[ODE] processing islands in different order

Thomas Yeh tomyeh at CS.UCLA.EDU
Fri Aug 25 01:31:34 MST 2006


Well, the position data starts diverging even in the first frame.

I've turned off the flags that result in non-deterministic behavior such as
REORDER_CONSTRAINTS and RANDOMLY_REORDER_CONSTRAINTS.

If I run the default ode(with the above flags turned off) , I can
reproduce exact results.

I am just trying to understand the engine better, so that I don't break
it with my modifications.

if anyone has an idea please let me know.

thx,

On Fri, 25 Aug 2006, Bram Stolk wrote:

> Thomas Yeh wrote:
> > I am seeing the behavior that if islands are processed in a different
> > order than the default ordering, the position data for the objects in the
> > simulation are slightly different. I am using 0.5, and I generate all the
> > islands before I start calling the stepper function.
> >
> > Can anyone explain this behavior? As far as I know, the islands'
> > computations should all be independent right?
>
> Do you mean that after many iterations (seconds), the result diverge?
>
> That is to be expected.
> As a matter of fact, simulations in ODE are not even reproducible
> (at least not on x86)
>
> Take e.g. test_basket from the ode tests... for me, sometimes
> it is a miss, sometimes it scores a basket.
>
>    bram
>
> >
> > thx
> >
> > _______________________________________________
> > 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