[ODE] ODE rocks!

[Alex Holkner]

> Speaking of a wall of bricks.
> How many stacked rigid bodies can ODE handle in real-time today
> on a, say, 2GHz P4?
> Assume I would pour spheres or boxes into one big box and let
> the system relax. How many would I be able to do: 50, 100, 500?

I have been running exactly this experiment -- creating a whole lot of boxes
in random positions -- with my python interface to ODE (which will
undoubtedly affect its performance):

- Up to 300 boxes.. no problems at all (very occaisonally a box will
"explode")
- 300 - 1600.. boxes periodically explode (a few every second)
- 1600 - 2500... boxes move in an unpredictable fashion
- More than 2500 the program crashes with a stack overflow (I have not
investigated this further yet.. it may well be a problem with my interface)

This is on Win 2000 platform, P3 850Mhz.  The time steps are set up so that
at gravity = -15 everything moves at a realistic speed (not terribly
methodical, I know).  I wonder if it would be worth setting up some tests to
drill ODE without the overhead of a renderer or (in my case) the Python
interface?

Alex.