URL
https://opencores.org/ocsvn/sparc64soc/sparc64soc/trunk
Subversion Repositories sparc64soc
[/] [sparc64soc/] [trunk/] [T1-FPU/] [fpu_add_frac_dp.v] - Rev 3
Go to most recent revision | Compare with Previous | Blame | View Log
// ========== Copyright Header Begin ========================================== // // OpenSPARC T1 Processor File: fpu_add_frac_dp.v // Copyright (c) 2006 Sun Microsystems, Inc. All Rights Reserved. // DO NOT ALTER OR REMOVE COPYRIGHT NOTICES. // // The above named program is free software; you can redistribute it and/or // modify it under the terms of the GNU General Public // License version 2 as published by the Free Software Foundation. // // The above named program is distributed in the hope that it will be // useful, but WITHOUT ANY WARRANTY; without even the implied warranty of // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU // General Public License for more details. // // You should have received a copy of the GNU General Public // License along with this work; if not, write to the Free Software // Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA. // // ========== Copyright Header End ============================================ /////////////////////////////////////////////////////////////////////////////// // // Add pipeline fraction datapath. // /////////////////////////////////////////////////////////////////////////////// module fpu_add_frac_dp ( inq_in1, inq_in2, a1stg_step, a1stg_sngop, a1stg_expadd3_11, a1stg_norm_dbl_in1, a1stg_denorm_dbl_in1, a1stg_norm_sng_in1, a1stg_denorm_sng_in1, a1stg_norm_dbl_in2, a1stg_denorm_dbl_in2, a1stg_norm_sng_in2, a1stg_denorm_sng_in2, a1stg_intlngop, a2stg_frac1_in_frac1, a2stg_frac1_in_frac2, a1stg_2nan_in_inv, a1stg_faddsubop_inv, a2stg_frac1_in_qnan, a2stg_frac1_in_nv, a2stg_frac1_in_nv_dbl, a6stg_step, a2stg_frac2_in_frac1, a2stg_frac2_in_qnan, a2stg_shr_cnt_in, a2stg_shr_cnt_5_inv_in, a2stg_shr_frac2_shr_int, a2stg_shr_frac2_shr_dbl, a2stg_shr_frac2_shr_sng, a2stg_shr_frac2_max, a2stg_expadd_11, a2stg_sub_step, a2stg_fracadd_frac2_inv_in, a2stg_fracadd_frac2_inv_shr1_in, a2stg_fracadd_frac2, a2stg_fracadd_cin_in, a2stg_exp, a2stg_expdec_neq_0, a3stg_faddsubopa, a3stg_sub_in, a3stg_exp10_0_eq0, a3stg_exp10_1_eq0, a3stg_exp_0, a4stg_rnd_frac_add_inv, a3stg_fdtos_inv, a4stg_fixtos_fxtod_inv, a4stg_rnd_sng, a4stg_rnd_dbl, a4stg_shl_cnt_in, add_frac_out_rndadd, add_frac_out_rnd_frac, a4stg_in_of, add_frac_out_shl, a4stg_to_0, fadd_clken_l, rclk, a1stg_in2_neq_in1_frac, a1stg_in2_gt_in1_frac, a1stg_in2_eq_in1_exp, a2stg_frac2_63, a2stg_frac2hi_neq_0, a2stg_frac2lo_neq_0, a3stg_fsdtoix_nx, a3stg_fsdtoi_nx, a3stg_denorm, a3stg_denorm_inv, a3stg_lead0, a4stg_round, a4stg_shl_cnt, a4stg_denorm_inv, a3stg_inc_exp_inv, a3stg_same_exp_inv, a3stg_dec_exp_inv, a4stg_rnd_frac_40, a4stg_rnd_frac_39, a4stg_rnd_frac_11, a4stg_rnd_frac_10, a4stg_rndadd_cout, a4stg_frac_9_0_nx, a4stg_frac_dbl_nx, a4stg_frac_38_0_nx, a4stg_frac_sng_nx, a4stg_frac_neq_0, a4stg_shl_data_neq_0, add_of_out_cout, add_frac_out, se, si, so ); input [62:0] inq_in1; // request operand 1 to op pipes input [63:0] inq_in2; // request operand 2 to op pipes input a1stg_step; // add pipe load input a1stg_sngop; // single precision operation- add 1 stg input a1stg_expadd3_11; // exponent adder sign out- add 1 stg input a1stg_norm_dbl_in1; // select line to normalized fraction 1 input a1stg_denorm_dbl_in1; // select line to normalized fraction 1 input a1stg_norm_sng_in1; // select line to normalized fraction 1 input a1stg_denorm_sng_in1; // select line to normalized fraction 1 input a1stg_norm_dbl_in2; // select line to normalized fraction 2 input a1stg_denorm_dbl_in2; // select