Re: A85: 16-bit integer Square Root

To: assembly-85@lists.ticalc.org
Subject: Re: A85: 16-bit integer Square Root
From: Humberto Yeverino <humberto@engr.csufresno.edu>
Date: Sat, 14 Mar 1998 22:47:46 -0800 (PST)
Delivered-To: assembly-85-outgoing@towerguard.unix.edu.sollentuna.se
In-Reply-To: <Pine.LNX.3.95.980312094449.17427A-100000@opal.utu.fi>
Reply-To: assembly-85@lists.ticalc.org

On Thu, 12 Mar 1998, Jarno Rajala wrote:

> 
> The GUESS=(GUESS+X/GUESS)/2 algorithm works fine on floating points, but
> is no good for intergers. I used a binary search kind of method. It
> calculates square root of an integer and rounds it down.
> #include "ti-85.h"
> 
> VAR1    = TEXT_MEM
> 
> .org
> .db "SQRT by Jarno Rajala",0
> 
>         ld hl,0
...
> 
> .end

I finished the 16-bit version of my sqrt algorithym.  I'm not sure how
fast Jarno's Sqrt is but I can say that my Sqrt is smaller an instead of
rounding down it rounds to the nearest integer.
Speedwise I tested the worst case and my calc on 8 contrast did 256 worst
case sqrt16s in about .7 seconds.
The worst case by the way is sqrt of 1.

Anyways here it is:
Sqrt16:	
  push de
  push bc
  ld de,0
  ld bc,0
  call CP_HL_DE
  jr z,Sqrt16Done  ;0 is a special case.
  ld de,65280
  ld bc,256
  sbc hl,de
  jr nc,Sqrt16done
  dec bc           ;bc=255
  scf
  rr h
  rr l		   ;shift hl
  call Sqrt16loop
Sqrt16done:
  ld h,b
  ld l,c
  pop bc
  pop de
  ret
Sqrt16loop:
  add hl,bc
  ret c
  dec c
  jp Sqrt16loop  

That's it.
If anyone tested the other algorithym see which is faster.

-Humberto Yeverino Jr.

"Small potatos make the steak look bigger."

http://www.engr.csufresno.edu/~humberto/Home.html
humberto@engr.csufresno.edu

<font size=3>For browsers:<br>
<a href="http://www.engr.csufresno.edu/~humberto/Home.html">
Have a look.</a></font><br>

Follow-Ups:

Re: A85: 16-bit integer Square Root

From: egillespie@juno.com (Erik L Gillespie)

References:

A85: 16-bit integer Square Root

From: Jarno Rajala <jarraj@opal.tnkpub.utu.fi>