Since updating to 3.4 I'm experiencing an access violation whenever I remove a dynamic rigid body from the simulation, either by changing it's eDISABLE_SIMULATION flag state, or when I delete the actor. Only occurs when I build in Release mode for x64 targets; all other combinations don't experience this issue.

The calling code seems related to touch events, but the problem occurs even if there is no simulationEventCallback set and eNOTIFY_TOUCH_FOUND/eNOTIFY_TOUCH_LOST are not included in filter pair flags. Also doesn't matter what form of CCD I'm using (or lack thereof), or if I'm using GPU rigid bodies or broad phase.

Access violation: Exception thrown at 0x00007FFE9D6B483E (PhysX3_x64.dll) in Bricks.exe: 0xC0000005: Access violation writing location 0x0000000000000002. occurred (in ScNPhaseCore.cpp, line 2171)

Stack trace:
(sidebar: if anyone can tell me a less clunky way of sharing a stack trace for Visual Studio's debugger, I'd be grateful!)

Memory dump: https://drive.google.com/file/d/0B6cTOEVTlAMJVHhadkV4NGlZREE/view?usp=sharing

I deployed a big scene on PhysX. RayCast is used for ray detection, but after using OpenMP for multi-core acceleration, although the parallel performance is improved, the speed of ray detection is significantly reduced. Below are statistics on the number of threads and time spent per ray.

i7-4790 (max threads:8) 1 thread: 0.055ms per ray. 2 threads: 0.057ms per ray. 4 threads: 0.075ms per ray. 8 threads: 0.104ms per ray.

i9-7980XE (max threads: 36) 8 threads: 0.074ms per ray. 36 threads: 0.442ms per ray.

RayCast is a process of read-only calculations of the scene, why performance is so much degraded in parallel. How can I modify my program to make it better? Thank you.

I've been playing around with GPU rigid bodies in a sandbox application, and every so often I experience a soft-crash when modifying many rigid bodies at once (between simulation ticks). You can see an example of it occurring in this video (I get it to occur at 1:21, but watch from at least 1:00 for context). It doesn't happen very often (say, 1 in 30 times doing the above), but when it does it brings down PhysX entirely.

You can see the error log spam in the error console after it occurs, but the very first errors reported are:

[PxgNarrowphaseCore.cpp:1039] memcpy failed fail! 2  700
[PxgNarrowphaseCore.cpp:1049] GPU cudaMainGjkEpa or prepareLostFoundPairs kernel fail!
  700
[PxgNarrowphaseCore.cpp:1274] memcpy failed fail!
  700
[PxgCudaUtils.h:53] SynchronizeStreams failed
[PxgCudaSolverCore.cpp:1021] GPU contactConstraintPrepare fail to launch kernel!!

The 'grenade explosion' is a fairly simple loop over all dynamic rigid bodies and adding force based on their distance and direction from the point of origin (source). I haven't seen the same/similar issue with GPU rigid bodies disabled.

The relevant code is linked above, but ~~[here's the sandbox]~~ I'm using in the video.

Hi

I'm trying building in target for android under windows, using tegra-toolkit. how build it?

ndk-build returned "Could not find application project directory"

Hi, My application is dynamically linked at run-time therefore I am compiling all PhysX libraries with /MD but, I am getting this error when compiling my application with PhysX integration:

2>PhysX3ExtensionsPROFILE.lib(ExtD6Joint.obj) : error LNK2001: unresolved external symbol __libm_sse2_sincosf_
2>PhysX3ExtensionsPROFILE.lib(ExtFixedJoint.obj) : error LNK2001: unresolved external symbol __libm_sse2_sincosf_
2>PhysX3ExtensionsPROFILE.lib(ExtPrismaticJoint.obj) : error LNK2001: unresolved external symbol __libm_sse2_sincosf_
2>PhysX3ExtensionsPROFILE.lib(ExtRevoluteJoint.obj) : error LNK2001: unresolved external symbol __libm_sse2_sincosf_
2>PhysX3VehiclePROFILE.lib(PxVehicleUpdate.obj) : error LNK2019: unresolved external symbol __libm_sse2_sincosf_ referenced in function "unsigned int __cdecl physx::PxVehicleModifyWheelContacts(class physx::PxVehicleWheels const &,unsigned int,float,float,class physx::PxContactModifyPair &)" ([email protected]@@[email protected]@[email protected]@@Z)

But, from what I could find it seem I would need Intel's libm.lib but, I can't get this library since it's not freely available. I would be very grateful if you could give me some pointers on what I am doing wrong in my integration or if there is any alternative to fix this issue.

