Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxt!houxm!whuxl!whuxlm!akgua!gatech!seismo!ut-sally!pyramid!amiga!dale
From: dale@amiga.UUCP
Newsgroups: net.micro.amiga
Subject: Re: AmigaFest--San Diego
Message-ID: <833@amiga.amiga.UUCP>
Date: Sun, 9-Mar-86 00:18:32 EST
Article-I.D.: amiga.833
Posted: Sun Mar  9 00:18:32 1986
Date-Received: Tue, 11-Mar-86 01:14:30 EST
References: <1361@caip.RUTGERS.EDU> <122@hp-sdd.UUCP> <367@mips.UUCP> <792@amiga.amiga.UUCP> <128@hp-sdd.UUCP>
Reply-To: dale@tooter.UUCP (Dale Luck)
Organization: Commodore-Amiga Inc., 983 University Ave #D, Los Gatos CA 95030
Lines: 38

In article <128@hp-sdd.UUCP> nick@hp-sdd.UUCP (Nick Flor) writes:
>Gee, you should have told me that sooner.  The salesman was really
>excited that the CSA board was "speeding up" the bouncing ball
>demo.

Amiga engineering does not claim any responsibility for uninformed
salesmen for anybodies machine. I do not want to state the obvious
comparisons engineers make between themselves and salesmen. I hope
it was just an uninformed but enthusiastic amigafile.

>Also,  the blitter hardware draws the lines, not the 68000
>ergo, there shouldn't be a significant speed increase.  
>(Go to the head of the class Nicko for stating the obvious)
>
>So, what's the use of that board in the first place. 
>

The blitter does not magically realize that an application wants a
line drawn.  It needs a little coaxing. The graphics takes about
200usec setting everything up.  Because we actually run the blitter
async to the graphics routines we achieve some parrallism and can
approach 4000-5000 vectors per second as long as the vectors are
less the 200 pixel vectors. Greater than that and limiting factor
is the blitter.  Well when you have 68020 running at 14mhz this
68k overhead reduces to about (guessing here)100usec's and for
vectors in the 100-200 pixel range we see an improvement in speed
that is a function of the length of the vector. 100 pixel vectors
may be almost twice as fast. For vectors <100 pixels the time is
again dominated by overhead and setting up the blitter. But this
time the overhead is halved and the result should be a near doubling
in speed.
The above discussion is true for all graphics operations. The throughput
of the graphics is determined not by the blitter alone but also by
the efficiency and speed of the code that drives the blitter and
the program itself.

Dale Luck
Amiga