OpenCores
URL https://opencores.org/ocsvn/openrisc/openrisc/trunk

Subversion Repositories openrisc

[/] [openrisc/] [trunk/] [orpsocv2/] [rtl/] [verilog/] [or1200/] [or1200_fpu_post_norm_addsub.v] - Diff between revs 350 and 360

Go to most recent revision | Show entire file | Details | Blame | View Log

Rev 350 Rev 360
Line 1... Line 1...
 
//////////////////////////////////////////////////////////////////////
 
////                                                              ////
 
////  or1200_fpu_post_norm_addsub                                 ////
 
////                                                              ////
 
////  This file is part of the OpenRISC 1200 project              ////
 
////  http://opencores.org/project,or1k                           ////
 
////                                                              ////
 
////  Description                                                 ////
 
////  post-normalization entity for the addition/subtraction unit ////
 
////                                                              ////
 
////  To Do:                                                      ////
 
////                                                              ////
 
////                                                              ////
 
////  Author(s):                                                  ////
 
////      - Original design (FPU100) -                            ////
 
////        Jidan Al-eryani, jidan@gmx.net                        ////
 
////      - Conv. to Verilog and inclusion in OR1200 -            ////
 
////        Julius Baxter, julius@opencores.org                   ////
 
////                                                              ////
 
//////////////////////////////////////////////////////////////////////
 
//
 
//  Copyright (C) 2006, 2010
 
//
 
//      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. 
 
//
 
 
 
module or1200_fpu_post_norm_addsub
 
  (
 
   clk_i,
 
   opa_i,
 
   opb_i,
 
   fract_28_i,
 
   exp_i,
 
   sign_i,
 
   fpu_op_i,
 
   rmode_i,
 
   output_o,
 
   ine_o
 
   );
 
 
 
   parameter FP_WIDTH = 32;
 
   parameter MUL_SERIAL = 0; // 0 for parallel multiplier, 1 for serial
 
   parameter MUL_COUNT = 11; //11 for parallel multiplier, 34 for serial
 
   parameter FRAC_WIDTH = 23;
 
   parameter EXP_WIDTH = 8;
 
   parameter ZERO_VECTOR = 31'd0;
 
   parameter INF = 31'b1111111100000000000000000000000;
 
   parameter QNAN = 31'b1111111110000000000000000000000;
 
   parameter SNAN = 31'b1111111100000000000000000000001;
 
 
 
   input     clk_i;
 
   input [FP_WIDTH-1:0] opa_i;
 
   input [FP_WIDTH-1:0] opb_i;
 
   input [FRAC_WIDTH+4:0] fract_28_i;
 
   input [EXP_WIDTH-1:0]  exp_i;
 
   input                  sign_i;
 
   input                  fpu_op_i;
 
   input [1:0]             rmode_i;
 
   output reg [FP_WIDTH-1:0]       output_o;
 
   output reg             ine_o;
 
 
 
   wire [FP_WIDTH-1:0]     s_opa_i;
 
   wire [FP_WIDTH-1:0]     s_opb_i;
 
   wire [FRAC_WIDTH+4:0] s_fract_28_i;
 
   wire [EXP_WIDTH-1:0] s_exp_i;
 
   wire s_sign_i;
 
   wire s_fpu_op_i;
 
   wire [1:0] s_rmode_i;
 
   wire [FP_WIDTH-1:0] s_output_o;
 
   wire s_ine_o;
 
   wire s_overflow;
 
 
 
   wire [5:0] s_zeros;
 
   reg [5:0] s_shr1;
 
   reg [5:0] s_shl1;
 
   wire s_shr2, s_carry;
 
 
 
   wire [9:0] s_exp10;
 
   reg [EXP_WIDTH:0] s_expo9_1;
 
   wire [EXP_WIDTH:0] s_expo9_2;
 
   wire [EXP_WIDTH:0] s_expo9_3;
 
 
 
   reg [FRAC_WIDTH+4:0] s_fracto28_1;
 
   wire [FRAC_WIDTH+4:0] s_fracto28_2;
 
   wire [FRAC_WIDTH+4:0] s_fracto28_rnd;
 
 
 
   wire s_roundup;
 
   wire s_sticky;
 
