ruby-bullet 0.0.2-x86-mingw32

data/deps/include/bullet/BulletMultiThreaded/SpuLibspe2Support.h

@@ -0,0 +1,180 @@
+/*
+Bullet Continuous Collision Detection and Physics Library
+Copyright (c) 2003-2007 Erwin Coumans  http://bulletphysics.com
+
+This software is provided 'as-is', without any express or implied warranty.
+In no event will the authors be held liable for any damages arising from the use of this software.
+Permission is granted to anyone to use this software for any purpose, 
+including commercial applications, and to alter it and redistribute it freely, 
+subject to the following restrictions:
+
+1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.
+2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.
+3. This notice may not be removed or altered from any source distribution.
+*/
+
+
+#ifndef BT_SPU_LIBSPE2_SUPPORT_H
+#define BT_SPU_LIBSPE2_SUPPORT_H
+
+#include <LinearMath/btScalar.h> //for uint32_t etc.
+
+#ifdef USE_LIBSPE2
+
+#include <stdlib.h>
+#include <stdio.h>
+//#include "SpuNarrowPhaseCollisionTask/SpuGatheringCollisionTask.h"
+#include "PlatformDefinitions.h"
+
+
+//extern struct SpuGatherAndProcessPairsTaskDesc;
+
+enum
+{
+	Spu_Mailbox_Event_Nothing = 0,
+	Spu_Mailbox_Event_Task = 1,
+	Spu_Mailbox_Event_Shutdown = 2,
+	
+	Spu_Mailbox_Event_ForceDword = 0xFFFFFFFF
+	
+};
+
+enum
+{
+	Spu_Status_Free = 0,
+	Spu_Status_Occupied = 1,
+	Spu_Status_Startup = 2,
+	
+	Spu_Status_ForceDword = 0xFFFFFFFF
+	
+};
+
+
+struct btSpuStatus
+{
+	uint32_t	m_taskId;
+	uint32_t	m_commandId;
+	uint32_t	m_status;
+
+	addr64 m_taskDesc;
+	addr64 m_lsMemory;
+	
+}
+__attribute__ ((aligned (128)))
+;
+
+
+
+#ifndef __SPU__
+
+#include "LinearMath/btAlignedObjectArray.h"
+#include "SpuCollisionTaskProcess.h"
+#include "SpuSampleTaskProcess.h"
+#include "btThreadSupportInterface.h"
+#include <libspe2.h>
+#include <pthread.h>
+#include <sched.h>
+
+#define MAX_SPUS 4 
+
+typedef struct ppu_pthread_data 
+{
+	spe_context_ptr_t context;
+	pthread_t pthread;
+	unsigned int entry;
+	unsigned int flags;
+	addr64 argp;
+	addr64 envp;
+	spe_stop_info_t stopinfo;
+} ppu_pthread_data_t;
+
+
+static void *ppu_pthread_function(void *arg)
+{
+    ppu_pthread_data_t * datap = (ppu_pthread_data_t *)arg;
+    /*
+    int rc;
+    do 
+    {*/
+        spe_context_run(datap->context, &datap->entry, datap->flags, datap->argp.p, datap->envp.p, &datap->stopinfo);
+        if (datap->stopinfo.stop_reason == SPE_EXIT) 
+        {
+           if (datap->stopinfo.result.spe_exit_code != 0) 
+           {
+             perror("FAILED: SPE returned a non-zero exit status: \n");
+             exit(1);
+           }
+         } 
+        else 
+         {
+           perror("FAILED: SPE abnormally terminated\n");
+           exit(1);
+         }
+        
+        
+    //} while (rc > 0); // loop until exit or error, and while any stop & signal
+    pthread_exit(NULL);
+}
+
+
+
+
+
+
+///SpuLibspe2Support helps to initialize/shutdown libspe2, start/stop SPU tasks and communication
+class SpuLibspe2Support : public btThreadSupportInterface
+{
+
+	btAlignedObjectArray<btSpuStatus>	m_activeSpuStatus;
+	
+public:
+	//Setup and initialize SPU/CELL/Libspe2
+	SpuLibspe2Support(spe_program_handle_t *speprog,int numThreads);
+	
+	// SPE program handle ptr.
