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

Subversion Repositories openrisc

[/] [openrisc/] [trunk/] [or1200/] [rtl/] [verilog/] [or1200_fpu_arith.v] - Rev 622

Go to most recent revision | Compare with Previous | Blame | View Log

//////////////////////////////////////////////////////////////////////
////                                                              ////
////  OR1200 FPU arith                                            ////
////                                                              ////
////  This file is part of the OpenRISC 1200 project              ////
////  http://opencores.org/project,or1k                           ////
////                                                              ////
////  Description                                                 ////
////  Wrapper for floating point arithmetic units.                ////
////                                                              ////
////  To Do:                                                      ////
////   - lf.rem.s and lf.madd.s instruction support               ////
////                                                              ////
////  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_arith
  (
   clk_i,
   opa_i,
   opb_i,
   fpu_op_i,
   rmode_i,
   output_o,
   start_i,
   ready_o,
   ine_o,
   overflow_o,
   underflow_o,
   div_zero_o,
   inf_o,
   zero_o,
   qnan_o,
   snan_o
   );
 
   parameter FP_WIDTH = 32;
   parameter MUL_SERIAL = 1; // 0 for parallel multiplier, 1 for serial
   parameter MUL_COUNT = 34; //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'b11111111_10000000000000000000000;
   parameter SNAN = 31'b11111111_00000000000000000000001;
 
   // fpu operations (fpu_op_i):
   // ========================
   // 000 = add, 
   // 001 = substract, 
   // 010 = multiply, 
   // 011 = divide,
   // 100 = square root - DISABLED - JPB
   // 101 = unused
   // 110 = unused
   // 111 = unused
 
   // Rounding Mode: 
   // ==============
   // 00 = round to nearest even (default), 
   // 01 = round to zero, 
   // 10 = round up, 
   // 11 = round down
 
   input  clk_i;
   input [FP_WIDTH-1:0] opa_i;
   input [FP_WIDTH-1:0] opb_i;
   input [2:0] 		fpu_op_i;
   input [1:0] 		rmode_i;
   input 		start_i;
   output reg 		ready_o;
   output reg [FP_WIDTH-1:0] output_o;
   output reg 		     ine_o;
   output reg 		     overflow_o;
   output reg 		     underflow_o;
   output reg 		     div_zero_o;
   output reg 		     inf_o;
   output reg 		     zero_o;
   output reg 		     qnan_o;
   output reg 		     snan_o;
 
   reg [FP_WIDTH-1:0] 	     s_opa_i;
   reg [FP_WIDTH-1:0] 	     s_opb_i;
   reg [2:0] 		     s_fpu_op_i;
   reg [1:0] 		     s_rmode_i;
   reg 			     s_start_i;
   reg [5:0] 		     s_count; // Max value of 64
 
   reg [FP_WIDTH-1:0] 	     s_output1;   
   reg [FP_WIDTH-1:0] 	     s_output_o; // Comb
 
   reg 			     s_ine_o;
 
   wire 		     s_overflow_o, 
			     s_underflow_o, 
			     s_div_zero_o, 
			     s_inf_o, s_zero_o, s_qnan_o, s_snan_o;
 
   wire 		     s_infa, s_infb;
 
   parameter t_state_waiting = 0,
	       t_state_busy = 1;
 
   reg 			     s_state;
 
   ////	***Add/Substract units signals***
   wire [27:0] 		     prenorm_addsub_fracta_28_o;
   wire [27:0] 		     prenorm_addsub_fractb_28_o;
 
   wire [7:0] 		     prenorm_addsub_exp_o; 
 
   wire [27:0] 		     addsub_fract_o; 
   wire 		     addsub_sign_o;
 
   wire [31:0] 		     postnorm_addsub_output_o; 
   wire 		     postnorm_addsub_ine_o;
 
