#Dead?
8 messages in this thread
This section had something Very Scary happen in it about two months ago, and
ever since then it has been haunted. A real house can get haunted if the
corpse of someone popular is found hanging in the rec room. It's still a fine
house, the TV still works, but people get more and more reluctant to visit
because they keep flashing back on ghastly images.
I'd be sorry to see this section vanish, because I own both the TDI and
Benchmark Modulas and neither has 'clicked' yet. I keep hoping to read what
others are doing 'live' rather than polished examples, believing that I'll
eventually go 'aha!' and be able to start creating.
Hi Tom! Nice to see you here. What are you doing with Modula-2? Have you seen
my CHAOS program? It's written in Modula-2. Some of us here in Brooklyn are
working on a new Mandelbrot program (using Benchmark). We should be done in
couple of weeks. It will use a new recursive (!) algorithm and I hope it will
run faster than any other Mandlebrot program out there – this would be only at
iterations set at a high level as this is where you get the great pictures.
Steve F. is working on a new game (also in M2) and he should be done soon
(right Steve?). What sort of things would you like to see……
…take care ….Richie
Rich,
Good luck on the Mandelbrot program… I might warn you though, that you will
really have to do some tight coding to make it faster than MandFXP, which is a
real barn burner.
Regards, Larry.
Richie's goal in writing this Mandelbrot program is to prove that you can
achieve a greater computational power by using better algorithms, not resorting
to brute force. MandFXP is a brute force approach; it is coded in assembler (I
think) and uses fixed point math. Richie's program is in M2 and uses the
standard FFP routines. It is indeed much faster than a conventional Mandelbrot
program, but I have my doubts whether it can match the brute force method in
this case. Now, I would LOVE to see Richie's algorithm used in the MandFXP
program; the thing should FLY!
(to set the record straight, the algorithm was originally conceived by
a coworker of ours, Peter Trei).
– Steve –
Steve,
I don't consider MandFXP to be in any way 'brute force'. The algorithms used
are carefully thought out and a great many insights have been applied to the
problem. I would be willing to bet that any algorithm using standard
floating point (FFP or IEEE) would not be faster than the integer math used in
MandFXP. Just because something is coded in assembler, or because the floating
point functions have been replaced by faster and more accurate equivalents does
not make it 'brute force'.
Regards, Larry.
Larry,
I don't think that Steve was trying to give that impression. I would,
however, like to see his algorithm.
Shhh!!! Richie, don't give my surprizes away!!
Have Fun in DECUS, and bring me a T-shirt from DisneyLand (or is it
DisneyWorld?? I never remember)
– Steve –
Richard,
Recursive? algorithm?? Could you elaborate please?