+	spe_program_handle_t *program;
+	
+	// SPE program data
+	ppu_pthread_data_t data[MAX_SPUS];
+	
+	//cleanup/shutdown Libspe2
+	~SpuLibspe2Support();
+
+	///send messages to SPUs
+	void sendRequest(uint32_t uiCommand, uint32_t uiArgument0, uint32_t uiArgument1=0);
+
+	//check for messages from SPUs
+	void waitForResponse(unsigned int *puiArgument0, unsigned int *puiArgument1);
+
+	//start the spus (can be called at the beginning of each frame, to make sure that the right SPU program is loaded)
+	virtual void startSPU();
+
+	//tell the task scheduler we are done with the SPU tasks
+	virtual void stopSPU();
+
+	virtual void setNumTasks(int numTasks)
+	{
+		//changing the number of tasks after initialization is not implemented (yet)
+	}
+
+private:
+	
+	///start the spus (can be called at the beginning of each frame, to make sure that the right SPU program is loaded)
+	void internal_startSPU();
+
+
+	
+	
+	int numThreads;
+
+};
+
+#endif // NOT __SPU__
+
+#endif //USE_LIBSPE2
+
+#endif //BT_SPU_LIBSPE2_SUPPORT_H
+
+
+
+

data/deps/include/bullet/BulletMultiThreaded/SpuNarrowPhaseCollisionTask/Box.h

@@ -0,0 +1,167 @@
+/*
+   Copyright (C) 2006, 2008 Sony Computer Entertainment Inc.
+   All rights reserved.
+
+This software is provided 'as-is', without any express or implied warranty.
+In no event will the authors be held liable for any damages arising from the use of this software.
+Permission is granted to anyone to use this software for any purpose, 
+including commercial applications, and to alter it and redistribute it freely, 
+subject to the following restrictions:
+
+1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.
+2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.
+3. This notice may not be removed or altered from any source distribution.
+
+*/
+
+#ifndef __BOX_H__
+#define __BOX_H__
+
+
+#ifndef PE_REF
+#define PE_REF(a) a&
+#endif
+
+#include <math.h>
+
+
+#include "../PlatformDefinitions.h"
+
+
+
+
+enum FeatureType { F, E, V };
+
+//----------------------------------------------------------------------------
+// Box
+//----------------------------------------------------------------------------
+///The Box is an internal class used by the boxBoxDistance calculation.
+class Box
+{
+public:
+	vmVector3 mHalf;
+
+	inline Box()
+	{}
+	inline Box(PE_REF(vmVector3) half_);
+	inline Box(float hx, float hy, float hz);
+
+	inline void Set(PE_REF(vmVector3) half_);
+	inline void Set(float hx, float hy, float hz);
+
+	inline vmVector3 GetAABB(const vmMatrix3& rotation) const;
+};
+
+inline
+Box::Box(PE_REF(vmVector3) half_)
+{
+	Set(half_);
+}
+
+inline
+Box::Box(float hx, float hy, float hz)
+{
+	Set(hx, hy, hz);
+}
+
+inline
+void
+Box::Set(PE_REF(vmVector3) half_)
+{
+	mHalf = half_;
+}
+
+inline
+void
+Box::Set(float hx, float hy, float hz)
+{
+	mHalf = vmVector3(hx, hy, hz);
+}
+
+inline
+vmVector3
+Box::GetAABB(const vmMatrix3& rotation) const
+{
+	return absPerElem(rotation) * mHalf;
+}
+
+//-------------------------------------------------------------------------------------------------
+// BoxPoint
+//-------------------------------------------------------------------------------------------------
+
+///The BoxPoint class is an internally used class to contain feature information for boxBoxDistance calculation.
