[ODE] is ODE multi-threaded?
Jon Watte (ODE)
hplus-ode at mindcontrol.org
Thu Oct 4 07:04:45 MST 2007
David Black wrote:
> I believe there can be value splitting the physics further than a
> seperate thread for just physics, although mainly if you are (un)lucky
> enough to be developing for a platform with more than two cores(I guess
> this will include PC in the next year or two as well).
There absolutely can be value. However, ODE, as currently implemented,
doesn't quite allow you to separate collision and simulation, because of
the dGeomMoved() call. If that could be buffered and deferred, you could
split into at least two threads. You could even use more threads, after
collision broad-phase is done for a step.
More information about the ODE