added hierarchical profiling (needs more cleanup)
avoid dynamic allocations in btRaycastVehicle
This commit is contained in:
@@ -300,8 +300,6 @@ void CcdPhysicsDemo::displayCallback(void) {
|
||||
|
||||
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
|
||||
|
||||
m_dynamicsWorld->updateAabbs();
|
||||
|
||||
|
||||
renderme();
|
||||
|
||||
@@ -431,11 +429,13 @@ int maxNumOutstandingTasks = 4;
|
||||
createSolverLocalStoreMemory,
|
||||
maxNumOutstandingTasks));
|
||||
|
||||
m_solver = new btParallelSequentialImpulseSolver(threadSupportSolver,maxNumOutstandingTasks);
|
||||
m_solver = new btParallelSequentialImpulseSolver(m_threadSupportSolver,maxNumOutstandingTasks);
|
||||
#else
|
||||
m_solver = new btSequentialImpulseConstraintSolver;
|
||||
btSequentialImpulseConstraintSolver* solver = new btSequentialImpulseConstraintSolver();
|
||||
|
||||
m_solver = solver;
|
||||
//default solverMode is SOLVER_RANDMIZE_ORDER. Warmstarting seems not to improve convergence, see
|
||||
//m_solver->setSolverMode(btSequentialImpulseConstraintSolver::SOLVER_USE_WARMSTARTING | btSequentialImpulseConstraintSolver::SOLVER_RANDMIZE_ORDER);
|
||||
//solver->setSolverMode(0);//btSequentialImpulseConstraintSolver::SOLVER_USE_WARMSTARTING | btSequentialImpulseConstraintSolver::SOLVER_RANDMIZE_ORDER);
|
||||
|
||||
#endif //USE_PARALLEL_SOLVER
|
||||
|
||||
|
||||
Reference in New Issue
Block a user