+class BoxPoint
+{
+public:
+	BoxPoint() : localPoint(0.0f) {}
+
+	vmPoint3      localPoint;
+	FeatureType featureType;
+	int         featureIdx;
+
+	inline void setVertexFeature(int plusX, int plusY, int plusZ);
+	inline void setEdgeFeature(int dim0, int plus0, int dim1, int plus1);
+	inline void setFaceFeature(int dim, int plus);
+
+	inline void getVertexFeature(int & plusX, int & plusY, int & plusZ) const;
+	inline void getEdgeFeature(int & dim0, int & plus0, int & dim1, int & plus1) const;
+	inline void getFaceFeature(int & dim, int & plus) const;
+};
+
+inline
+void
+BoxPoint::setVertexFeature(int plusX, int plusY, int plusZ)
+{
+	featureType = V;
+	featureIdx = plusX << 2 | plusY << 1 | plusZ;
+}
+
+inline
+void
+BoxPoint::setEdgeFeature(int dim0, int plus0, int dim1, int plus1)
+{
+	featureType = E;
+
+	if (dim0 > dim1) {
+		featureIdx = plus1 << 5 | dim1 << 3 | plus0 << 2 | dim0;
+	} else {
+		featureIdx = plus0 << 5 | dim0 << 3 | plus1 << 2 | dim1;
+	}
+}
+
+inline
+void
+BoxPoint::setFaceFeature(int dim, int plus)
+{
+	featureType = F;
+	featureIdx = plus << 2 | dim;
+}
+
+inline
+void
+BoxPoint::getVertexFeature(int & plusX, int & plusY, int & plusZ) const
+{
+	plusX = featureIdx >> 2;
+	plusY = featureIdx >> 1 & 1;
+	plusZ = featureIdx & 1;
+}
+
+inline
+void
+BoxPoint::getEdgeFeature(int & dim0, int & plus0, int & dim1, int & plus1) const
+{
+	plus0 = featureIdx >> 5;
+	dim0 = featureIdx >> 3 & 3;
+	plus1 = featureIdx >> 2 & 1;
+	dim1 = featureIdx & 3;
+}
+
+inline
+void
+BoxPoint::getFaceFeature(int & dim, int & plus) const
+{
+	plus = featureIdx >> 2;
+	dim = featureIdx & 3;
+}
+
+#endif /* __BOX_H__ */

data/deps/include/bullet/BulletMultiThreaded/SpuNarrowPhaseCollisionTask/SpuCollisionShapes.cpp

@@ -0,0 +1,302 @@
+/*
+Bullet Continuous Collision Detection and Physics Library
+Copyright (c) 2003-2006 Erwin Coumans  http://continuousphysics.com/Bullet/
+
+This software is provided 'as-is', without any express or implied warranty.
+In no event will the authors be held liable for any damages arising from the use of this software.
+Permission is granted to anyone to use this software for any purpose, 
+including commercial applications, and to alter it and redistribute it freely, 
+subject to the following restrictions:
+
+1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required.
+2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.
+3. This notice may not be removed or altered from any source distribution.
+*/
+
+
+#include "SpuCollisionShapes.h"
+
+///not supported on IBM SDK, until we fix the alignment of btVector3
+#if defined (__CELLOS_LV2__) && defined (__SPU__)
+#include <spu_intrinsics.h>
+static inline vec_float4 vec_dot3( vec_float4 vec0, vec_float4 vec1 )
+{
+    vec_float4 result;
+    result = spu_mul( vec0, vec1 );
+    result = spu_madd( spu_rlqwbyte( vec0, 4 ), spu_rlqwbyte( vec1, 4 ), result );
+    return spu_madd( spu_rlqwbyte( vec0, 8 ), spu_rlqwbyte( vec1, 8 ), result );
+}
+#endif //__SPU__
+
+
+void computeAabb (btVector3& aabbMin, btVector3& aabbMax, btConvexInternalShape* convexShape, ppu_address_t convexShapePtr, int shapeType, const btTransform& xform)
+{
+	//calculate the aabb, given the types...