line to normalized fraction 2 input a1stg_norm_sng_in2; // select line to normalized fraction 2 input a1stg_denorm_sng_in2; // select line to normalized fraction 2 input a1stg_intlngop; // integer/long input- add 1 stage input a2stg_frac1_in_frac1; // select line to a2stg_frac1 input a2stg_frac1_in_frac2; // select line to a2stg_frac1 input a1stg_2nan_in_inv; // 2 NaN inputs- a1 stage input a1stg_faddsubop_inv; // add/subtract- a1 stage input a2stg_frac1_in_qnan; // make fraction 1 a QNaN input a2stg_frac1_in_nv; // NV- make a new prec QNaN input a2stg_frac1_in_nv_dbl; // NV- make a new double prec QNaN input a6stg_step; // advance the add pipe input a2stg_frac2_in_frac1; // select line to a2stg_frac2 input a2stg_frac2_in_qnan; // make fraction 2 a QNaN input [5:0] a2stg_shr_cnt_in; // right shift count input- add 1 stage input a2stg_shr_cnt_5_inv_in; // right shift count input[5]- add 1 stg input a2stg_shr_frac2_shr_int; // select line to a3stg_frac2 input a2stg_shr_frac2_shr_dbl; // select line to a3stg_frac2 input a2stg_shr_frac2_shr_sng; // select line to a3stg_frac2 input a2stg_shr_frac2_max; // select line to a3stg_frac2 input a2stg_expadd_11; // exponent adder[11]- add 2 stage input a2stg_sub_step; // select line to a3stg_frac2 input a2stg_fracadd_frac2_inv_in; // sel line to main adder input 2 input a2stg_fracadd_frac2_inv_shr1_in; // sel line to main adder in 2 input a2stg_fracadd_frac2; // select line to main adder input 2 input a2stg_fracadd_cin_in; // carry in to main adder- add 1 stage input [5:0] a2stg_exp; // exponent add 2 stage bits[5:0] input a2stg_expdec_neq_0; // exponent will be < 54 input [1:0] a3stg_faddsubopa; // denorm compare lead0[10] input select input a3stg_sub_in; // subtract in main adder- add 3 stage input a3stg_exp10_0_eq0; // exponent[10:0]==0- add 3 stg input a3stg_exp10_1_eq0; // exponent[10:1]==0- add 3 stg input a3stg_exp_0; // exponent[0]- add 3 stg input a4stg_rnd_frac_add_inv; // select line to a4stg_rnd_frac input a3stg_fdtos_inv; // double to single convert- add 3 stg input a4stg_fixtos_fxtod_inv; // int to single/double cvt- add 4 stg input a4stg_rnd_sng; // round to single precision- add 4 stg input a4stg_rnd_dbl; // round to double precision- add 4 stg input [9:0] a4stg_shl_cnt_in; // postnorm shift left count- add 3 stg input add_frac_out_rndadd; // select line to add_frac_out input add_frac_out_rnd_frac; // select line to add_frac_out input a4stg_in_of; // add overflow- select fraction out input add_frac_out_shl; // select line to add_frac_out input a4stg_to_0; // result to max finite on overflow input fadd_clken_l; // add pipe clk enable - asserted low input rclk; // global clock output a1stg_in2_neq_in1_frac; // operand 2 fraction != oprnd 1 frac output a1stg_in2_gt_in1_frac; // operand 2 fraction > oprnd 1 frac output a1stg_in2_eq_in1_exp; // operand 2 exponent == oprnd 1 exp output a2stg_frac2_63; // fraction 2 bit[63]- add 2 stage output a2stg_frac2hi_neq_0; // fraction 2[62:32]in add 2 stage != 0 output a2stg_frac2lo_neq_0; // fraction 2[31:11] in add 2 stage != 0 output a3stg_fsdtoix_nx; // inexact result for flt -> ints output a3stg_fsdtoi_nx; // inexact result for flt -> 32b ints output a3stg_denorm; // denorm output- add 3 stage output a3stg_denorm_inv; // result is not a denorm- add 3 stage output [5:0] a3stg_lead0; // leading 0's count- add 3 stage output a4stg_round; // round the result- add 4 stage output [5:0] a4stg_shl_cnt; // subtract in main adder- add 4 stage output a4stg_denorm_inv; // 0 the exponent output a3stg_inc_exp_inv; // increment the exponent- add 3 stg output a3stg_same_exp_inv; // keep the exponent- add 3 stg output