http://www.bmath.net/bmath/

I have a professional game programer here that says it is not any help to game engines. But it may be useful to someone.

It reminds me of my bit-based image scaling algorithm I made in 68k.
Posted on 2001-08-27 20:33:21 by ThoughtCriminal
They sure look like...havent tested them 100% yet but they sure look like ;)

First you must learn to think for yourself

Second you must experiment to see with your own eyes

Third you must understand that usually "profesional" means doing a job for the money...and that is "a kind of prostitution"

do you belive ? or do you know? that is the question...
Posted on 2001-08-27 20:56:59 by BogdanOntanu
Hehehe...

:grin: Always remember... "wax on", "wax off"... :grin:


Just having fun.. Little more serious tho, Im not much of an expert here, but I do know this sort of thing have been an asset in past 3d engines. As BogdanOntanu sugested, do some clock tests for your self and compare 1 million floating point cosines to 1 million of 'thier' cosines and see what is faster...

I know some engines also use lookup tables for trig op's so perhaps a random algabraic equation would be better...

Anywho, thanx for the link at the very least..

NaN
Posted on 2001-08-27 22:45:23 by NaN
hm, just wanted to comment on that site, but then I remembered that 'new rule' ... too bad ;)
Posted on 2001-08-28 09:59:19 by Tola
Yeah I hate that new rule, whats wrong with a bit of religious preaching.... But staying away from the religious aspect of the site, the thought that people out there believe that crap genuinely scares me.

Anyway if your using OpenGL or DirectX to render you engine then you'll find very few uses for trig left.

In the game I'm writing I have huge maths sections for controling homing missiles, ship, particles, etc yet the amount of trig used is quite small compared to vector maths and the like.

These function would probably only help a 3d engine if you were writing it yourself from scratch.

Also what sort of accuracy do they offer?
Posted on 2001-08-28 12:31:00 by Eóin
Zadkiel, I typed a long reply, but I had to delete it. :tongue:

The numbers look like they're scaled by 10-bits, so the resolution is 1/2^10, and the max value is 2^21 (plus one bit for sign).
Posted on 2001-08-28 12:58:08 by bitRAKE
Apart from a serious case of eyestrain from looking at the site, if you need the capacity, test it and see if it does the job.

I have met enough people who have religious views that have no problems with evolution theory at all and I get the impression that those who try to rebuild empirical science an the basis of phony science are in the same boat as trying to prove the world is square.

As that expression has a very different meaning in my youth, depending on your assumptions, proving the world is square is probably easier. :)

Regards,

hutch@pbq.com.au
Posted on 2001-08-29 08:27:11 by hutch--
been there a while back & my opinion was hrummpf.

couldn't agree with BogdanOntanu more. ('cept i'm a whore not a prostitute because i don't get paid.)

seems to be one of those things that superficially makes sense... like the religious arguments... until you start to really think about it...

Like: Why 2^10 & not 2^16 or 2^32 bitians/twirl? if you're going for machine compatiblity. pi doesn't make any more sense for base 10 (or 1024) than any other number system except for base 3.1415...

test & see (the light)... i could become a believer ;)
Posted on 2001-08-29 11:53:43 by rafe