+	switch (shapeType)
+	{
+	case CYLINDER_SHAPE_PROXYTYPE:
+		/* fall through */
+	case BOX_SHAPE_PROXYTYPE:
+	{
+		btScalar margin=convexShape->getMarginNV();
+		btVector3 halfExtents = convexShape->getImplicitShapeDimensions();
+		halfExtents += btVector3(margin,margin,margin);
+		const btTransform& t = xform;
+		btMatrix3x3 abs_b = t.getBasis().absolute();  
+		btVector3 center = t.getOrigin();
+		btVector3 extent = btVector3(abs_b[0].dot(halfExtents),abs_b[1].dot(halfExtents),abs_b[2].dot(halfExtents));
+		
+		aabbMin = center - extent;
+		aabbMax = center + extent;
+		break;
+	}
+	case CAPSULE_SHAPE_PROXYTYPE:
+	{
+		btScalar margin=convexShape->getMarginNV();
+		btVector3 halfExtents = convexShape->getImplicitShapeDimensions();
+		//add the radius to y-axis to get full height
+		btScalar radius = halfExtents[0];
+		halfExtents[1] += radius;
+		halfExtents += btVector3(margin,margin,margin);
+#if 0
+		int capsuleUpAxis = convexShape->getUpAxis();
+		btScalar halfHeight = convexShape->getHalfHeight();
+		btScalar radius = convexShape->getRadius();
+		halfExtents[capsuleUpAxis] = radius + halfHeight;
+#endif
+		const btTransform& t = xform;
+		btMatrix3x3 abs_b = t.getBasis().absolute();  
+		btVector3 center = t.getOrigin();
+		btVector3 extent = btVector3(abs_b[0].dot(halfExtents),abs_b[1].dot(halfExtents),abs_b[2].dot(halfExtents));
+		
+		aabbMin = center - extent;
+		aabbMax = center + extent;
+		break;
+	}
+	case SPHERE_SHAPE_PROXYTYPE:
+	{
+		btScalar radius = convexShape->getImplicitShapeDimensions().getX();// * convexShape->getLocalScaling().getX();
+		btScalar margin = radius + convexShape->getMarginNV();
+		const btTransform& t = xform;
+		const btVector3& center = t.getOrigin();
+		btVector3 extent(margin,margin,margin);
+		aabbMin = center - extent;
+		aabbMax = center + extent;
+		break;
+	}
+	case CONVEX_HULL_SHAPE_PROXYTYPE:
+	{
+		ATTRIBUTE_ALIGNED16(char convexHullShape0[sizeof(btConvexHullShape)]);
+		cellDmaGet(&convexHullShape0, convexShapePtr  , sizeof(btConvexHullShape), DMA_TAG(1), 0, 0);
+		cellDmaWaitTagStatusAll(DMA_MASK(1));
+		btConvexHullShape* localPtr = (btConvexHullShape*)&convexHullShape0;
+		const btTransform& t = xform;
+		btScalar margin = convexShape->getMarginNV();
+		localPtr->getNonvirtualAabb(t,aabbMin,aabbMax,margin);
+		//spu_printf("SPU convex aabbMin=%f,%f,%f=\n",aabbMin.getX(),aabbMin.getY(),aabbMin.getZ());
+		//spu_printf("SPU convex aabbMax=%f,%f,%f=\n",aabbMax.getX(),aabbMax.getY(),aabbMax.getZ());
+		break;
+	}
+	default:
+		{
+	//	spu_printf("SPU: unsupported shapetype %d in AABB calculation\n");
+		}
+	};
+}
+
+void dmaBvhShapeData (bvhMeshShape_LocalStoreMemory* bvhMeshShape, btBvhTriangleMeshShape* triMeshShape)
+{
+	register int dmaSize;
+	register ppu_address_t	dmaPpuAddress2;
+
+	dmaSize = sizeof(btTriangleIndexVertexArray);
+	dmaPpuAddress2 = reinterpret_cast<ppu_address_t>(triMeshShape->getMeshInterface());
+	//	spu_printf("trimeshShape->getMeshInterface() == %llx\n",dmaPpuAddress2);
+#ifdef __SPU__
+	cellDmaGet(&bvhMeshShape->gTriangleMeshInterfaceStorage, dmaPpuAddress2  , dmaSize, DMA_TAG(1), 0, 0);