Thank you very much

I try to init GPU Rigid Bodies :

	PxGpuDispatcher * GPUDispatcher;
	PxCudaContextManager* gCudaContextManager;

	PxCudaContextManagerDesc cudaContextManagerDesc;
	ZeroMemory(&cudaContextManagerDesc, sizeof ( PxCudaContextManagerDesc ));
	cudaContextManagerDesc.graphicsDevice = MY_DX11_device;
	cudaContextManagerDesc.interopMode = PxCudaInteropMode::D3D11_INTEROP;
	cudaContextManagerDesc.ctx = NULL;
	cudaContextManagerDesc.appGUID = "APName";

	gCudaContextManager = PxCreateCudaContextManager( *gFoundation, cudaContextManagerDesc );

	if ( gCudaContextManager )	
	GPUDispatcher = gCudaContextManager->getGpuDispatcher();

	if ( GPUDispatcher )		
	{
		desc.gpuDispatcher = GPUDispatcher;
		desc.flags |= PxSceneFlag::eENABLE_GPU_DYNAMICS;
		desc.flags |= PxSceneFlag::eENABLE_PCM;
		desc.flags |= PxSceneFlag::eENABLE_STABILIZATION;
		desc.broadPhaseType = PxBroadPhaseType::eGPU;
		desc.gpuMaxNumPartitions = 8;
		desc.gpuDynamicsConfig.constraintBufferCapacity = (32 * 1024 * 1024);
		desc.gpuDynamicsConfig.contactBufferCapacity = (24 * 1024 * 1024);
		desc.gpuDynamicsConfig.tempBufferCapacity = (16 * 1024 * 1024);
		desc.gpuDynamicsConfig.contactStreamSize = (6 * 1024 * 1024);
		desc.gpuDynamicsConfig.patchStreamSize = (5 * 1024 * 1024);
		desc.gpuDynamicsConfig.forceStreamCapacity = (1 * 1024 * 1024);
		desc.gpuDynamicsConfig.heapCapacity = (64 * 1024 * 1024);
		desc.gpuDynamicsConfig.foundLostPairsCapacity = (256 * 1024);
	}

	gScene = gPhysics->createScene(desc);

Initialization was successful, but RigidBodies - Boxes behavior are wrong : Something wrong with Boxes gravity and frictions if colliding with Heightfield or with each other. Boxes just skating on Heightfield surface as there no frictions at all.

If i use CPU PhysX i have no problem with RigidBodies and Heightfield behavior.

What is wrong with initialisation or whatever?

Hi, is there a way to get torque applied by a motor in a joint?

I also experience a problem where motor set to zero velocity never gets to stop the wheel when that wheel is propelled by friction from the ground. Could I fix that somehow? Is that because the motor only applies the torque needed to stop the rotating mass of given inertia? I hoped that motors have more info on the post-friction velocity and can be more precise then applying and calculating the torque myself in some "braking PID controller". Could you help me understand?

I am using Visual Studio 2017 professional and I used the solution in \PhysX_3.4\Source\compiler\vc15win64 since that was the latest version (so it is logically the one that is most likely to be compiled successfully). Can you edit the description and make it easier for everyone to understand what they have to do to compile the SDK? The documentation in https://docs.nvidia.com/gameworks/content/gameworkslibrary/physx/guide/Manual/Index.html is also very poorly written and vague...

Hey, I have a height-field terrain and a convex road that emerges from below the terrain. Then I have a sphere that rolls on a heightfield towards the road. When it rolls on HF it generates a nice contact with near zero separation. Then as it rolls over the road, it detects another collision with the convex with 6 cm separation. Both of those collisions have the same normal.

However, instead of giving me two contacts (with 0 and 6 cm separation), PhysX seems to merge the two contacts into one (with 6 cm separation) which results in no force applied. Then a couple frames later, when separation with convex gets to 0, the force quickly grows and pushes the sphere up - flying. I expect the sphere to roll smoothly in that transition area between the terrain and the road.

Can I selectively disable merging contacts into patches? Or disable it in PhysX source code? Can I disable merging via onContactModify maybe? And why is PhysX using 6 cm separation instead of 0 when merging those contacts? That doesn't make sense to me...

