[ODE] Syncing ODE over a Network (client side prediction)

Jaroslav Sinecky jsinecky at tiscali.cz
Thu Dec 1 02:46:37 MST 2005


Check out this page: http://ode.org/cgi-bin/wiki.pl?SaveAndRestoreHowTo
You can also search the archives, this topic has been discussed. In any case
you need to send as many decimal places as your numbers can store (float
resp. double) (sending them in their binary representation is the best
IMHO). If there is difference even just on the last decimal place, you can
finally arrive at really different results.

Jaroslav

> -----Original Message-----
> From: ode-bounces at q12.org [mailto:ode-bounces at q12.org]On Behalf
> Of Emmanuel Astier
> Sent: Thursday, December 01, 2005 9:54 AM
> To: ode at q12.org
> Subject: Re: [ODE] Syncing ODE over a Network (client side prediction)
>
>
> You must also have the same dt in order to have the same simulation.
>
> Event with the same dt, two different processors can achieve different
> results, because of rounding differences in the FPU.
>
> On PC, you can set the rouding mode you want to have the same results on
> Intel and AMD.
>
> Hope it helps,
>
> Emmanuel
>
> -----Message d'origine-----
> De : ode-bounces at q12.org [mailto:ode-bounces at q12.org] De la part de Dave
> Kerr
> Envoyé : jeudi 1 décembre 2005 08:40
> À : ode at q12.org
> Objet : [ODE] Syncing ODE over a Network (client side prediction)
>
> Hi, I'm building a multiplayer game engine and am trying to sync
> ODE objects
>
> over a network.
> All is fine if I sync static objects...
> But with moving objects, the clients and server usually end up
> out of sync.
> Ie the simulations rarely run the same course.
>
> I was wondering if this is part of ODE (random numbers)?
> Or am I missing some variables to sync:
>
> When a client connects, they get Position, Linear Velocity, Angular
> Velocity, Mass, Density, Size/shape and the Quaternion
> to 12 decimal places, of every body.
> It works if all objects are resting (or at least looks like its working),
> but once a moving object is added, the simulations take different paths.
> How many decimal places is necessary?
> Am I missing some data?
>
> I would like to minimize the network traffic by trusting the clients to
> correctly predict most of the ODE physics.
>
> Any thoughts on this subject would be appreciated.
>
> cheers,
> Dave
>
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode
>
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode



More information about the ODE mailing list