+	bvhMeshShape->gTriangleMeshInterfacePtr = &bvhMeshShape->gTriangleMeshInterfaceStorage;
+#else
+	bvhMeshShape->gTriangleMeshInterfacePtr = (btTriangleIndexVertexArray*)cellDmaGetReadOnly(&bvhMeshShape->gTriangleMeshInterfaceStorage, dmaPpuAddress2  , dmaSize, DMA_TAG(1), 0, 0);
+#endif
+
+	//cellDmaWaitTagStatusAll(DMA_MASK(1));
+	
+	///now DMA over the BVH
+	
+	dmaSize = sizeof(btOptimizedBvh);
+	dmaPpuAddress2 = reinterpret_cast<ppu_address_t>(triMeshShape->getOptimizedBvh());
+	//spu_printf("trimeshShape->getOptimizedBvh() == %llx\n",dmaPpuAddress2);
+	cellDmaGet(&bvhMeshShape->gOptimizedBvh, dmaPpuAddress2  , dmaSize, DMA_TAG(2), 0, 0);
+	//cellDmaWaitTagStatusAll(DMA_MASK(2));
+	cellDmaWaitTagStatusAll(DMA_MASK(1) | DMA_MASK(2));
+}
+
+void dmaBvhIndexedMesh (btIndexedMesh* IndexMesh, IndexedMeshArray& indexArray, int index, uint32_t dmaTag)
+{		
+	cellDmaGet(IndexMesh, (ppu_address_t)&indexArray[index]  , sizeof(btIndexedMesh), DMA_TAG(dmaTag), 0, 0);
+	
+}
+
+void dmaBvhSubTreeHeaders (btBvhSubtreeInfo* subTreeHeaders, ppu_address_t subTreePtr, int batchSize, uint32_t dmaTag)
+{
+	cellDmaGet(subTreeHeaders, subTreePtr, batchSize * sizeof(btBvhSubtreeInfo), DMA_TAG(dmaTag), 0, 0);
+}
+
+void dmaBvhSubTreeNodes (btQuantizedBvhNode* nodes, const btBvhSubtreeInfo& subtree, QuantizedNodeArray&	nodeArray, int dmaTag)
+{
+	cellDmaGet(nodes, reinterpret_cast<ppu_address_t>(&nodeArray[subtree.m_rootNodeIndex]) , subtree.m_subtreeSize* sizeof(btQuantizedBvhNode), DMA_TAG(2), 0, 0);
+}
+
+///getShapeTypeSize could easily be optimized, but it is not likely a bottleneck
+int		getShapeTypeSize(int shapeType)
+{
+
+
+	switch (shapeType)
+	{
+	case CYLINDER_SHAPE_PROXYTYPE:
+		{
+			int shapeSize = sizeof(btCylinderShape);
+			btAssert(shapeSize < MAX_SHAPE_SIZE);
+			return shapeSize;
+		}
+	case BOX_SHAPE_PROXYTYPE:
+		{
+			int shapeSize = sizeof(btBoxShape);
+			btAssert(shapeSize < MAX_SHAPE_SIZE);
+			return shapeSize;
+		}
+	case SPHERE_SHAPE_PROXYTYPE:
+		{
+			int shapeSize = sizeof(btSphereShape);
+			btAssert(shapeSize < MAX_SHAPE_SIZE);
+			return shapeSize;
+		}
+	case TRIANGLE_MESH_SHAPE_PROXYTYPE:
+		{
+			int shapeSize = sizeof(btBvhTriangleMeshShape);
+			btAssert(shapeSize < MAX_SHAPE_SIZE);
+			return shapeSize;
+		}
+	case CAPSULE_SHAPE_PROXYTYPE:
+		{
+			int shapeSize = sizeof(btCapsuleShape);
+			btAssert(shapeSize < MAX_SHAPE_SIZE);
+			return shapeSize;
+		}
+
+	case CONVEX_HULL_SHAPE_PROXYTYPE:
+		{
+			int shapeSize = sizeof(btConvexHullShape);
+			btAssert(shapeSize < MAX_SHAPE_SIZE);
+			return shapeSize;
+		}
+
+	case COMPOUND_SHAPE_PROXYTYPE:
+		{
+			int shapeSize = sizeof(btCompoundShape);
+			btAssert(shapeSize < MAX_SHAPE_SIZE);
+			return shapeSize;
+		}
+	case STATIC_PLANE_PROXYTYPE:
+		{
+			int shapeSize = sizeof(btStaticPlaneShape);
+			btAssert(shapeSize < MAX_SHAPE_SIZE);
+			return shapeSize;
+		}
+
+	default:
+		btAssert(0);
+		//unsupported shapetype, please add here
+		return 0;
+	}
+}
+
+void dmaConvexVertexData (SpuConvexPolyhedronVertexData* convexVertexData, btConvexHullShape* convexShapeSPU)
+{
+	convexVertexData->gNumConvexPoints = convexShapeSPU->getNumPoints();