   ////	***Multiply units signals***
 
   wire [9:0] 		     pre_norm_mul_exp_10;
   wire [23:0] 		     pre_norm_mul_fracta_24 ;
   wire [23:0] 		     pre_norm_mul_fractb_24 ;
   wire [47:0] 		     mul_fract_48;
   wire [47:0] 		     mul_24_fract_48;
   wire 		     mul_24_sign;   
   wire [47:0] 		     serial_mul_fract_48;
   wire 		     serial_mul_sign;   
   wire 		     mul_sign;
   wire [31:0] 		     post_norm_mul_output   ;
   wire 		     post_norm_mul_ine;
 
 
   ////	***Division units signals***
 
   wire [49:0] 		     pre_norm_div_dvdnd;
   wire [26:0] 		     pre_norm_div_dvsor;
   wire [EXP_WIDTH+1:0]      pre_norm_div_exp;   
   wire [26:0] 		     serial_div_qutnt;
   wire [26:0] 		     serial_div_rmndr;
   wire 		     serial_div_sign;   
   wire 		     serial_div_div_zero;
   wire [31:0] 		     post_norm_div_output;
   wire 		     post_norm_div_ine;
 
 
   ////	***Square units***
 
   wire [51:0] 		     pre_norm_sqrt_fracta_o;   
   wire [7:0] 		     pre_norm_sqrt_exp_o;   
   wire [25:0] 		     sqrt_sqr_o;
   wire 		     sqrt_ine_o;
 
   wire [31:0] 		     post_norm_sqrt_output  ;
   wire 		     post_norm_sqrt_ine_o;
 
   //***Add/Substract units***
 
   or1200_fpu_pre_norm_addsub fpu_prenorm_addsub
     (
      .clk_i(clk_i)  ,
      .opa_i(s_opa_i)  ,
      .opb_i(s_opb_i)  ,
      .fracta_28_o(prenorm_addsub_fracta_28_o)  ,
      .fractb_28_o(prenorm_addsub_fractb_28_o)  ,
      .exp_o(prenorm_addsub_exp_o) );
 
   or1200_fpu_addsub fpu_addsub
     (      
	    .clk_i(clk_i)  , 			
	    .fpu_op_i(s_fpu_op_i[0]),		 
	    .fracta_i(prenorm_addsub_fracta_28_o)	 ,	
	    .fractb_i(prenorm_addsub_fractb_28_o)	 ,		
	    .signa_i( s_opa_i[31]),			
	    .signb_i( s_opb_i[31]),				
	    .fract_o(addsub_fract_o)  ,			
	    .sign_o(addsub_sign_o)  );	
 
   or1200_fpu_post_norm_addsub fpu_postnorm_addsub
     (
      .clk_i(clk_i)  ,		
      .opa_i(s_opa_i)  ,
      .opb_i(s_opb_i)  ,	
      .fract_28_i(addsub_fract_o)  ,
      .exp_i(prenorm_addsub_exp_o)  ,
      .sign_i(addsub_sign_o)  ,
      .fpu_op_i(s_fpu_op_i[0]), 
      .rmode_i(s_rmode_i)  ,
      .output_o(postnorm_addsub_output_o)  ,
      .ine_o(postnorm_addsub_ine_o)  
      );
 
   //***Multiply units***
 
   or1200_fpu_pre_norm_mul fpu_pre_norm_mul
     (
      .clk_i(clk_i),
      .opa_i(s_opa_i),
      .opb_i(s_opb_i),
      .exp_10_o(pre_norm_mul_exp_10),
      .fracta_24_o(pre_norm_mul_fracta_24),
      .fractb_24_o(pre_norm_mul_fractb_24));
   /*   
    mul_24 i_mul_24 
    (
    .clk_i(clk_i)  ,
    .fracta_i(pre_norm_mul_fracta_24)  ,
    .fractb_i(pre_norm_mul_fractb_24)  ,
    .signa_i(s_opa_i[31]),
    .signb_i(s_opb_i[31]),
    .start_i(start_i)  ,
    .fract_o(mul_24_fract_48)  , 
    .sign_o(mul_24_sign) 	,
    .ready_o()  );	
    */
   // Serial multiply is default and only one included here
   or1200_fpu_mul fpu_mul
     (
      .clk_i(clk_i)  ,
      .fracta_i(pre_norm_mul_fracta_24)  ,
      .fractb_i(pre_norm_mul_fractb_24)  ,
      .signa_i(s_opa_i[31]),
      .signb_i(s_opb_i[31]),
      .start_i(s_start_i)  ,
      .fract_o(serial_mul_fract_48)  ,
      .sign_o(serial_mul_sign) 	,
      .ready_o()
      );
 