a3stg_dec_exp_inv; // decrement the exponent- add 3 stg output a4stg_rnd_frac_40; // rounded fraction[40]- add 4 stage output a4stg_rnd_frac_39; // rounded fraction[39]- add 4 stage output a4stg_rnd_frac_11; // rounded fraction[11]- add 4 stage output a4stg_rnd_frac_10; // rounded fraction[10]- add 4 stage output a4stg_rndadd_cout; // fraction rounding adder carry out output a4stg_frac_9_0_nx; // inexact double precision result output a4stg_frac_dbl_nx; // inexact double precision result output a4stg_frac_38_0_nx; // inexact single precision result output a4stg_frac_sng_nx; // inexact single precision result output a4stg_frac_neq_0; // fraction != 0- add 4 stage output a4stg_shl_data_neq_0; // left shift result != 0- add 4 stage output add_of_out_cout; // fraction rounding adder carry out output [63:0] add_frac_out; // add fraction output input se; // scan_enable input si; // scan in output so; // scan out wire [62:0] a1stg_in1; wire [54:0] a1stg_in1a; wire a1stg_in1_31_0_neq_0; wire a1stg_in1_50_32_neq_0; wire a1stg_in1_50_0_neq_0; wire a1stg_in1_53_32_neq_0; wire a1stg_in1_51; wire a1stg_in1_54; wire [63:0] a1stg_in2; wire [54:0] a1stg_in2a; wire a1stg_in2_31_0_neq_0; wire a1stg_in2_50_32_neq_0; wire a1stg_in2_50_0_neq_0; wire a1stg_in2_53_32_neq_0; wire a1stg_in2_51; wire a1stg_in2_54; wire a1stg_in2_neq_in1_frac; wire a1stg_in2_gt_in1_frac; wire a1stg_in2_gt_in1; wire a1stg_in2_eq_in1_exp; wire [63:0] a1stg_norm_frac1; wire [63:0] a1stg_norm_frac2; wire [63:0] a2stg_frac1_in; wire [63:0] a2stg_frac1; wire [63:0] a2stg_frac2_in; wire [63:0] a2stg_frac2; wire [63:0] a2stg_frac2a; wire a2stg_frac2_63; wire a2stg_frac2hi_neq_0; wire a2stg_frac2lo_neq_0; wire [115:52] a2stg_shr; wire a2stg_fsdtoix_nx; wire a2stg_fsdtoi_nx; wire a2stg_shr_60_0_neq_0; wire [63:0] a2stg_shr_frac2_inv; wire [63:0] a3stg_frac2_in; wire [63:0] a3stg_frac2; wire [63:0] a3stg_frac1; wire [63:0] a2stg_fracadd_in2; wire [63:0] a2stg_fracadd; wire [63:0] a3stg_ld0_frac; wire [53:0] a2stg_expdec_tmp; wire [53:0] a2stg_expdec; wire [53:0] a3stg_expdec; wire a3stg_ld0_dnrm_10; wire a3stg_denorm; wire a3stg_denorm_inv; wire a3stg_denorma; wire a3stg_denorm_inva; wire [5:0] a3stg_lead0; wire [63:0] a3stg_fracadd; wire a4stg_round_in; wire a4stg_round; wire [5:0] a2stg_shr_cnt; wire [5:3] a2stg_shr_cnta; wire [2:0] a2stg_shr_cnta_5; wire [3:0] a2stg_shr_cnt_5_inv; wire [3:0] a2stg_shr_cnt_5; wire [4:0] a2stg_shr_cnt_4; wire [4:0] a2stg_shr_cnt_3; wire [1:0] a2stg_shr_cnt_2; wire [1:0] a2stg_shr_cnt_1; wire [1:0] a2stg_shr_cnt_0; wire a3stg_sub; wire a3stg_suba; wire [2:0] a4stg_shl_cnt_dec54_0; wire [2:0] a4stg_shl_cnt_dec54_1; wire [2:0] a4stg_shl_cnt_dec54_2; wire [2:0] a4stg_shl_cnt_dec54_3; wire [5:0] a4stg_shl_cnt; wire a2stg_fracadd_frac2_inv; wire a2stg_fracadd_frac2_inv_shr1; wire a4stg_denorm_inv; wire a3stg_fsdtoix_nx; wire a3stg_fsdtoi_nx; wire a2stg_fracadd_cin; wire [63:0] astg_xtra_regs; wire a3stg_inc_exp_inv; wire a3stg_same_exp_inv; wire a3stg_dec_exp_inv; wire a3stg_inc_exp_inva; wire a3stg_fsame_exp_inv; wire a3stg_fdec_exp_inv; wire [63:0] a4stg_rnd_frac_pre1_in; wire [63:0] a4stg_rnd_frac_pre1; wire [63:0] a4stg_rnd_frac_pre2_in; wire [63:0] a4stg_rnd_frac_pre2; wire [63:0] a4stg_rnd_frac_pre3_in; wire [63:0] a4stg_rnd_frac_pre3; wire [63:0] a4stg_rnd_frac; wire [63:0] a4stg_rnd_fraca; wire a4stg_rnd_frac_40; wire a4stg_rnd_frac_39; wire a4stg_rnd_frac_11; wire a4stg_rnd_frac_10; wire [63:0] a4stg_shl_data_in; wire [63:0] a4stg_shl_data; wire [52:0] a4stg_rndadd_tmp; wire a4stg_rndadd_cout; wire [51:0] a4stg_rndadd; wire a4stg_frac_9_0_nx; wire a4stg_frac_dbl_nx; wire a4stg_frac_38_0_nx; wire a4stg_frac_sng_nx; wire