I'm using a non PCM version of contact detection.

When I use xcode 10.1 to compile PhysX, tell me compile errer with : 'requires' is a keyword in C++2a where use is Source/SceneQuery/include/SqPruningStructure.h virtual void requires(PxProcessPxBaseCallback&);

https://github.com/NVIDIAGameWorks/PhysX-3.4/blob/7f12de60542edc8f1c6683e6b4cdce8570e51456/PhysX_3.4/Source/PhysXCooking/src/convex/ConvexMeshBuilder.cpp#L264 cooking always fails when 16-bit indices are provided because it tries to process 3 at a time

				*dest++ = trig16[0];
				*dest++ = trig16[1];
				*dest++ = trig16[2];

should be replaced with; *dest++ = *trig16;

Program received signal SIGSEGV, Segmentation fault.
0x00007ffff69c2860 in __memset_sse2 () from /lib64/libc.so.6
Missing separate debuginfos, use: debuginfo-install glibc-2.17-196.tl2.3.x86_64 libgcc-7.3.1-6.tl2.x86_64 libstdc++-7.3.1-6.tl2.x86_64 zlib-1.2.7-15.el7.x86_64
(gdb) bt
#0  0x00007ffff69c2860 in __memset_sse2 () from /lib64/libc.so.6
#1  0x0000000000b79d6b in memSet (count=<optimized out>, c=<optimized out>, dest=<optimized out>) at /data/rwbyguo/trunk/server/external/PhysX/PxShared/include/foundation/unix/PxUnixIntrinsics.h:166
#2  physx::shdfnd::internal::HashBase<physx::shdfnd::Pair<physx::PxBase* const, unsigned long>, physx::PxBase*, physx::shdfnd::Hash<physx::PxBase*>, physx::shdfnd::internal::HashMapBase<physx::PxBase*, unsigned long, physx::shdfnd::Hash<physx::PxBase*>, physx::shdfnd::NonTrackingAllocator>::GetKey, physx::shdfnd::NonTrackingAllocator, true>::reserveInternal ([email protected]=0x680af90, size=268435456) at /data/rwbyguo/trunk/server/external/PhysX/PxShared/src/foundation/include/PsHashInternals.h:381
#3  0x0000000000b778fc in reserve (size=<optimized out>, this=0x680af90) at /data/rwbyguo/trunk/server/external/PhysX/PxShared/src/foundation/include/PsHashInternals.h:209
#4  reserve (size=<optimized out>, this=0x680af90) at /data/rwbyguo/trunk/server/external/PhysX/PxShared/src/foundation/include/PsHashInternals.h:777
#5  physx::Cm::Collection::add (this=0x680af50, _collection=...) at /data/rwbyguo/trunk/server/external/PhysX/PhysX_3.4/Source/Common/src/CmCollection.cpp:117

Collection::add reserved too many entries. After added collection serveral times, reserve size reached 268435456.

When I add multiple skeleton rigidbodies of ragdoll to a PxAggregate, it crashes when ragdoll enter the trigger range. In fact, whenever Aggregate contains a shape greater than 1 and intersects another trigger Shape, it will inevitably collapse.

I now have a step-shaped PxTriangleMesh and set its MeshScale value to PxVec3(-1, 1, 1). This results in a TriangleMesh that is a mirror image of the X-axis. Then roll a small ball down the top of the step. Each time the ball passes the edge of the step, it falls into the TriangleMesh and then pops out of the TriangleMesh. It looks like the edge of the steps didn't collide.

Does the API provide any means to copy a cloth fabric created by NvClothCookFabricFromMesh? It takes some time to run so, I would like to cache it and copy it's when spawning multiple copies of the same mesh

I have a revolute joint attached to an actor and set a velocity target in the motor. No problems, until when external forces resist the motor. The motor will only ever exert the amount force it would take to bring the mass of the actor it's attached to up to speed. It does not scale up the force to account for other actors attached, collisions or gravity. This means that it will likely never get to the desired speed even though it is not exerting all its allotted force. In my case I've set the max force to PX_MAX_F32. I have the same problem when I want the motor to keep the actor steady at velocity zero (given that it's got sufficient force to apply).

I'm trying to build cranes in my simulation and I don't know how to solve this the best way. Should I somehow implement my own motor that can handle external forces? Any other tips?

Cheers