Real-Time Physics Simulation Forum 2023-02-17T13:52:48+00:00 https://pybullet.org/Bullet/phpBB3/app.php/feed/topic/13537 2023-02-17T13:52:48+00:00 2023-02-17T13:52:48+00:00 https://pybullet.org/Bullet/phpBB3/viewtopic.php?p=44256#p44256 <![CDATA[Re: Properly achieving determinism (on same machine)]]> - m_deterministicOverlappingPairs = true;
- performDiscreteCollisionDetection()

Keep in mind, I tried all these configurations together and different combinations of them

Statistics: Posted by anascim — Fri Feb 17, 2023 1:52 pm

]]> 2023-02-17T13:51:24+00:00 2023-02-17T13:51:24+00:00 https://pybullet.org/Bullet/phpBB3/viewtopic.php?p=44255#p44255 <![CDATA[Properly achieving determinism (on same machine)]]> However, to achieve determinism, I currently have to delete all rigidbodies and recreate them, which is a bit slow.
It appears that something about the rigidbodies is cached from the previous frame but I couldn't figure it out.
Things that I already tried and failed the determinism test:
- calling resetPool on the broadphase
- calling refreshBroadphaseProxy for every collision object
- calling cleanProxyFromPairs passing each collision object
- calling reset on the constraint solver
- disabling warmstaring on the m_solverMode

Is there nothing else I can do to make deterministic simulations on reset? Or do I really have to delete everything and recreate?

Statistics: Posted by anascim — Fri Feb 17, 2023 1:51 pm

]]>