+	if (convexVertexData->gNumConvexPoints>MAX_NUM_SPU_CONVEX_POINTS)
+	{
+		btAssert(0);
+	//	spu_printf("SPU: Error: MAX_NUM_SPU_CONVEX_POINTS(%d) exceeded: %d\n",MAX_NUM_SPU_CONVEX_POINTS,convexVertexData->gNumConvexPoints);
+		return;
+	}
+			
+	register int dmaSize = convexVertexData->gNumConvexPoints*sizeof(btVector3);
+	ppu_address_t pointsPPU = (ppu_address_t) convexShapeSPU->getUnscaledPoints();
+	cellDmaGet(&convexVertexData->g_convexPointBuffer[0], pointsPPU  , dmaSize, DMA_TAG(2), 0, 0);
+}
+
+void dmaCollisionShape (void* collisionShapeLocation, ppu_address_t collisionShapePtr, uint32_t dmaTag, int shapeType)
+{
+	register int dmaSize = getShapeTypeSize(shapeType);
+	cellDmaGet(collisionShapeLocation, collisionShapePtr  , dmaSize, DMA_TAG(dmaTag), 0, 0);
+	//cellDmaGetReadOnly(collisionShapeLocation, collisionShapePtr  , dmaSize, DMA_TAG(dmaTag), 0, 0);
+	//cellDmaWaitTagStatusAll(DMA_MASK(dmaTag));
+}
+
+void dmaCompoundShapeInfo (CompoundShape_LocalStoreMemory* compoundShapeLocation, btCompoundShape* spuCompoundShape, uint32_t dmaTag)
+{
+	register int dmaSize;
+	register	ppu_address_t	dmaPpuAddress2;
+	int childShapeCount = spuCompoundShape->getNumChildShapes();
+	dmaSize = childShapeCount * sizeof(btCompoundShapeChild);
+	dmaPpuAddress2 = (ppu_address_t)spuCompoundShape->getChildList();
+	cellDmaGet(&compoundShapeLocation->gSubshapes[0], dmaPpuAddress2, dmaSize, DMA_TAG(dmaTag), 0, 0);
+}
+
+void dmaCompoundSubShapes (CompoundShape_LocalStoreMemory* compoundShapeLocation, btCompoundShape* spuCompoundShape, uint32_t dmaTag)
+{
+	int childShapeCount = spuCompoundShape->getNumChildShapes();
+	int i;
+	// DMA all the subshapes 
+	for ( i = 0; i < childShapeCount; ++i)
+	{
+		btCompoundShapeChild& childShape = compoundShapeLocation->gSubshapes[i];
+		dmaCollisionShape (&compoundShapeLocation->gSubshapeShape[i],(ppu_address_t)childShape.m_childShape, dmaTag, childShape.m_childShapeType);
+	}
+}
+
+
+void	spuWalkStacklessQuantizedTree(btNodeOverlapCallback* nodeCallback,unsigned short int* quantizedQueryAabbMin,unsigned short int* quantizedQueryAabbMax,const btQuantizedBvhNode* rootNode,int startNodeIndex,int endNodeIndex)
+{
+
+	int curIndex = startNodeIndex;
+	int walkIterations = 0;
+#ifdef BT_DEBUG
+	int subTreeSize = endNodeIndex - startNodeIndex;
+#endif
+
+	int escapeIndex;
+
+	unsigned int aabbOverlap, isLeafNode;
+
+	while (curIndex < endNodeIndex)
+	{
+		//catch bugs in tree data
+		btAssert (walkIterations < subTreeSize);
+
+		walkIterations++;
+		aabbOverlap = spuTestQuantizedAabbAgainstQuantizedAabb(quantizedQueryAabbMin,quantizedQueryAabbMax,rootNode->m_quantizedAabbMin,rootNode->m_quantizedAabbMax);
+		isLeafNode = rootNode->isLeafNode();
+
+		if (isLeafNode && aabbOverlap)
+		{
+			//printf("overlap with node %d\n",rootNode->getTriangleIndex());
+			nodeCallback->processNode(0,rootNode->getTriangleIndex());
+			//			spu_printf("SPU: overlap detected with triangleIndex:%d\n",rootNode->getTriangleIndex());
+		} 
+
+		if (aabbOverlap || isLeafNode)
+		{
+			rootNode++;
+			curIndex++;
+		} else
+		{
+			escapeIndex = rootNode->getEscapeIndex();
+			rootNode += escapeIndex;
+			curIndex += escapeIndex;
+		}
+	}
+
+}