[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference hydra::amiga_v1

Title:AMIGA NOTES
Notice:Join us in the *NEW* conference - HYDRA::AMIGA_V2
Moderator:HYDRA::MOORE
Created:Sat Apr 26 1986
Last Modified:Wed Feb 05 1992
Last Successful Update:Fri Jun 06 1997
Number of topics:5378
Total number of notes:38326

3849.0. "polygon fill tests" by WJG::GUINEAU () Sun Jun 17 1990 12:41


wjg::amiga:poly_fill.lzh

has some code that shows the difference in polygon filling between a 68000
and the amiga's blitter.


This was posted to Usenet to settle a (yet another :-) dispute.

john
T.RTitleUserPersonal
Name
DateLines
3849.1Quick test programWHAMMY::SPODARYKScaring the pedestrians...Wed Jun 20 1990 03:4266
    I put together a quick test program, then realized that this has
    probably already been done.  Is such a beast available on Usenet?
    
    Anyway, I have uploaded an archive that contains a test program.  The
    program does a basic comparison between the Blitter and the CPU for
    rendering simple polygons (triangles, pick the number).  Source 
    included.
    
    If anyone has any ideas for getting more conclusive results, let me
    know.  The archive is in WJG::AMIGA:POLY_TEST.ZOO.
    
    Steve
    
    ---- comments from the source ----
    
/*
 *  poly_test.c
 * 
 *  Function:	This program is used to compare several Intuition graphics
 *		functions (using the Amiga custom chips) against graphics
 *		routines that only use the CPU (M680x0).
 *
 *		This probably isn't a completely fair test, but it's probably
 *		fairly reasonable.  The amount of time spent setting up
 *		the drawing calls should be small compared to the time 
 *		actually spent doing the rendering.  Typically we're only 
 *		doing about 8 adds and a couple of compares for each 
 *		triangle.  Test for yourself.
 *
 *		My results when rendering 100,000 triangles (standard size):   
 *
 *				7MHz 68010	14MHz 68020
 *		-------------------------------------------
 *		Blitter		295 seconds	143 seconds
 *		CPU		300 seconds	103 seconds
 *
 *		When using triangles of the larger variety the results
 *		were comparable.  When the triangles were not being moved,
 *		the results were almost the same.
 *
 *		All this test probably proves is that I'm great at writing
 *		inconclusive tests.  If anyone has any improvements to 
 * 		make please let me know about them.  It might be nice to
 *		add other graphics operations, but it's getting late.
 *
 *	Note:   The custom routines are written for a 320x200x4 screen,
 *		so changing the defaults in this program might cause some
 *		problems.  Not flexible, but don't blame me.
 *
 *		These routines are also optimized for "regular" polygons.
 *		The Intuition routines are more generalized, and that would
 *		add some overhead.  The Intuition routine might also require
 *		more calling overhead.  
 *
 *
 *  Input:	One CLI argument, the # of triangles to render
 *		the default is 5000 
 *
 *  Authors:	Stephen Spodaryk	19-Jun-1990
 *		spodaryk@smaug.dec.com
 *
 *		Custom graphics routines (c) 1990 by John Schultz
 *
 *  Modification History:
 *
 */