| |
Kisiel
Account closed
Registered: Jul 2003
Posts: 56 |
Floating Point BASIC Routines
Hello,
I'm looking for basic/not basic programs which are using FP routines (only BASIC routines, clear jmps). I'm going to test my FP routines to compare speed improve. Can you help me? |
|
| |
Krill
Registered: Apr 2002
Posts: 2980 |
Hi,
not sure what exactly you're looking for, but this
Artefacts
uses the ROM routines for some float calculations (basically unpacking lossily-compressed look-up tables using polynomials).
Check math.s in the source folder, featuring the poly unpack loop.
Apart from that, the test app of my loader uses ROM fp routines to calculate the kB/s figure, check
Krill's Loader, repository version 58 (But this version is outdated, will release a new one soonish.) |
| |
Frantic
Registered: Mar 2003
Posts: 1648 |
I did some FP division through calling the BASIC routines about a year ago or two, and found very little info on how to do that when browsing the net. In the end, Krill came to the rescue. In case you find proper info on how to do this, it would be nice to know about it. It would be even nicer if that info ended up on http://codebase64.org
//FTC |
| |
chatGPZ
Registered: Dec 2001
Posts: 11386 |
http://hitmen.c02.at/temp/cc65-floats.tar.gz might give you some hints :) |
| |
Kisiel
Account closed
Registered: Jul 2003
Posts: 56 |
Thanks for all, I really know routines of basic FP. I've waste 2 days to figure out how it's done(with a little help from my friends). I'm looking for programs like Krill sad (I think 4K it's good example...very good job;)). I'm going to check my own routines. Maybe some of You made some huge calculations in FP format in some interesting program so it helps me to compare my routines. |
| |
Marauder/GSS
Account closed
Registered: Jul 2006
Posts: 224 |
I think I used it in Art Bouncer Preview to speed up calculating sines, based on the basic routines of Art Bouncer V2.0...
I may also have some old Turbo Ass code for it... a pity only the preview version found so far... but maybe helps and is what you're looking for.
|
| |
TNT
Account closed
Registered: Oct 2004
Posts: 189 |
There was a Mandelbrot generator in some old Tietokone-magazine back in 1987 or 1988. It used FP routines directly for speed, but funnily enough did not have ML routines for clearing the bitmap or plotting the pixels. You can find the program here. If you remove all the graphic stuff and check only the mandelbrot calculation time then you should have a reasonable benchmark for *some* FP routines. t1=ti before, print ti-t1 after, initial set takes about 3 hours with ROM routines. |
| |
Kisiel
Account closed
Registered: Jul 2003
Posts: 56 |
Yes, TNT it's best test ever :) Graphic routine is worst ever too ;) |
