Re: [Algorithms] Matching Cross platform FP

[Drizzt <dri...@li...>]

---------- Initial Header -----------

>From      : gda...@li...
Date      : Tue, 18 Jul 2006 16:18:08 +0100

> > 5KByte * 50 * 60 =3D 15 megs for a 
> > full minute of replay, that is not much.
> > And we are talking about 300 objects 
> > moving each frame, not that few...
> 
> Ouch ... i assme you are a PC developer?  
> For the platforms i work on 15 meg 
> is a HELL of a lot memory ... Thats about 
> 7 times larger than i'd want for 5 
> minutes!! 

On a platform like the one -I suppose- you are working for, 300 moving ob=
jects in FP precision at 50 frame per second is a dream (again, I suppose=
).


However, do your math...for each moving object you need 6 "values" to put=
 it back into the world (3 for the pos and 3 for the rotation).
You can use 4 byte each "value" to reach a good precision, 2 byte each fo=
r an acceptable precision (but it depends on the world used), or you can =
even use 2 byte for each of the 3 pos values and 1 byte for each of the r=
otation one.


Sum it all, mult for the number of objects that you think can be moving e=
ach frame, add a few hundreds byte for storing commands, plus 2 byte agai=
n for each object moving and you have your total per frame.




(and no, I'm not a PC Developer. At least, I'm not a developer at all whe=
n on this list because if you call me that way, you don't have words in a=
 dictionary to classify all the other real programmers that are here :-) =
)
=0A=0A=0A________________________________________________________________=
___=0ADimagrire in 30 giorni con New Giorno & Notte! Garanzia 100% Soddis=
fatti o Rimborsati! =0Ahttp://click.libero.it/diete=0A