a4stg_frac_neq_0; wire a4stg_shl_data_neq_0; wire [126:0] a4stg_shl_tmp; wire [63:0] a4stg_shl; wire add_of_out_cout; wire a5stg_frac_out_rndadd; wire a5stg_frac_out_rnd_frac; wire a5stg_in_of; wire a5stg_frac_out_shl; wire a5stg_to_0; wire [51:0] a5stg_rndadd; wire [63:0] a5stg_rnd_frac; wire [63:0] a5stg_shl; wire [63:0] add_frac_out; wire [63:0] a2stg_shr_tmp2; wire [63:0] a2stg_shr_tmp4; wire [63:0] a2stg_shr_tmp6; wire [63:0] a2stg_shr_tmp8; wire [63:0] a2stg_shr_tmp10; wire [63:0] a2stg_shr_tmp13; wire [63:0] a2stg_shr_tmp18; wire [63:20] a2stg_nx_neq0_84_tmp_1; wire [63:36] a2stg_nx_neq0_84_tmp_2; wire [63:44] a2stg_nx_neq0_84_tmp_3; wire [63:48] a2stg_nx_neq0_84_tmp_4; wire [61:50] a2stg_nx_neq0_84_tmp_5; wire [60:59] a2stg_nx_neq0_84_tmp_6; wire a2stg_nx_neq0_84_tmp_6_51; wire [63:0] a4stg_shl_tmp4; wire se_l; assign se_l = ~se; clken_buf ckbuf_add_frac_dp ( .clk(clk), .rclk(rclk), .enb_l(fadd_clken_l), .tmb_l(se_l) ); /////////////////////////////////////////////////////////////////////////////// // // Add fraction inputs. // // Add input stage. // /////////////////////////////////////////////////////////////////////////////// dffe_s #(63) i_a1stg_in1 ( .din (inq_in1[62:0]), .en (a1stg_step), .clk (clk), .q (a1stg_in1[62:0]), .se (se), .si (), .so () ); dffe_s #(55) i_a1stg_in1a ( .din (inq_in1[54:0]), .en (a1stg_step), .clk (clk), .q (a1stg_in1a[54:0]), .se (se), .si (), .so () ); dffe_s #(64) i_a1stg_in2 ( .din (inq_in2[63:0]), .en (a1stg_step), .clk (clk), .q (a1stg_in2[63:0]), .se (se), .si (), .so () ); dffe_s #(55) i_a1stg_in2a ( .din (inq_in2[54:0]), .en (a1stg_step), .clk (clk), .q (a1stg_in2a[54:0]), .se (se), .si (), .so () ); /////////////////////////////////////////////////////////////////////////////// // // Add normalization and special input injection. // // Add stage 1. // /////////////////////////////////////////////////////////////////////////////// fpu_in2_gt_in1_frac i_a1stg_in2_gt_in1_frac ( .din1 (a1stg_in1a[54:0]), .din2 (a1stg_in2a[54:0]), .sngop (a1stg_sngop), .expadd11 (a1stg_expadd3_11), .expeq (a1stg_in2_eq_in1_exp), .din2_neq_din1 (a1stg_in2_neq_in1_frac), .din2_gt_din1 (a1stg_in2_gt_in1_frac), .din2_gt1_din1 (a1stg_in2_gt_in1) ); assign a1stg_in2_eq_in1_exp= (&{(~(a1stg_in1[62:55] ^ a1stg_in2[62:55])), ((~(a1stg_in1[54:52] ^ a1stg_in2[54:52])) | {3{a1stg_sngop}})}); assign a1stg_norm_frac1[63:0]= ({64{a1stg_norm_dbl_in1}} & {1'b1, a1stg_in1[51:0], 11'b0}) | ({64{a1stg_denorm_dbl_in1}} & {a1stg_in1[51:0], 12'b0}) | ({64{a1stg_norm_sng_in1}} & {1'b1, a1stg_in1[54:32], 40'b0}) | ({64{a1stg_denorm_sng_in1}} & {a1stg_in1[54:32], 41'b0}); assign a1stg_norm_frac2[63:0]= ({64{a1stg_norm_dbl_in2}} & {1'b1, a1stg_in2[51:0], 11'b0}) | ({64{a1stg_denorm_dbl_in2}} & {a1stg_in2[51:0], 12'b0}) | ({64{a1stg_norm_sng_in2}} & {1'b1, a1stg_in2[54:32], 40'b0}) | ({64{a1stg_denorm_sng_in2}} & {a1stg_in2[54:32], 41'b0}) | ({64{a1stg_intlngop}} & a1stg_in2[63:0]); assign a2stg_frac1_in[63:0]= ({64{(a1stg_faddsubop_inv || (!((a1stg_in2_gt_in1 && a1stg_2nan_in_inv) || a2stg_frac1_in_frac1)))}} & {a1stg_norm_frac1[63], (a1stg_norm_frac1[62] || a2stg_frac1_in_qnan), (a1stg_norm_frac1[61:40] | {22{a2stg_frac1_in_nv}}), (a1stg_norm_frac1[39:11] | {29{a2stg_frac1_in_nv_dbl}}), a1stg_norm_frac1[10:0]}) | ({64{(a2stg_frac1_in_frac2 && (a1stg_in2_gt_in1 || a2stg_frac1_in_frac1))}} & {a1stg_norm_frac2[63], (a1stg_norm_frac2[62] || a2stg_frac1_in_qnan), (a1stg_norm_frac2[61:40] | {22{a2stg_frac1_in_nv}}), (a1stg_norm_frac2[39:11] | {29{a2stg_frac1_in_nv_dbl}}), a1stg_norm_frac2[10:0]}); dffe_s #(64) i_a2stg_frac1 ( .din (a2stg_frac1_in[63:0]), .en (a6stg_step), .clk (clk), .q (a2stg_frac1[63:0]), .se (se), .si (), .so () ); assign a2stg_frac2_in[63:0]= ({64{a1stg_faddsubop_inv}} & {a1stg_norm_frac2[63], (a1stg_norm_frac2[62] || a2stg_frac2_in_qnan), a1stg_norm_frac2[61:0]}) | ({64{(a2stg_frac2_in_frac1 && (!a1stg_in2_gt_in1))}} & {a1stg_norm_frac2[63], (a1stg_norm_frac2[62] || a2stg_frac2_in_qnan), a1stg_norm_frac2[61:0]}) | ({64{(a2stg_frac2_in_frac1 && a1stg_in2_gt_in1)}} & a1stg_norm_frac1[63:0]); dffe_s #(64) i_a2stg_frac2 ( .din (a2stg_frac2_in[63:0]), .en (a6stg_step), .clk (clk), .q (a2stg_frac2[63:0]), .se (se), .si (), .so () ); dffe_s #(64) i_a2stg_frac2a ( .din (a2stg_frac2_in[63:0]), .en (a6stg_step), .clk (clk), .q (a2stg_frac2a[63:0]), .se (se), .si (), .so () ); /////////////////////////////////////////////////////////////////////////////// // // Add pipe right shift // - shift the smaller fraction right for adds and subtracts // - shift the fraction right for float to integer conversion // // Add stage 2. // /////////////////////////////////////////////////////////////////////////////// assign a2stg_frac2_63= a2stg_frac2[63]; assign a2stg_frac2hi_neq_0= (|a2stg_frac2[62:32]); assign a2stg_frac2lo_neq_0= (|a2stg_frac2[31:11]); // a2 stage right shifter assign a2stg_shr_tmp2[63:0] = ({{24{a2stg_shr_cnt_5[0]}}, {16{a2stg_shr_cnt_5[1]}}, {13{a2stg_shr_cnt_5[2]}}, {11{a2stg_shr_cnt_5[3]}}} & {32'h00000000, a2stg_frac2a[63:32]}) | ({{24{a2stg_shr_cnt_5_inv[0]}}, {16{a2stg_shr_cnt_5_inv[1]}}, {13{a2stg_shr_cnt_5_inv[2]}}, {11{a2stg_shr_cnt_5_inv[3]}}} & a2stg_frac2a[63:0]); assign a2stg_shr_tmp4[63:0] = ({{24{a2stg_shr_cnt_4[0]}}, {16{a2stg_shr_cnt_4[1]}}, {13{a2stg_shr_cnt_4[2]}}, {11{a2stg_shr_cnt_4[3]}}} & {16'h0000, a2stg_shr_tmp2[63:16]}) | ({{43{~a2stg_shr_cnt_4[4]}}, {21{~a2stg_shr_cnt_4[4]}}} & a2stg_shr_tmp2[63:0]); assign a2stg_shr_tmp6[63:0] = ~(({{24{a2stg_shr_cnt_3[0]}}, {16{a2stg_shr_cnt_3[1]}}, {13{a2stg_shr_cnt_3[2]}}, {11{a2stg_shr_cnt_3[3]}}} & {8'h00, a2stg_shr_tmp4[63:8]}) | ({64{~a2stg_shr_cnt_3[4]}} & a2stg_shr_tmp4[63:0])); assign a2stg_shr_tmp8[63:0] = ~(({{43{a2stg_shr_cnt_2[0]}}, {21{a2stg_shr_cnt_2[0]}}} | a2stg_shr_tmp6[63:0]) & ({64{~a2stg_shr_cnt_2[1]}} | {4'hf, a2stg_shr_tmp6[63:4]})); assign a2stg_shr_tmp10[63:0] = ~(({{43{a2stg_shr_cnt_1[0]}}, {21{a2stg_shr_cnt_1[0]}}} & {2'b00, a2stg_shr_tmp8[63:2]}) | ({64{~a2stg_shr_cnt_1[1]}} & a2stg_shr_tmp8[63:0])); assign a2stg_shr[115:52] = ~(({{43{a2stg_shr_cnt_0[0]}}, {21{a2stg_shr_cnt_0[0]}}} | a2stg_shr_tmp10[63:0]) & ({64{~a2stg_shr_cnt_0[1]}} | {1'b1, a2stg_shr_tmp10[63:1]})); assign a2stg_shr_tmp18[63:0] = ~a2stg_shr_tmp2[63:0]; assign a2stg_shr_tmp13[63:0] = a2stg_shr_tmp4[63:0]; // a2 stage nx signals assign a2stg_fsdtoi_nx = (| a2stg_shr_tmp13[31:0]) | (~(& a2stg_shr_tmp6[31:24])) | (| a2stg_shr_tmp8[31:28]) | (~(& a2stg_shr_tmp10[31:30])) | a2stg_shr[83]; assign a2stg_nx_neq0_84_tmp_1[63:20] = ~((a2stg_frac2a[43:0] & {44{a2stg_shr_cnt[5]}}) | ({a2stg_frac2a[11:0], 32'h00000000} & {44{~a2stg_shr_cnt[5]}})); assign a2stg_nx_neq0_84_tmp_2[63:36] = ~(({a2stg_shr_tmp18[27:12], a2stg_nx_neq0_84_tmp_1[63:52]} | {28{~a2stg_shr_cnt[4]}}) & (a2stg_nx_neq0_84_tmp_1[63:36] | {28{a2stg_shr_cnt[4]}})); assign a2stg_nx_neq0_84_tmp_3[63:44] = ~(({a2stg_shr_tmp13[19:12], a2stg_nx_neq0_84_tmp_2[63:52]} & {20{a2stg_shr_cnt[3]}}) | (a2stg_nx_neq0_84_tmp_2[63:44] & {20{~a2stg_shr_cnt[3]}})); assign a2stg_nx_neq0_84_tmp_4[63:48] = ~(({a2stg_shr_tmp6[15:12], a2stg_nx_neq0_84_tmp_3[63:52]} | {16{~a2stg_shr_cnt[2]}}) & (a2stg_nx_neq0_84_tmp_3[63:48] | {16{a2stg_shr_cnt[2]}})); assign a2stg_nx_neq0_84_tmp_5[61:50] = ~((a2stg_nx_neq0_84_tmp_4[63:52] & {12{a2stg_shr_cnt[1]}}) | (a2stg_nx_neq0_84_tmp_4[61:50] & {12{~a2stg_shr_cnt[1]}})); assign