Path: chuka.playstation.co.uk!scea!greg_labrec@interactive.sony.com
From: "Wayne K. Werner" <wkwerner@con2.com>
Newsgroups: scea.yaroze.programming.3d_graphics
Subject: Re: How fast is fast?
Date: 6 Sep 1997 03:41:05 GMT
Organization: WermX Software
Lines: 36
Message-ID: <01bcba76$82a3a040$b6bf43ce@wkwerner>
References: <33B09E84.191E@charlie.cns.iit.edu> <5otr2k$fv51@chuka.playstation.co.uk> <33BFCA2D.1C4F@charlie.cns.iit.edu>
NNTP-Posting-Host: port82.con2.com
X-Newsreader: Microsoft Internet News 4.70.1161

Ed Federmeyer <fedeedw@charlie.cns.iit.edu> wrote:

> /* Don't try this at home, kids!!!  :-) */
> for ( angle=0 ; angle<360 ; angle++ ) {
> 	sin_table[i] = sin( (angle * PI) / 180 );
> 	cos_table[i] = cos( (angle * PI) / 180 );
                ^ shouldn't this be angle?
> }
>

This might work quite a bit faster (eliminating the fp * and especially the
/):

// assuming sin_table and cos_table are global
void make_trig_tables(void){

	float angle = 0;
	int degree;

	for(degree = 0; degree < 360; degree++){

		sin_table[degree] = sin(angle);
		cos_table[degree] = cos(angle);

		angle += (PI / 180);
	}
}

Since PI and 180 are constants, (PI / 180) is evaluated at compile time. 
The mult and divide have been replaced with a simple add.  It doesn't
address the real bottleneck, of course, the trig calcs.  If you try it,
please let me know how the results compare with your original calc!
-- 
Wayne K. Werner
wkwerner@con2.com