cvw/pipelined/src/fpu/fpudivsqrtrecurcore.sv
2022-01-07 12:58:40 +00:00

106 lines
3.9 KiB
Systemverilog

///////////////////////////////////////////
//
// Written: David Harris
// Modified: 11 September 2021
//
// Purpose: Recurrence-based SRT Division and Square Root
//
// A component of the Wally configurable RISC-V project.
//
// Copyright (C) 2021 Harvey Mudd College & Oklahoma State University
//
// MIT LICENSE
// Permission is hereby granted, free of charge, to any person obtaining a copy of this
// software and associated documentation files (the "Software"), to deal in the Software
// without restriction, including without limitation the rights to use, copy, modify, merge,
// publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons
// to whom the Software is furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in all copies or
// substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,
// INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR
// PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
// BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
// TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE
// OR OTHER DEALINGS IN THE SOFTWARE.
////////////////////////////////////////////////////////////////////////////////////////////////
`include "wally-config.vh"
// Bit counts:
// Inputs are originally normalized floating point numbers with NF fractional bits and a leading 1 integer bit
// x is right shifted by up to 2 to be in the range of 1/4 <= x < 1/2 for divide, 1/4 <= x < 1 for sqrt
// Hence, x now has NF+2 fractional bits and 0 integer bits
// d is right shifted by 1 to be in the range of 1/2 <= d < 1. It thus has NF+1 fractional bits and 0 integer bits
// q is eventually in the range of 1/4 < q < 1 and hence needs NF+2 bits to keep NF bits when normalized, plus some*** more bits for rounding
// The partial
/*
module fpudivsqrtrecurcore (
input logic clk,
input logic reset,
input logic start, // start a computation
input logic busy, // computation running
input logic fmt, // precision 1 = double 0 = single
input logic [`NF+1:0] x, // in range 1/4 <= x < 1/2 for divide, 1/4 <=x < 1 for sqrt
input logic [`NF+1:0] din, // in range 1/2 <= d < 1 for divide
input logic FDiv, FSqrt, // *** not yet used
output logic [`FLEN-1:0] FDivSqrtRecurRes // result
);
assign FDivSqrtRecurRes = 0;
logic [***] d, ws, wsout, wsnext, wc, wcout, wcnext;
logic [1:0] q; // 00 = 0, 01 = 1, 10 = -1
// Radix-2 SRT Division
// registers for divisor and partial remainder
flopen #(NF+1) dreg(clk, start, din, d);
mux2 #(NF+1) wsmux(wsout, x, start, wsnext);
flopen #(NF+1) wsreg(clk, busy, wsnext, ws);
mux2 #(NF+1) wcmux(wcout, 0, start, wcnext);
flopen #(NF+1) wcreg(clk, busy, wcnext, wc);
// quotient selection
qsel qsel(ws[***4bits], wc[***], q);
// partial remainder update
always_comb begin // select -d * q to add to partial remainder
if (q[1]) dq = d;
else if (q[0]) dq = ~d;
else dq = 0;
end
csa #(***) csa(ws, wc, dq, q[1], wsout, wcout);
endmodule
*/
/*
module csa #(parameter N=4) (
input logic [N-1:0] sin, cin, ain,
input logic carry,
output logic [N-1:0] sum, cout
);
logic [N-1:0] c;
assign c = {cin[N-2:0], carry}; // shift carries left and inject optional 1 into lsb
assign sum = sin ^ ain ^ c;
assign cout = sin & ain | sin & c | ain & c;
endmodule
*/
module qsel( // radix 2 SRT division quotient selection
input logic [3:0] wc, ws,
output logic [1:0] q
);
endmodule