[ODE] floats/doubles

[Tobias Thorsen]

I did compile ode.. My own system runs floats everywhere.
I changed to PRESCISION=SINGLE in the user-settings file, and reran make
ode-lib.

Tobias

----- Original Message -----
From: <skjold@cistron.nl>
To: <ode@q12.org>
Sent: Thursday, February 13, 2003 2:18 AM
Subject: Re: [ODE] floats/doubles


> Hi,
>
> I had similar problems at one time. This might seem trivial, but it's
possible: Did you forget to recompile the ODE library using single
precision? It's not enough to just recompile your own application, you
should recompile it with the same setting (single/double precision) as your
own application.
>
> Greets,
> Mark
>
>
> > Hi ODE list
> >
> > Ive implemented ode into a rudimentary engine i created, and i just used
the default settings for a while with double prescision. So now i wanted to
test how much performance i would get when compiling with single prescision
instead. But the problem is that it simply doesnt run anymore. I dont do any
"illegal" conversions between my internal numbers and dReals. Only an
occational pos.x = (float) pos[0] and simple stuff like that.
> > I know the problem can be anywhere in my program and might not have
anything to do with ODE, but has anyone experienced similar problems?
> > I get some warnings when i compile, but none that i would consider
dangerous.
> >
> > My app seems to start up, and runs very slow using a LOT of time in the
ODE simulate code. 25ms or more.  (it ran less than 8 in double prescision
mode) (800mhz pIII). All my models are gone (probably off to infinity) and
the program seems to crash on the strangest places where it really shouldnt
crash.
> >
> > Any ideas?
> >
> > Thanks.
> > Tobias Thorsen
> > www.spacetimefoam.com
> >
>
> _______________________________________________
> ODE mailing list
> ODE@q12.org
> http://q12.org/mailman/listinfo/ode
>