 
 
   wire s_zero_fract;
 
   wire s_lost;
 
   wire s_infa, s_infb;
 
   wire s_nan_in, s_nan_op, s_nan_a, s_nan_b, s_nan_sign;
 
 
 
   assign s_opa_i = opa_i;
 
   assign s_opb_i = opb_i;
 
   assign s_fract_28_i = fract_28_i;
 
   assign s_exp_i = exp_i;
 
   assign s_sign_i = sign_i;
 
   assign s_fpu_op_i = fpu_op_i;
 
   assign s_rmode_i = rmode_i;
 
 
 
   // Output Register
 
   always @(posedge clk_i)
 
     begin
 
        output_o <= s_output_o;
 
        ine_o <= s_ine_o;
 
     end
 
   //*** Stage 1 ****
 
   // figure out the output exponent and how much the fraction has to be 
 
   // shiftd right/left
 
 
 
   assign s_carry = s_fract_28_i[27];
 
 
 
   reg [5:0] lzeroes;
 
 
 
   always @(s_fract_28_i)
 
     casex(s_fract_28_i[26:0])   // synopsys full_case parallel_case
 
       27'b1??????????????????????????: lzeroes <=  0;
 
       27'b01?????????????????????????: lzeroes <=  1;
 
       27'b001????????????????????????: lzeroes <=  2;
 
       27'b0001???????????????????????: lzeroes <=  3;
 
       27'b00001??????????????????????: lzeroes <=  4;
 
       27'b000001?????????????????????: lzeroes <=  5;
 
       27'b0000001????????????????????: lzeroes <=  6;
 
       27'b00000001???????????????????: lzeroes <=  7;
 
       27'b000000001??????????????????: lzeroes <=  8;
 
       27'b0000000001?????????????????: lzeroes <=  9;
 
       27'b00000000001????????????????: lzeroes <=  10;
 
       27'b000000000001???????????????: lzeroes <=  11;
 
       27'b0000000000001??????????????: lzeroes <=  12;
 
       27'b00000000000001?????????????: lzeroes <=  13;
 
       27'b000000000000001????????????: lzeroes <=  14;
 
       27'b0000000000000001???????????: lzeroes <=  15;
 
       27'b00000000000000001??????????: lzeroes <=  16;
 
       27'b000000000000000001?????????: lzeroes <=  17;
 
       27'b0000000000000000001????????: lzeroes <=  18;
 
       27'b00000000000000000001???????: lzeroes <=  19;
 
       27'b000000000000000000001??????: lzeroes <=  20;
 
       27'b0000000000000000000001?????: lzeroes <=  21;
 
       27'b00000000000000000000001????: lzeroes <=  22;
 
       27'b000000000000000000000001???: lzeroes <=  23;
 
       27'b0000000000000000000000001??: lzeroes <=  24;
 
       27'b00000000000000000000000001?: lzeroes <=  25;
 
       27'b000000000000000000000000001: lzeroes <=  26;
 
       27'b000000000000000000000000000: lzeroes <=  27;
 
     endcase
 
 
 
   assign s_zeros = s_fract_28_i[27] ? 0 : lzeroes;
 
 
 
   // negative flag & large flag & exp          
 
   assign s_exp10 = {2'd0,s_exp_i} + {9'd0,s_carry} - {4'd0,s_zeros};
 
 
 
   always @(posedge clk_i)
 
     begin
 
        if (s_exp10[9] | !(|s_exp10))
 
          begin
 
             s_shr1 <= 0;
 
             s_expo9_1 <= 9'd1;
 
 
 
             if (|s_exp_i)
 
               s_shl1 <= s_exp_i[5:0] - 6'd1;
 
             else
 
               s_shl1 <= 0;
 
 
 
          end
 
        else if (s_exp10[8])
 
          begin
 
             s_shr1 <= 0;
 
             s_shl1 <= 0;
 
             s_expo9_1 <= 9'b011111111;
 
          end
 
        else
 
          begin
 
             s_shr1 <= {5'd0,s_carry};
 
             s_shl1 <= s_zeros;
 
             s_expo9_1 <= s_exp10[8:0];
 
          end // else: !if(s_exp10[8])
 
     end // always @ (posedge clk_i)
 
