- fixing various race conditions throughout (usage of static vars, etc)
- addition of a few lightweight mutexes (which are compiled out by default)
- slight code rearrangement in discreteDynamicsWorld to facilitate multithreading
- PoolAllocator::allocate() can now be called when pool is full without
crashing (null pointer returned)
- PoolAllocator allocate and freeMemory, are OPTIONALLY threadsafe
(default is un-threadsafe)
- CollisionDispatcher no longer checks if the pool allocator is full
before calling allocate(), instead it just calls allocate() and
checks if the return is null -- this avoids a race condition
- SequentialImpulseConstraintSolver OPTIONALLY uses different logic in
getOrInitSolverBody() to avoid a race condition with kinematic bodies
- addition of 2 classes which together allow simulation islands to be run
in parallel:
- btSimulationIslandManagerMt
- btDiscreteDynamicsWorldMt
- MultiThreadedDemo example in the example browser demonstrating use of
OpenMP, Microsoft PPL, and Intel TBB
- use multithreading for other demos
- benchmark demo: add parallel raycasting
For OSX:
cd build
./premake_osx xcode4
for iOS:
cd build
./ios_build.sh
./ios_run.sh
Also integrated the branches/StackAllocation to make it easier to multi-thread collision detection in the near future. It avoids changing the btCollisionObject while performing collision detection.
As this is a large patch, some stuff might be temporarily broken, I'll keep an eye out on issues.
+ Allow user to enable useConvexConservativeDistanceUtil . Use dynamicsWorld->getDispatchInfo().m_useConvexConservativeDistanceUtil = true;
(see Demos/Benchmarks/Benchmark4 (convex objects falling down)
+ Fix for plane drawing (just wire-frame)
+ Gimpact: use collision margin of 0.07 for demo (because BULLET_TRIANGLE_COLLISION is used)
+ replace dot,cross,distance,angle,triple in btVector3 by btDot, btCross,btDistance,btAngle,btDistance to avoid naming conflicts
+ Some fixes in GJK penetration depth normal direction (broken in a previous commit)
+ fix in calculateDiffAxisAngleQuaternion to make ConvexConservativeDistanceUtil work properly
+ allow debug drawing to debug btContinuousConvexCollision
+ add comment/warning that btTriangleMesh::findOrAddVertex is an internal method, users should use addTriangle instead
Add Concave Convexcast demo back in AllBulletDemos, and tweaked it a bit.
Fix view frustum of ForkLiftDemo (caused picking to fail)
Removed innerloop profiling for ray and convex casts, it hurts performance.
Set default #aabb's in CDTestFramework to 8192
Fix issues with btDbvt: use ATTRIBUTE_ALIGNED16
Removed several places where variable was declared within for statement: for (int i=0; should be int i; for (i=0
Fixed recently introduced issue: can't re-use stack as member variable (to reduce memory allocs) in btDbvt, due to recursion.
remove btGjkEpa.* from Makefile.am, Makefile and CMakeLists.txt
avoid division-by-zero in ODE boxbox contact reduction
btQuantizedBvh has a version number, memory layout might be different now (due to aligned btVector3)
reorganized some data members of some classes, to reduce memory footprint
Fix btConvexTriangleMeshShape::calculatePrincipalAxisTransform
Both thanks to Ole.
Use our own union to extract a scalar value from an SSE intrinsic (instead of relying on MSVC-only m128_u32
Thanks to BlindSide
and rayTestInternal re-uses precomputed invRayDirection/signs.
also did some performance comparison with different ray-AABB test, from
http://jgt.akpeters.com/papers/EisemannEtAl07/
In short: it is faster, but it is not clear how to cull ray segments using ray slopes: when rays starts inside the AABB, we get a negative t value, but negatives also get false t-values...
Renamed collideRay -> rayTest to reflect change to (rayFrom,rayTo) interface.
Re-use btRayAabb2 function from src/LinearMath/btAabbUtil2.h in btDbvt instead of duplicated implementation.
rename AddSingleResult to addSingleResult
moved collision filtering for btCollisionWorld::rayTest and btCollisionWorld::convexSweepTest from argument to the callback needsCollision
