Difference between revisions of "Programming:Fast 16 bit Square Root"

Latest revision as of 03:43, 1 September 2014

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,$ + 5
	dec h
	ld a,e
	cp (hl)
	ld a,l
	ret nc
	dec a
	ret