Code is available on my GitHub account: github.com/wbhart/PCRetroProg...
Пікірлер: 9
@JimLeonard2 жыл бұрын
This is excellent, and great work. Avoiding the "lake" is the big speedup as you mentioned. One of the ways I sped up mine for introJR (which was not as efficient as yours!) was to recognize that I only have 16 colors to paint with, so I limited iterations to 16. One of the most interesting ways I've seen to draw a fractal somehow generates empty outlines for the areas of solid color, then fills them in. If that method isn't sufficiently more calculations, then doing so and using a flood fill might be a speedup too. I regret I forgot where I saw that method used but it's probably in fractint somewhere.
@pcretroprogrammer26562 жыл бұрын
Thanks for the kind comments. Yes, I believe there was some version of Fractint that used the outline/floodfill optimisation. I remember being surprised as a kid that this didn't seem to go much faster.
@t00by00zer8 ай бұрын
@@pcretroprogrammer2656 The boundary is always close to max iteration, which is where the processing time is consumed. And you have to check every pixel for it's escape number.
@IExSet2 жыл бұрын
Dont look on number of comments and likes, just keep to burn hearts with your super content !!!!
@pcretroprogrammer26562 жыл бұрын
Thanks! I'm not worried about likes and comments here (though those do help grow the channel). I'm much more interested in getting good information out there that is clear and correct!
2 жыл бұрын
I can be wrong, but at 21:06, wouldn't it be easier to just remove INC DI and DEC DI, and putting an INC DI though before JMP done_iter instead? That way STOSB increments DI normally, and at the end of the iter_loop, before JMP done_iter, there is the new single INC DI which does the job.
@billhart38142 жыл бұрын
Yes, that seems like it should work just fine. It should give a small speedup, which is nice!
@nasigoreng5532 жыл бұрын
Really cool and yeh quite quick for an 8088 would having a 8087 make it perform better?
@pcretroprogrammer26562 жыл бұрын
I don't think it would be quicker, but I haven't tried a floating point version, so I don't know for sure.