   // Serial or parallel multiplier will be chosen depending on constant 
   // MUL_SERIAL
   assign mul_fract_48 = MUL_SERIAL ? serial_mul_fract_48 : mul_24_fract_48;
   assign mul_sign = MUL_SERIAL ? serial_mul_sign : mul_24_sign;
 
   or1200_fpu_post_norm_mul fpu_post_norm_mul
     (
      .clk_i(clk_i)  ,
      .opa_i(s_opa_i)  ,
      .opb_i(s_opb_i)  ,
      .exp_10_i(pre_norm_mul_exp_10)  ,
      .fract_48_i(mul_fract_48)	 , // Parallel multiplier input
      .sign_i(mul_sign)	 , // Parallel multiplier input
      .rmode_i(s_rmode_i)  ,
      .output_o(post_norm_mul_output)  ,
      .ine_o(post_norm_mul_ine)  
      );
 
   ////***Division units***
 
   or1200_fpu_pre_norm_div fpu_pre_norm_div
     (
      .clk_i(clk_i)  ,
      .opa_i(s_opa_i)  ,
      .opb_i(s_opb_i)  ,
      .exp_10_o(pre_norm_div_exp)  ,
      .dvdnd_50_o(pre_norm_div_dvdnd)	 ,
      .dvsor_27_o(pre_norm_div_dvsor)	 );
 
   or1200_fpu_div fpu_div
     (
      .clk_i(clk_i) ,
      .dvdnd_i(pre_norm_div_dvdnd)  ,
      .dvsor_i(pre_norm_div_dvsor)  ,
      .sign_dvd_i(s_opa_i[31]),
      .sign_div_i(s_opb_i[31]),
      .start_i(s_start_i)  ,
      .ready_o()  ,
      .qutnt_o(serial_div_qutnt)  ,
      .rmndr_o(serial_div_rmndr)  ,
      .sign_o(serial_div_sign)  ,
      .div_zero_o(serial_div_div_zero)	 );
 
   or1200_fpu_post_norm_div fpu_post_norm_div
     (
      .clk_i(clk_i)  ,
      .opa_i(s_opa_i)  ,
      .opb_i(s_opb_i)  ,
      .qutnt_i(serial_div_qutnt) 	,
      .rmndr_i(serial_div_rmndr)  ,
      .exp_10_i(pre_norm_div_exp)  ,
      .sign_i(serial_div_sign)	 ,
      .rmode_i(s_rmode_i) 	,
      .output_o(post_norm_div_output)  ,
      .ine_o(post_norm_div_ine)  );
 
   //////////////////////////////////////////////////////////////////-
 
   // Input Registers
   always @(posedge clk_i)
     begin
	s_opa_i <= opa_i;
	s_opb_i <= opb_i;
	s_fpu_op_i <= fpu_op_i;
	s_rmode_i <= rmode_i;
	s_start_i <= start_i;
     end
 
