Added conditional OTFC swap for simplified int postprocessing

pipelined/config/shared/wally-shared.vh

@@ -127,6 +127,7 @@
 `define DURLEN ($clog2(`FPDUR+1))
 `define QLEN (`FPDUR*`LOGR*`DIVCOPIES)
 `define DIVb (`QLEN-1)
+`define DIVa (`DIVb+4-`XLEN)
 `define DIVBLEN ($clog2(`DIVb+1)-1)
 
 

pipelined/src/fpu/fdivsqrt/fdivsqrt.sv

@@ -65,10 +65,11 @@ module fdivsqrt(
   logic WZero;
   logic SpecialCaseM;
   logic [`DIVBLEN:0] n, p, m;
+  logic OTFCSwap;
 
   fdivsqrtpreproc fdivsqrtpreproc(
     .clk, .DivStartE, .Xm(XmE), .QeM, .Xe(XeE), .Fmt(FmtE), .Ye(YeE), 
-    .Sqrt(SqrtE), .Ym(YmE), .XZero(XZeroE), .X, .Dpreproc, .n, .p, .m,
+    .Sqrt(SqrtE), .Ym(YmE), .XZero(XZeroE), .X, .Dpreproc, .n, .p, .m, .OTFCSwap,
     .ForwardedSrcAE, .ForwardedSrcBE, .Funct3E, .Funct3M, .MDUE, .W64E);
   fdivsqrtfsm fdivsqrtfsm(
     .clk, .reset, .FmtE, .XsE, .SqrtE, 
@@ -78,7 +79,7 @@ module fdivsqrt(
   fdivsqrtiter fdivsqrtiter(
     .clk, .Firstun, .D, .FirstU, .FirstUM, .FirstC, .SqrtE, .SqrtM, 
     .X,.Dpreproc, .FirstWS(WS), .FirstWC(WC),
-    .DivStartE, .Xe(XeE), .Ye(YeE), .XZeroE, .YZeroE,
+    .DivStartE, .Xe(XeE), .Ye(YeE), .XZeroE, .YZeroE, .OTFCSwap,
     .DivBusy);
   fdivsqrtpostproc fdivsqrtpostproc(
     .WS, .WC, .D, .FirstU, .FirstUM, .FirstC, .Firstun, 

pipelined/src/fpu/fdivsqrt/fdivsqrtiter.sv

@@ -38,6 +38,7 @@ module fdivsqrtiter(
   input  logic XZeroE, YZeroE, 
   input  logic SqrtE,
   input  logic SqrtM,
+  input  logic OTFCSwap,
   input  logic [`DIVb+3:0] X,
   input  logic [`DIVN-2:0] Dpreproc,
   output logic [`DIVN-2:0]  D, // U0.N-1

pipelined/src/fpu/fdivsqrt/fdivsqrtpreproc.sv

@@ -42,6 +42,7 @@ module fdivsqrtpreproc (
 	input  logic [2:0] 	Funct3E, Funct3M,
 	input  logic MDUE, W64E,
   output logic [`DIVBLEN:0] n, p, m,
+  output logic OTFCSwap,
   output logic [`NE+1:0] QeM,
   output logic [`DIVb+3:0] X,
   output logic [`DIVN-2:0] Dpreproc
@@ -56,7 +57,7 @@ module fdivsqrtpreproc (
   // Intdiv signals
   logic  [`DIVb-1:0] ZeroBufX, ZeroBufY;
   logic  [`XLEN-1:0] PosA, PosB;
-  logic  As, Bs;
+  logic  As, Bs, OTFCSwapTemp;
   logic  [`XLEN-1:0] A64, B64;
   logic  [`DIVBLEN:0] ZeroDiff, IntBits, RightShiftX;
   logic  [`DIVBLEN:0] pPlusr, pPrCeil;
@@ -70,6 +71,8 @@ module fdivsqrtpreproc (
   assign Bs = ForwardedSrcBE[`XLEN-1] & Funct3E[0];
   assign A64 = W64E ? {{(`XLEN-32){As}}, ForwardedSrcAE[31:0]} : ForwardedSrcAE;
   assign B64 = W64E ? {{(`XLEN-32){Bs}}, ForwardedSrcBE[31:0]} : ForwardedSrcBE;
+
+  assign OTFCSwapTemp = (As ^ Bs) & MDUE;
   
   assign PosA = As ? -A64 : A64;
   assign PosB = Bs ? -B64 : B64;
@@ -111,6 +114,7 @@ module fdivsqrtpreproc (
   // r = 1 or 2
   // DIVRESLEN/(r*`DIVCOPIES)
   flopen #(`NE+2) expflop(clk, DivStartE, Qe, QeM);
+  flopen #(1) swapflop(clk, DivStartE, OTFCSwapTemp, OTFCSwap);
   expcalc expcalc(.Fmt, .Xe, .Ye, .Sqrt, .XZero, .L, .m, .Qe);
 
 endmodule

pipelined/src/fpu/fdivsqrt/fdivsqrtstage2.sv

@@ -61,7 +61,7 @@ module fdivsqrtstage2 (
 	// 0001 = -2
   fdivsqrtqsel2 qsel2(WS[`DIVb+3:`DIVb], WC[`DIVb+3:`DIVb], up, uz, un);
 
-  // Sqrt F generatin
+  // Sqrt F generation
   fdivsqrtfgen2 fgen2(.up, .uz, .C(CNext), .U, .UM, .F);
 
   // Divisor multiple