i) when s_rmode_i="11" else -- round down '0'; -- round to zero(truncate = no rounding) process(clk_i) begin if rising_edge(clk_i) then if s_roundup='1' then s_frac_rnd <= (s_frac2a(47 downto 23)) + "1"; else s_frac_rnd <= (s_frac2a(47 downto 23)); end if; end if; end process; s_shr3 <= s_frac_rnd(24); s_expo3 <= s_expo2b + '1' when s_shr3='1' and s_expo2b /= "011111111" else s_expo2b; s_frac3 <= ("0"&s_frac_rnd(24 downto 1)) when s_shr3='1' and s_expo2b /= "011111111" else s_frac_rnd; ---***Stage 4**** -- Output s_op_0 <= not ( or_reduce(s_opa_i(30 downto 0)) and or_reduce(s_opb_i(30 downto 0)) ); s_infa <= '1' when s_expa="11111111" else '0'; s_infb <= '1' when s_expb="11111111" else '0'; s_nan_a <= '1' when (s_infa='1' and or_reduce (s_opa_i(22 downto 0))='1') else '0'; s_nan_b <= '1' when (s_infb='1' and or_reduce (s_opb_i(22 downto 0))='1') else '0'; s_nan_in <= '1' when s_nan_a='1' or s_nan_b='1' else '0'; s_nan_op <= '1' when (s_infa or s_infb)='1' and s_op_0='1' else '0';-- 0 * inf = nan s_overflow <= '1' when s_expo3 = "011111111" and (s_infa or s_infb)='0' else '0'; s_ine_o <= '1' when s_op_0='0' and (s_lost or or_reduce(s_frac2a(22 downto 0)) or s_overflow)='1' else '0'; process(s_sign_i, s_expo3, s_frac3, s_nan_in, s_nan_op, s_infa, s_infb, s_overflow, s_r_zeros) begin if (s_nan_in or s_nan_op)='1' then s_output_o <= s_sign_i & QNAN; elsif (s_infa or s_infb)='1' or s_overflow='1' then s_output_o <= s_sign_i & INF; elsif s_r_zeros=48 then s_output_o <= s_sign_i & ZERO_VECTOR; else s_output_o <= s_sign_i & s_expo3(7 downto 0) & s_frac3(22 downto 0); end if; end process; end rtl;