   // Output registers
   always @(posedge clk_i)
     begin
	output_o <= s_output_o;
	ine_o <= s_ine_o;
	overflow_o <= s_overflow_o;
	underflow_o <= s_underflow_o;
	div_zero_o <= s_div_zero_o & !s_infa;
	inf_o <= s_inf_o;
	zero_o <= s_zero_o;
	qnan_o <= s_qnan_o;
	snan_o <= s_snan_o;
     end
 
   always @(posedge clk_i)
     begin
	if (s_start_i)  begin
	   s_state <= t_state_busy;
	   s_count <= 0;
	end
	else if (s_state == t_state_busy) begin
	   // Ready cases
	   if (((s_count == 6) & ((fpu_op_i==3'd0) | (fpu_op_i==3'd1))) |
	       ((s_count==MUL_COUNT) & (fpu_op_i==3'd2)) |
	       ((s_count==33) & (fpu_op_i==3'd3))) 
	     begin
		s_state <= t_state_waiting;
		ready_o <= 1;
		s_count <= 0;
	     end
	   else
	     s_count <= s_count + 1;
	end // if (s_state == t_state_busy)
	else begin
	   s_state <= t_state_waiting;
	   ready_o <= 0;
	end // else: !if(s_state == t_state_busy)
     end // else: !if(s_start_i)
 
   //// Output Multiplexer
   always @(posedge clk_i)
     begin
	case(fpu_op_i)
	  3'd0,
	    3'd1: begin
	       s_output1 <= postnorm_addsub_output_o;
	       s_ine_o <= postnorm_addsub_ine_o;
	    end
	  3'd2: begin
	     s_output1 <= post_norm_mul_output;
	     s_ine_o <= post_norm_mul_ine;
	  end
	  3'd3: begin
	     s_output1 <= post_norm_div_output;
	     s_ine_o <= post_norm_div_ine;
	  end
	  //	  3'd4: begin
	  //	        s_output1 	<= post_norm_sqrt_output;
	  //		s_ine_o 	<= post_norm_sqrt_ine_o;
	  //	end
	  default: begin
	     s_output1 <= 0;
	     s_ine_o <= 0;
	  end
	endcase // case (fpu_op_i)
     end // always @ (posedge clk_i)
 
   // Infinte exponent
   assign s_infa = &s_opa_i[30:23];
   assign s_infb = &s_opb_i[30:23];
 
   always @*
     begin
	if (s_rmode_i==2'd0 | s_div_zero_o | s_infa | s_infb | s_qnan_o |
	    s_qnan_o) // Round to nearest even
	  s_output_o = s_output1;
	else if (s_rmode_i==2'd1 & (&s_output1[30:23]))
	  // In round-to-zero: the sum of two non-infinity operands is never 
	  // infinity,even if an overflow occures
	  s_output_o = {s_output1[31], 31'b1111111_01111111_11111111_11111111};
	else if (s_rmode_i==2'd2 & (&s_output1[31:23]))
	  // In round-up: the sum of two non-infinity operands is never 
	  // negative infinity,even if an overflow occures
	  s_output_o = {32'b11111111_01111111_11111111_11111111};
	else if (s_rmode_i==2'd3) begin
	   if (((s_fpu_op_i==3'd0) | (s_fpu_op_i==3'd1)) & s_zero_o & 
	       (s_opa_i[31] | (s_fpu_op_i[0] ^ s_opb_i[31])))
	     // In round-down: a-a= -0
	     s_output_o = {1'b1,s_output1[30:0]};
	   else if (s_output1[31:23]==9'b0_11111111)
	     s_output_o = 32'b01111111011111111111111111111111;
	   else
	     s_output_o = s_output1;
	end
	else
	  s_output_o = s_output1;
     end // always @ *
 
   // Exception generation
   assign s_underflow_o = (s_output1[30:23]==8'h00) & s_ine_o;
   assign s_overflow_o = (s_output1[30:23]==8'hff) & s_ine_o;
   assign s_div_zero_o = serial_div_div_zero & fpu_op_i==3'd3;
   assign s_inf_o = s_output1[30:23]==8'hff & !(s_qnan_o | s_snan_o);
   assign s_zero_o = !(|s_output1[30:0]);
   assign s_qnan_o = s_output1[30:0]==QNAN;
   assign s_snan_o = s_output1[30:0]==SNAN;
 
endmodule // or1200_fpu_arith
 
 
 

Go to most recent revision | Compare with Previous | Blame | View Log

powered by: WebSVN 2.1.0

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