I've found the Mandelbrot set always interesting since it became widely known in the late 1980s. This is clearly shown by the fact that I wrote three programs on three different systems to create and view Mandelbrot fractals.
I've found the Mandelbrot set always interesting since it became
widely known in the late 1980s. This is clearly shown by the fact
that I wrote three programs on three different systems to create
and view Mandelbrot fractals.
The first one (early 1990) was written in TMS9900 assembly and ran on the TI99/4A home computer under my home-grown OS. Its screen resolution was a phantastic 256x192 monochrome pixels, and its 32 bit floating point emulation (can't remember whether I switched to fixed point later) needed an hour to calculate even a simple image. It could load and save images, but not print them (I didn't have a printer anyway). It even had an algorithm to put a copyright notice in a part of the image where it would not disturb anything!
The second one was also my first C program. It is written for DOS only. But later it had rudimentary support for the Linux console. It has fast graphics output routines which unfortunately are specific for each graphic card. Therefore it only supports ET3000 (and probably ET4000), TVGA9000 and - recently - some S3 cards. This program features fixed point integer routines in 32 and 64 bit (useless on 486ers and up, but very convenient for 386s), automatic checkpoints, 16 colors, and screen resolutions of up to 1024x1024 or so. The image must always be a square. And the 16 colors are predefined to shades of blue - looks quite good but cannot be changed. Interactive zooming is available. Images can be saved in a format invented by myself and exported in PCX and BMP format.
This program still exists and works, but is probably not interesting enough for anyone. It could be useful as a performance test for FPUs, however.
2000 iterations
The third one was written in summer 1995 and runs under X11. It can load and save files portably in ppm format and supports all color depths up to 24 bit.
However, there is not much functionality. Except that it uses separate processes for the GUI and the calculations and offers a zoom function which is available while the image is calculated.
A few years ago I changed the program so that it can calculate an image on several computers in parallel. It uses the MPI implementation LAM for this. I have also renamed the program to netchaos. The current version is available here as a Tar archive.