a2stg_nx_neq0_84_tmp_6[59] = ~(a2stg_shr_cnt[0] | a2stg_nx_neq0_84_tmp_5[60]); assign a2stg_nx_neq0_84_tmp_6[60] = ~(~a2stg_shr_cnt[0] | a2stg_nx_neq0_84_tmp_5[61]); assign a2stg_nx_neq0_84_tmp_6_51 = ~((a2stg_nx_neq0_84_tmp_5[52] | ~a2stg_shr_cnt[0]) & (a2stg_nx_neq0_84_tmp_5[51] | a2stg_shr_cnt[0])); assign a2stg_fsdtoix_nx = (~(& a2stg_nx_neq0_84_tmp_1[51:20]) | (| a2stg_nx_neq0_84_tmp_2[51:36]) | ~(& a2stg_nx_neq0_84_tmp_3[51:44]) | (| a2stg_nx_neq0_84_tmp_4[51:48]) | ~(& a2stg_nx_neq0_84_tmp_5[51:50]) | a2stg_nx_neq0_84_tmp_6_51); assign a2stg_shr_60_0_neq_0 = (~(& a2stg_nx_neq0_84_tmp_1[60:20]) | (| a2stg_nx_neq0_84_tmp_2[60:45]) | ~(& a2stg_nx_neq0_84_tmp_3[60:53]) | (| a2stg_nx_neq0_84_tmp_4[60:57]) | ~(& a2stg_nx_neq0_84_tmp_5[60:59]) | (| a2stg_nx_neq0_84_tmp_6[60:59])); assign a2stg_shr_frac2_inv[63:0]= (~(({64{a2stg_shr_frac2_shr_int}} & {1'b0, a2stg_shr[115:61], a2stg_shr_60_0_neq_0, 7'b0}) | ({64{(a2stg_shr_frac2_shr_dbl && a2stg_expadd_11)}} & a2stg_shr[115:52]) | ({64{(a2stg_shr_frac2_shr_sng && a2stg_expadd_11)}} & {a2stg_shr[115:84], 32'b0}) | ({64{(a2stg_shr_frac2_max && (!a2stg_expadd_11))}} & 64'h7fffffffffffffff) | ({64{(!a6stg_step)}} & a3stg_frac2[63:0]))); assign a3stg_frac2_in[63:0]= (~(a2stg_shr_frac2_inv[63:0] ^ {64{a2stg_sub_step}})); dff_s #(64) i_a3stg_frac2 ( .din (a3stg_frac2_in[63:0]), .clk (clk), .q (a3stg_frac2[63:0]), .se (se), .si (), .so () ); /////////////////////////////////////////////////////////////////////////////// // // Pipe the other/larger fraction to stage 3. // // Add stage 2. // /////////////////////////////////////////////////////////////////////////////// dffe_s #(64) i_a3stg_frac1 ( .din ({1'b0, a2stg_frac1[63:1]}), .en (a6stg_step), .clk (clk), .q (a3stg_frac1[63:0]), .se (se), .si (), .so () ); /////////////////////////////////////////////////////////////////////////////// // // Add pipe adder. // // Add stage 2. // /////////////////////////////////////////////////////////////////////////////// assign a2stg_fracadd_in2[63:0]= ({64{a2stg_fracadd_frac2_inv}} & (~a2stg_frac2[63:0])) | ({64{a2stg_fracadd_frac2_inv_shr1}} & (~{1'b0, a2stg_frac2[63:1]})) | ({64{a2stg_fracadd_frac2}} & a2stg_frac2[63:0]); assign a2stg_fracadd[63:0]= (a2stg_frac1[63:0] + a2stg_fracadd_in2[63:0] + {63'b0, a2stg_fracadd_cin}); dffe_s #(64) i_a3stg_ld0_frac ( .din (a2stg_fracadd[63:0]), .en (a6stg_step), .clk (clk), .q (a3stg_ld0_frac[63:0]), .se (se), .si (), .so () ); /////////////////////////////////////////////////////////////////////////////// // // Add pipe exponent decode- used to identify denorm results. // // Add stage 2. // /////////////////////////////////////////////////////////////////////////////// // assign a2stg_expdec_tmp[107:0]= ({54'b0, 54'h20000000000000} >> a2stg_exp[5:0]); assign a2stg_expdec_tmp[53:0] = 54'h20000000000000 >> a2stg_exp[5:0] ; assign a2stg_expdec[53:0]= a2stg_expdec_tmp[53:0] & {54{a2stg_expdec_neq_0}}; dffe_s #(54) i_a3stg_expdec ( .din (a2stg_expdec[53:0]), .en (a6stg_step), .clk (clk), .q (a3stg_expdec[53:0]), .se (se), .si (), .so () ); /////////////////////////////////////////////////////////////////////////////// // // Add pipe denorm comparator to identify denorm results. // // Add stage 3. // /////////////////////////////////////////////////////////////////////////////// assign a3stg_ld0_dnrm_10= (a3stg_faddsubopa[0] && a3stg_ld0_frac[10]) || ((!a3stg_faddsubopa[0]) && (|a3stg_ld0_frac[10:0])); fpu_denorm_frac i_a3stg_denorm ( .din1 ({a3stg_ld0_frac[63:11], a3stg_ld0_dnrm_10}), .din2 (a3stg_expdec[53:0]), .din2_din1_denorm (a3stg_denorm), .din2_din1_denorm_inv (a3stg_denorm_inv), .din2_din1_denorma (a3stg_denorma), .din2_din1_denorm_inva (a3stg_denorm_inva) ); /////////////////////////////////////////////////////////////////////////////// // // Add pipe leading 0's counter. // // Add stage 3. // /////////////////////////////////////////////////////////////////////////////// fpu_cnt_lead0_64b i_a3stg_lead0 ( .din (a3stg_ld0_frac[63:0]), .lead0 (a3stg_lead0[5:0]) ); /////////////////////////////////////////////////////////////////////////////// // // Add pipe main adder. // // Add stage 3. // /////////////////////////////////////////////////////////////////////////////// assign a3stg_fracadd[63:0]= (a3stg_frac1[63:0] + a3stg_frac2[63:0] + {63'b0, a3stg_suba}); dffe_s #(64) i_astg_xtra_regs ( .din ({{4{a2stg_shr_cnt_5_inv_in}}, {4{a2stg_shr_cnt_in[5]}}, a2stg_shr_cnt_in[5:3], {5{a2stg_shr_cnt_in[4]}}, {5{a2stg_shr_cnt_in[3]}}, a2stg_shr_cnt_in[5:0], a4stg_round_in, {2{a2stg_shr_cnt_in[2]}}, {2{a2stg_shr_cnt_in[1]}}, {2{a2stg_shr_cnt_in[0]}}, {3{a4stg_shl_cnt_in[6]}}, {3{a4stg_shl_cnt_in[7]}}, {3{a4stg_shl_cnt_in[8]}}, {3{a4stg_shl_cnt_in[9]}}, a4stg_shl_cnt_in[5:0], {3{a2stg_shr_cnt_in[5]}}, a2stg_fracadd_frac2_inv_in, a2stg_fracadd_frac2_inv_shr1_in, a3stg_denorm_inva, a2stg_fsdtoix_nx, a2stg_fsdtoi_nx, 1'b0, a2stg_fracadd_cin_in, {2{a3stg_sub_in}}}), .en (a6stg_step), .clk (clk), .q ({a2stg_shr_cnt_5_inv[3:0], a2stg_shr_cnt_5[3:0], a2stg_shr_cnta[5:3], a2stg_shr_cnt_4[4:0], a2stg_shr_cnt_3[4:0], a2stg_shr_cnt[5:0], a4stg_round, a2stg_shr_cnt_2[1:0], a2stg_shr_cnt_1[1:0], a2stg_shr_cnt_0[1:0], a4stg_shl_cnt_dec54_0[2:0], a4stg_shl_cnt_dec54_1[2:0], a4stg_shl_cnt_dec54_2[2:0], a4stg_shl_cnt_dec54_3[2:0], a4stg_shl_cnt[5:0], a2stg_shr_cnta_5[2:0], a2stg_fracadd_frac2_inv, a2stg_fracadd_frac2_inv_shr1, a4stg_denorm_inv, a3stg_fsdtoix_nx, a3stg_fsdtoi_nx, astg_xtra_regs[3], a2stg_fracadd_cin, a3stg_sub, a3stg_suba}), .se (se), .si (), .so () ); assign a4stg_round_in= a3stg_fracadd[61] || a3stg_fracadd[62] || a3stg_fracadd[63]; assign a3stg_inc_exp_inv= (!a3stg_fracadd[63]); assign a3stg_same_exp_inv= (!(((!a3stg_fracadd[63]) && a3stg_fracadd[62]) || ((!a3stg_fracadd[63]) && a3stg_exp10_0_eq0))); assign a3stg_dec_exp_inv= (!((!a3stg_fracadd[63]) && (!a3stg_fracadd[62]) && a3stg_fracadd[61] && (!a3stg_exp10_0_eq0))); assign a3stg_inc_exp_inva= (!a3stg_fracadd[63]); assign a3stg_fsame_exp_inv= (!(((!a3stg_fracadd[63]) && (!a3stg_fracadd[62]) && a3stg_fracadd[61] && a3stg_exp10_1_eq0 && a3stg_exp_0) || ((!a3stg_fracadd[63]) && a3stg_fracadd[62] && (!a3stg_exp10_0_eq0)))); assign a3stg_fdec_exp_inv= (!((!a3stg_fracadd[63]) && (!a3stg_fracadd[62]) && a3stg_fracadd[61] && (!a3stg_exp10_1_eq0))); assign a4stg_rnd_frac_pre1_in[63:0]= ({64{(a3stg_faddsubopa[1] && a6stg_step && (!a3stg_fdec_exp_inv))}} & {a3stg_fracadd[61:0], 2'b00}); dff_s #(64) i_a4stg_rnd_frac_pre1 ( .din (a4stg_rnd_frac_pre1_in[63:0]), .clk (clk), .q (a4stg_rnd_frac_pre1[63:0]), .se (se), .si (), .so () ); assign a4stg_rnd_frac_pre3_in[63:0]= ({64{(a3stg_faddsubopa[1] && a6stg_step && (!a3stg_fsame_exp_inv))}} & {a3stg_fracadd[62:0], 1'b0}); dff_s #(64) i_a4stg_rnd_frac_pre3 ( .din (a4stg_rnd_frac_pre3_in[63:0]), .clk (clk), .q (a4stg_rnd_frac_pre3[63:0]), .se (se), .si (), .so () ); assign a4stg_rnd_frac_pre2_in[63:0]= ({64{(a3stg_faddsubopa[1] && a6stg_step && (!a3stg_inc_exp_inva))}} & a3stg_fracadd[63:0]) | ({64{((!a4stg_rnd_frac_add_inv) && a6stg_step)}} & a3stg_fracadd[63:0]) | ({64{((!a3stg_fdtos_inv) && a6stg_step)}} & {a3stg_fracadd[62:0], 1'b0}) | ({64{((!a4stg_fixtos_fxtod_inv) && a6stg_step)}} & a4stg_shl[63:0]) | ({64{(!a6stg_step)}} & a4stg_rnd_frac[63:0]); dff_s #(64) i_a4stg_rnd_frac_pre2 ( .din (a4stg_rnd_frac_pre2_in[63:0]), .clk (clk), .q (a4stg_rnd_frac_pre2[63:0]), .se (se), .si (), .so () ); /////////////////////////////////////////////////////////////////////////////// // // Add pipe input to left shift. // // Add stage 3. // /////////////////////////////////////////////////////////////////////////////// assign a4stg_shl_data_in[63:0]= ({64{a3stg_denorm_inva}} & a3stg_ld0_frac[63:0]) | ({64{a3stg_denorma}} & {1'b0, a3stg_ld0_frac[63:1]}); dffe_s #(64) i_a4stg_shl_data ( .din (a4stg_shl_data_in[63:0]), .en (a6stg_step), .clk (clk), .q (a4stg_shl_data[63:0]), .se (se), .si (), .so () ); /////////////////////////////////////////////////////////////////////////////// // // Add pipe rounding adder. // // Add stage 4. // /////////////////////////////////////////////////////////////////////////////// assign a4stg_rnd_frac[63:0]= (a4stg_rnd_frac_pre1[63:0] | a4stg_rnd_frac_pre2[63:0] | a4stg_rnd_frac_pre3[63:0]); assign a4stg_rnd_frac_40= a4stg_rnd_frac[40]; assign a4stg_rnd_frac_39= a4stg_rnd_frac[39]; assign a4stg_rnd_frac_11= a4stg_rnd_frac[11]; assign a4stg_rnd_frac_10= a4stg_rnd_frac[10]; assign a4stg_frac_9_0_nx= (|a4stg_rnd_frac[9:0]); assign a4stg_frac_dbl_nx= a4stg_frac_9_0_nx || a4stg_rnd_frac[10]; assign a4stg_frac_38_0_nx= a4stg_frac_dbl_nx || (|a4stg_rnd_frac[38:11]); assign a4stg_frac_sng_nx= a4stg_frac_38_0_nx || a4stg_rnd_frac[39]; assign a4stg_frac_neq_0= a4stg_frac_sng_nx || (|a4stg_rnd_frac[63:40]); assign a4stg_rndadd_tmp[52:0]= {1'b0, a4stg_rnd_frac[62:11]} + {23'b0, a4stg_rnd_sng, 28'b0, a4stg_rnd_dbl}; assign a4stg_rndadd_cout= a4stg_rndadd_tmp[52]; assign a4stg_rndadd[51:0]= a4stg_rndadd_tmp[51:0]; /////////////////////////////////////////////////////////////////////////////// // // Add pipe left shift. // // Add stage 4. // /////////////////////////////////////////////////////////////////////////////// assign a4stg_shl_data_neq_0= (|a4stg_shl_data[63:0]); assign a4stg_shl_tmp4[63:0] = ({{32{a4stg_shl_cnt_dec54_0[0]}}, {21{a4stg_shl_cnt_dec54_0[1]}}, {11{a4stg_shl_cnt_dec54_0[2]}}} & a4stg_shl_data[63:0]) | ({{32{a4stg_shl_cnt_dec54_1[0]}}, {21{a4stg_shl_cnt_dec54_1[1]}}, {11{a4stg_shl_cnt_dec54_1[2]}}} & {a4stg_shl_data[47:0], 16'h0000}) | ({{32{a4stg_shl_cnt_dec54_2[0]}}, {21{a4stg_shl_cnt_dec54_2[1]}}, {11{a4stg_shl_cnt_dec54_2[2]}}} & {a4stg_shl_data[31:0], 32'h00000000}) | ({{32{a4stg_shl_cnt_dec54_3[0]}}, {21{a4stg_shl_cnt_dec54_3[1]}}, {11{a4stg_shl_cnt_dec54_3[2]}}} & {a4stg_shl_data[15:0], 32'h00000000, 16'h0000}); assign a4stg_shl[63:0] = a4stg_shl_tmp4[63:0] << a4stg_shl_cnt[3:0]; /////////////////////////////////////////////////////////////////////////////// // // Add pipe fraction output. // // Add stage 4. // /////////////////////////////////////////////////////////////////////////////// dffe_s #(58) i_a5stg_rndadd ( .din ({a4stg_rndadd_cout, add_frac_out_rndadd, add_frac_out_rnd_frac, a4stg_in_of, add_frac_out_shl, a4stg_to_0, a4stg_rndadd[51:0]}), .en (a6stg_step), .clk (clk), .q ({add_of_out_cout, a5stg_frac_out_rndadd, a5stg_frac_out_rnd_frac, a5stg_in_of, a5stg_frac_out_shl, a5stg_to_0, a5stg_rndadd[51:0]}), .se (se), .si (), .so () ); dffe_s #(64) i_a5stg_rnd_frac ( .din (a4stg_rnd_frac[63:0]), .en (a6stg_step), .clk (clk), .q (a5stg_rnd_frac[63:0]), .se (se), .si (), .so () ); dffe_s #(64) i_a5stg_shl ( .din (a4stg_shl[63:0]), .en (a6stg_step), .clk (clk), .q (a5stg_shl[63:0]), .se (se), .si (), .so () ); assign add_frac_out[63:0]= ({64{a5stg_frac_out_rndadd}} & {1'b0, a5stg_rndadd[51:0], 11'b0}) | ({64{a5stg_frac_out_rnd_frac}} & a5stg_rnd_frac[63:0]) | ({64{a5stg_in_of}} & {64{a5stg_to_0}}) | ({64{a5stg_frac_out_shl}} & a5stg_shl[63:0]); endmodule
Go to most recent revision | Compare with Previous | Blame | View Log