Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site caip.RUTGERS.EDU
Path: utzoo!watmath!clyde!burl!ulysses!gatech!seismo!caip!french
From: french@caip.RUTGERS.EDU
Newsgroups: net.micro.amiga
Subject: Mandelbrot improvements
Message-ID: <751@caip.RUTGERS.EDU>
Date: Tue, 17-Dec-85 05:36:35 EST
Article-I.D.: caip.751
Posted: Tue Dec 17 05:36:35 1985
Date-Received: Wed, 18-Dec-85 05:46:58 EST
Sender: daemon@caip.RUTGERS.EDU
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 31

From: "french robert%c.mfenet"@LLL-MFE.ARPA

DEC.BANKS@MARLBORO.DEC.COM writes:
>...you could perhaps speed up your inner loop...if you replaced
>the computation with something similar to what we used in our ABasiC
>Mandelbrot program...

Dawn, first let me thank you for taking the time to try out my version
of the program.  Yours is the first response I have gotten.

I have no doubt that your 68000 assembly language code is faster than
my C code, even using the FFP routines.  The only reasons I didn't write
that section in 68000 are: 1) I wanted to get a working version out the
"door" as soon as possible, and 2) I don't speak 68000 assembly language
yet (am learning it, though).

I would look at the code that you included with your AbasiC version, but
unfortunately, due to a mailer error (he he), I was unable to download
it.  If you would send it to me again (at least the 68000 portion),
I would be glad to analyze it and place it in my program.

I have been thinking about writing my own specialized FP routines, since
so much in the Mandelbrot equation can be optimized on a binary level
(like multiplication by two), but haven't gotten around to it yet.  A
friend of mine wrote an integer version of the program (using 3 digit
precision...not terrific, but better than nothing) and got a marked
increase in speed.

         Thanks for your input!
            Robert French
            French#Robert%d@LLL-MFE.ARPA