Added floating point conversion functions for 128 quads

m3-libs/m3core/src/float/Common/FloatUtils.i3

@@ -0,0 +1,62 @@
+(* Copyright (C) 2016 Peter McKinna. All rights reserved. *)
+(* See file COPYRIGHT-BSD for details. *)
+
+INTERFACE FloatUtils;
+
+IMPORT Word;
+IMPORT DragonInt AS DI;
+
+CONST 
+  Bias = 1023;
+
+TYPE
+  Int32 = [-16_7FFFFFFF-1 .. 16_7FFFFFFF];
+
+(* return the lowest 32 bit word of a longreal *)  
+PROCEDURE Word0(r : LONGREAL) : Int32;
+
+(* set the lowest 32 bit word of a longreal *)
+PROCEDURE Word0Set(r : LONGREAL; r0 : Int32) : LONGREAL;
+
+(* checks if all bits are set in a bigint *)
+PROCEDURE allOn(s : DI.Session; b : DI.T; n : INTEGER) : BOOLEAN;
+
+(* checks if any bits are set in a bigint *)
+PROCEDURE anyOn(s : DI.Session; b : DI.T; n : INTEGER) : BOOLEAN;
+
+(* set all bits of a bigint to 1 *)
+PROCEDURE setOnes(s : DI.Session; b : DI.T; n : INTEGER) : DI.T;
+
+(* removes the lowest zero bits ie shifts all the zeros right
+and returns the number shifted and the result of the shift *)
+PROCEDURE lo0bits(VAR y : Word.T) : INTEGER;
+
+(* removes the highest zero bits ie shifts all the zeros left
+and returns the number shifted and the result of the shift *)
+PROCEDURE hi0bits(x : Word.T) : INTEGER;
+
+(* returns the mantissa bits *)
+PROCEDURE mantbits(d : LONGREAL) : INTEGER;
+
+(* bigint diff ensuring a > b swapping if needed *)
+PROCEDURE diff(s : DI.Session; a,b : DI.T; VAR dsign : [-1..1]) : DI.T;
+
+(* multiply a by the kth power of 5 *)
+PROCEDURE pow5mult(s : DI.Session; a : DI.T; k : INTEGER) : DI.T;
+
+(* returns a longreal as a bigint *)
+PROCEDURE d2b(s : DI.Session; l : LONGREAL; VAR exp,bits : INTEGER) : DI.T;
+
+(* returns a string as a bigint *)
+PROCEDURE s2b(s : DI.Session; READONLY a : ARRAY OF CHAR; nd0,nd : INTEGER; y9 : INTEGER) : DI.T;
+
+(* returns ratio of a to b *)
+PROCEDURE ratio(s : DI.Session; a,b : DI.T) : LONGREAL;
+
+(* returns number of trailing zeros of a bigint *)
+PROCEDURE trailz(s : DI.Session; b : DI.T) : INTEGER;
+
+(* return the sign-extended bottom 32 bits of 'x' *)
+PROCEDURE Fix32 (x: Word.T): Int32;
+
+END FloatUtils.