This routine was written by The Executioner based on the fast pre-calculated square routine and an algorithm by Milos "baze" Bazelides, with 16 bit arithmetic removed for faster performance.
The initialisation routine, used to build the squares table (same as for the Pre-calculated Square table (my version)).
ld de,1 ld hl,0 ld ix,sqtab sqloop: ld (ix + 0),l inc hx ld (ix + 0),h dec hx add hl,de inc de inc de inc lx jr nz,sqloop
The actual 16 bit square root routine.
Input: DE = The number you want to find the square root of
Output: A = Square root of the number supplied in DE
Destroys: HL
Call: CALL SqrtDE
SqrtDE: ld hl,sqtab + #180 ld a,d cp (hl) jr nz,$ + 6 dec h ld a,e cp (hl) inc h set 6,l jr nc,$ + 4 res 7,l ld a,d cp (hl) jr nz,$ + 6 dec h ld a,e cp (hl) inc h set 5,l jr nc,$ + 4 res 6,l ld a,d cp (hl) jr nz,$ + 6 dec h ld a,e cp (hl) inc h set 4,l jr nc,$ + 4 res 5,l ld a,d cp (hl) jr nz,$ + 6 dec h ld a,e cp (hl) inc h set 3,l jr nc,$ + 4 res 4,l ld a,d cp (hl) jr nz,$ + 6 dec h ld a,e cp (hl) inc h set 2,l jr nc,$ + 4 res 3,l ld a,d cp (hl) jr nz,$ + 6 dec h ld a,e cp (hl) inc h set 1,l jr nc,$ + 4 res 2,l ld a,d cp (hl) jr nz,$ + 6 dec h ld a,e cp (hl) inc h inc l jr nc,$ + 4 res 1,l ld a,d cp (hl) jr nz,$ + 6 dec h ld a,e cp (hl) ld a,l ret nc dec a ret