 
 
   //-
 
   // *** Stage 2 ***
 
   // Shifting the fraction and rounding
 
 
 
   always @(posedge clk_i)
 
     if (|s_shr1)
 
       s_fracto28_1 <= s_fract_28_i >> s_shr1;
 
     else
 
       s_fracto28_1 <= s_fract_28_i << s_shl1;
 
 
 
   assign s_expo9_2 = (s_fracto28_1[27:26]==2'b00) ?
 
                      s_expo9_1 - 9'd1 : s_expo9_1;
 
 
 
   // round
 
   //check last bit, before and after right-shift
 
   assign s_sticky = s_fracto28_1[0] | (s_fract_28_i[0] & s_fract_28_i[27]);
 
 
 
   assign s_roundup = s_rmode_i==2'b00 ?
 
                      // round to nearset even
 
                      s_fracto28_1[2] & ((s_fracto28_1[1] | s_sticky) |
 
                                         s_fracto28_1[3]) :
 
                      s_rmode_i==2'b10 ?
 
                      // round up
 
                      (s_fracto28_1[2] | s_fracto28_1[1] | s_sticky) & !s_sign_i:
 
                      s_rmode_i==2'b11 ?
 
                      // round down
 
                      (s_fracto28_1[2] | s_fracto28_1[1] | s_sticky) & s_sign_i :
 
                      // round to zero(truncate = no rounding)
 
                      0;
 
 
 
   assign s_fracto28_rnd = s_roundup ?
 
                           s_fracto28_1+28'b0000_0000_0000_0000_0000_0000_1000 :
 
                           s_fracto28_1;
 
 
 
   // ***Stage 3***
 
   // right-shift after rounding (if necessary)
 
   assign s_shr2 = s_fracto28_rnd[27];
 
 
 
   assign s_expo9_3 = (s_shr2 &  s_expo9_2!=9'b011111111) ?
 
                      s_expo9_2 + 9'b000000001 : s_expo9_2;
 
 
 
   assign s_fracto28_2 = s_shr2 ? {1'b0,s_fracto28_rnd[27:1]} : s_fracto28_rnd;
 
 
 
   ////-
 
 
 
   assign s_infa = &s_opa_i[30:23];
 
   assign s_infb = &s_opb_i[30:23];
 
 
 
   assign s_nan_a = s_infa &  (|s_opa_i[22:0]);
 
   assign s_nan_b = s_infb &  (|s_opb_i[22:0]);
 
 
 
   assign s_nan_in = s_nan_a | s_nan_b;
 
 
 
   // inf-inf=Nan
 
   assign s_nan_op = (s_infa & s_infb) &
 
                     (s_opa_i[31] ^ (s_fpu_op_i ^ s_opb_i[31]));
 
 
 
   assign s_nan_sign = (s_nan_a & s_nan_b) ? s_sign_i :
 
                       s_nan_a ?
 
                       s_opa_i[31] : s_opb_i[31];
 
 
 
   // check if result is inexact;
 
   assign s_lost = (s_shr1[0] & s_fract_28_i[0]) |
 
                   (s_shr2 & s_fracto28_rnd[0]) | (|s_fracto28_2[2:0]);
 
 
 
   assign s_ine_o = (s_lost | s_overflow) & !(s_infa | s_infb);
 
 
 
   assign s_overflow = s_expo9_3==9'b011111111 & !(s_infa | s_infb);
 
 
 
   // '1' if fraction result is zero
 
   assign s_zero_fract = s_zeros==27 & !s_fract_28_i[27];
 
 
 
 
 
   // Generate result
 
   assign s_output_o = (s_nan_in | s_nan_op) ?
 
                       {s_nan_sign,QNAN} :
 
                       (s_infa | s_infb) | s_overflow ?
 
                       {s_sign_i,INF} :
 
                       s_zero_fract ?
 
                       {s_sign_i,ZERO_VECTOR} :
 
                       {s_sign_i,s_expo9_3[7:0],s_fracto28_2[25:3]};
 
 
 
endmodule // or1200_fpu_post_norm_addsub
 
 
 
 
 
 
 No newline at end of file
 No newline at end of file

powered by: WebSVN 2.1.0

© copyright 1999-2024 OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.