Real-Time Physics Simulation Forum 2007-03-21T15:26:50+00:00 https://pybullet.org/Bullet/phpBB3/app.php/feed/topic/1063 2007-03-21T15:24:38+00:00 2007-03-21T15:24:38+00:00 https://pybullet.org/Bullet/phpBB3/viewtopic.php?p=3979#p3979 <![CDATA[Bullet 2.47 released, small constraint solver optimizations]]>
Some features like setAngularFactor, and custom friction model callbacks don't work yet with the 'cache friendly' solver mode.

But you can switch to original (non-cache-friendly) solver mode, and then it should works again:

Code: 

btSequentialImpulseConstraintSolver* solver = new btSequentialImpulseConstraintSolver;//default solverMode is  SOLVER_CACHE_FRIENDLY and SOLVER_RANDMIZE_ORDER.//Switch back to original solver (non-cache friendly):solver->setSolverMode(btSequentialImpulseConstraintSolver::SOLVER_RANDMIZE_ORDER);
Hope this helps,
Erwin

Statistics: Posted by Erwin Coumans — Wed Mar 21, 2007 3:24 pm

]]> 2007-03-19T23:25:35+00:00 2007-03-19T23:25:35+00:00 https://pybullet.org/Bullet/phpBB3/viewtopic.php?p=3960#p3960 <![CDATA[Bullet 2.47 released, small constraint solver optimizations]]> Statistics: Posted by Hamstray — Mon Mar 19, 2007 11:25 pm

]]> 2007-03-19T12:22:12+00:00 2007-03-19T12:22:12+00:00 https://pybullet.org/Bullet/phpBB3/viewtopic.php?p=3945#p3945 <![CDATA[Bullet 2.47 released, small constraint solver optimizations]]>
Compilation fails on linux / systems with case sensitive file names:
Rename
src/BulletDynamics/ConstraintSolver/btsolverbody.h to btSolverBody.h
src/BulletDynamics/ConstraintSolver/btsolverconstraint.h to btSolverConstraint.h

Statistics: Posted by Codepoet — Mon Mar 19, 2007 12:22 pm

]]> 2007-03-21T15:26:50+00:00 2007-03-17T08:56:56+00:00 https://pybullet.org/Bullet/phpBB3/viewtopic.php?p=3931#p3931 <![CDATA[Bullet 2.47 released, small constraint solver optimizations]]>
Constraint solver has been optimized with a second path. Avoiding cross products during the iterations and gathering rigid body information into contiguous memory are the main changes. No huge benefits yet, but a good step in the right direction.

It can be disabled by removing 'SOLVER_CACHE_FRIENDLY' from

Code: 

solver->setSolverMode(SOLVER_RANDMIZE_ORDER | SOLVER_CACHE_FRIENDLY)
Another stability fix is enabled, that adds some damping for rigid bodies with linear/angular velocity below a treshold. This heuristic is experimental, and can be disabled using

Code: 

gJitterVelocityDampingFactor = 0.f;
No proper API for this, as it is experimental.

Enjoy,
Erwin

Statistics: Posted by Erwin Coumans — Sat Mar 17, 2007 8:56 am

]]>