URL
https://opencores.org/ocsvn/thor/thor/trunk
Subversion Repositories thor
[/] [thor/] [trunk/] [rtl/] [verilog/] [Thor_execute_combo.v] - Rev 48
Go to most recent revision | Compare with Previous | Blame | View Log
// ============================================================================ // __ // \\__/ o\ (C) 2013-2016 Robert Finch, Stratford // \ __ / All rights reserved. // \/_// robfinch<remove>@finitron.ca // || // // This source file is free software: you can redistribute it and/or modify // it under the terms of the GNU Lesser General Public License as published // by the Free Software Foundation, either version 3 of the License, or // (at your option) any later version. // // This source file 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 program. If not, see <http://www.gnu.org/licenses/>. // // // Thor SuperScalar // Execute combinational logic // // ============================================================================ // wire [DBW-1:0] alu0_out, alu1_out; wire alu0_done,alu1_done; wire alu0_idle,alu1_idle; wire alu0_divByZero, alu1_divByZero; wire alu0_abort,alu1_abort; Thor_alu #(.DBW(DBW),.BIG(1)) ualu0 ( .corenum(corenum), .rst(rst_i), .clk(clk), .alu_ld(alu0_ld), .alu_abort(alu0_abort), .alu_op(alu0_op), .alu_fn(alu0_fn), .alu_argA(alu0_argA), .alu_argB(alu0_argB), .alu_argC(alu0_argC), .alu_argI(alu0_argI), .alu_pc(alu0_pc), .insnsz(alu0_insnsz), .o(alu0_out), .alu_done(alu0_done), .alu_idle(alu0_idle), .alu_divByZero(alu0_divByZero) ); Thor_alu #(.DBW(DBW),.BIG(ALU1BIG)) ualu1 ( .corenum(corenum), .rst(rst_i), .clk(clk), .alu_ld(alu1_ld), .alu_abort(alu1_abort), .alu_op(alu1_op), .alu_fn(alu1_fn), .alu_argA(alu1_argA), .alu_argB(alu1_argB), .alu_argC(alu1_argC), .alu_argI(alu1_argI), .alu_pc(alu1_pc), .insnsz(alu1_insnsz), .o(alu1_out), .alu_done(alu1_done), .alu_idle(alu1_idle), .alu_divByZero(alu1_divByZero) ); function fnPredicate; input [3:0] pr; input [3:0] cond; case(cond) PF: fnPredicate = 1'b0; PT: fnPredicate = 1'b1; PEQ: fnPredicate = pr[0]; PNE: fnPredicate = !pr[0]; PLE: fnPredicate = pr[0]|pr[1]; PGT: fnPredicate = !(pr[0]|pr[1]); PLT: fnPredicate = pr[1]; PGE: fnPredicate = !pr[1]; PLEU: fnPredicate = pr[0]|pr[2]; PGTU: fnPredicate = !(pr[0]|pr[2]); PLTU: fnPredicate = pr[2]; PGEU: fnPredicate = !pr[2]; default: fnPredicate = 1'b1; endcase endfunction wire alu0_cmtw = fnPredicate(alu0_pred, alu0_cond); wire alu1_cmtw = fnPredicate(alu1_pred, alu1_cond); always @* begin alu0_cmt <= alu0_cmtw; alu1_cmt <= alu1_cmtw; alu0_bus <= alu0_cmtw ? alu0_out : alu0_argT; alu1_bus <= alu1_cmtw ? alu1_out : alu1_argT; alu0_v <= alu0_dataready; alu1_v <= alu1_dataready; alu0_id <= alu0_sourceid; alu1_id <= alu1_sourceid; end assign alu0_abort = !alu0_cmt; assign alu1_abort = !alu1_cmt; // Special flag bit is used for INT and SYS instructions in order to turn off // segmentation while the vector jump is taking place. always @(alu0_op or alu0_fn or alu0_argA or alu0_argI or alu0_insnsz or alu0_pc or alu0_bt) case(alu0_op) `JSR,`JSRS,`JSRZ,`RTD,`RTE,`RTI,`RTS2: alu0_misspc <= alu0_argA + alu0_argI; `LOOP,`SYNC: alu0_misspc <= alu0_pc + alu0_insnsz; `RTS: alu0_misspc <= alu0_argA + alu0_fn[3:0]; `SYS,`INT,`RTF,`JSF: alu0_misspc <= {1'b1,alu0_argA + alu0_argI}; default: alu0_misspc <= (alu0_bt ? alu0_pc + alu0_insnsz : alu0_pc + alu0_insnsz + alu0_argI); endcase always @(alu1_op or alu1_fn or alu1_argA or alu1_argI or alu1_insnsz or alu1_pc or alu1_bt) case(alu1_op) `JSR,`JSRS,`JSRZ,`RTD,`RTE,`RTI,`RTS2: alu1_misspc <= alu1_argA + alu1_argI; `LOOP,`SYNC: alu1_misspc <= alu1_pc + alu1_insnsz; `RTS: alu1_misspc <= alu1_argA + alu1_fn[3:0]; `SYS,`INT,`RTF,`JSF: alu1_misspc <= {1'b1,alu1_argA + alu1_argI}; default: alu1_misspc <= (alu1_bt ? alu1_pc + alu1_insnsz : alu1_pc + alu1_insnsz + alu1_argI); endcase always @(dram0_fn or dram0_misspc or dram_bus) case (dram0_fn[1:0]) 2'd1: jmpi_misspc <= {dram0_misspc[DBW-1:16],dram_bus[15:0]}; 2'd2: jmpi_misspc <= (DBW==32) ? dram_bus[31:0] : {dram0_misspc[63:32],dram_bus[31:0]}; 2'd3: jmpi_misspc <= dram_bus[DBW-1:0]; default: jmpi_misspc <= 32'h00000FA0; // unimplemented instruction vector endcase wire dbze0 = alu0_op==`DIVI || alu0_op==`MODI || (alu0_op==`RR && (alu0_fn==`DIV || alu0_fn==`MOD)); wire dbze1 = alu1_op==`DIVI || alu1_op==`MODI || (alu1_op==`RR && (alu1_fn==`DIV || alu1_fn==`MOD)); assign alu0_exc = (fnIsKMOnly(alu0_op) && !km && alu0_cmt) ? `EX_PRIV : (alu0_done && dbze0 && alu0_divByZero && alu0_cmt) ? `EX_DBZ : ((alu0_op==`CHKXI||(alu0_op==`RR && alu0_fn==`CHKX)) && !alu0_out && alu0_cmt) ? `EX_CHK : (((alu0_op==`MTSPR && alu0_fn==0)|| alu0_op==`LDIS) && iqentry_tgt[alu0_id[2:0]][7:3]==5'h6) ? {6'h20,iqentry_tgt[alu0_id[2:0]][2:0]} : `EX_NONE; assign alu1_exc = (fnIsKMOnly(alu1_op) && !km && alu1_cmt) ? `EX_PRIV : (alu1_done && dbze1 && alu1_divByZero && alu1_cmt) ? `EX_DBZ : ((alu1_op==`CHKXI ||(alu1_op==`RR && alu1_fn==`CHKX)) && !alu1_out && alu1_cmt) ? `EX_CHK : (((alu1_op==`MTSPR && alu1_fn==0)|| alu1_op==`LDIS) && iqentry_tgt[alu1_id[2:0]][7:3]==5'h6) ? {6'h20,iqentry_tgt[alu1_id[2:0]][2:0]} : `EX_NONE; assign alu0_branchmiss = alu0_dataready && ((fnIsBranch(alu0_op)) ? ((alu0_cmt && !alu0_bt) || (!alu0_cmt && alu0_bt)) : !alu0_cmt ? (alu0_op==`LOOP) : (alu0_cmt && (alu0_op==`SYNC || alu0_op == `JSR || alu0_op == `JSRS || alu0_op == `JSRZ || alu0_op==`SYS || alu0_op==`INT || alu0_op==`RTF || alu0_op==`JSF || alu0_op==`RTS || alu0_op==`RTS2 || alu0_op==`RTD || alu0_op == `RTE || alu0_op==`RTI || ((alu0_op==`LOOP) && (alu0_argA == 64'd0))))); assign alu1_branchmiss = alu1_dataready && ((fnIsBranch(alu1_op)) ? ((alu1_cmt && !alu1_bt) || (!alu1_cmt && alu1_bt)) : !alu1_cmt ? (alu1_op==`LOOP) : (alu1_cmt && (alu1_op==`SYNC || alu1_op == `JSR || alu1_op == `JSRS || alu1_op == `JSRZ || alu1_op==`SYS || alu1_op==`INT || alu1_op==`RTF || alu1_op==`JSF || alu1_op==`RTS || alu1_op==`RTS2 || alu1_op==`RTD || alu1_op == `RTE || alu1_op==`RTI || ((alu1_op==`LOOP) && (alu1_argA == 64'd0))))); assign branchmiss = (alu0_branchmiss | alu1_branchmiss | mem_stringmiss | jmpi_miss), misspc = (jmpi_miss ? jmpi_misspc : mem_stringmiss ? dram0_misspc : alu0_branchmiss ? alu0_misspc : alu1_misspc), missid = (jmpi_miss ? dram0_id : mem_stringmiss ? dram0_id : alu0_branchmiss ? alu0_sourceid : alu1_sourceid); `ifdef FLOATING_POINT wire fp0_exception; fpUnit ufp0 ( .rst(rst_i), .clk(clk), .ce(1'b1), .op(fp0_op), .fn(fp0_fn), .ld(fp0_ld), .a(fp0_argA), .b(fp0_argB), .o(fp0_bus), .exception(fp0_exception) ); reg [7:0] cnt; always @(posedge clk) if (rst_i) cnt <= 8'h00; else begin if (fp0_ld) cnt <= 8'h00; else begin if (cnt < 8'hff) cnt <= cnt + 8'd1; end end always @* begin case(fp0_op) `FLOAT: case(fp0_fn) `FCMP,`FCMPS: fp0_done = 1'b1; // These ops are done right away `FADD,`FSUB,`FMUL,`FADDS,`FSUBS,`FMULS: fp0_done = cnt > 8'd4; `FDIV: fp0_done = cnt > 8'h70; `FDIVS: fp0_done = cnt > 8'h37; default: fp0_done = 1'b1; endcase `SINGLE_R: case(fp0_fn) `FNEGS,`FABSS,`FSIGNS,`FMOVS, `FNABSS,`FMANS: fp0_done = 1'b1; // These ops are done right away `FTOIS,`ITOFS: fp0_done = cnt > 8'd1; default: fp0_done = 1'b1; endcase `DOUBLE_R: case(fp0_fn) `FMOV,`FNEG,`FABS,`FNABS,`FSIGN,`FMAN: fp0_done = 1'b1; // These ops are done right away `FTOI,`ITOF: fp0_done = cnt > 8'd1; default: fp0_done = 1'b1; endcase default: fp0_done = 1'b1; endcase end assign fp0_cmt = fnPredicate(fp0_pred, fp0_cond); assign fp0_exc = fp0_exception ? `EX_FP : 9'd0; assign fp0_v = fp0_dataready; assign fp0_id = fp0_sourceid; `endif
Go to most recent revision | Compare with Previous | Blame | View Log