2 Copyright (c) 2012 Advanced Micro Devices, Inc.
4 This software is provided 'as-is', without any express or implied warranty.
5 In no event will the authors be held liable for any damages arising from the use of this software.
6 Permission is granted to anyone to use this software for any purpose,
7 including commercial applications, and to alter it and redistribute it freely,
8 subject to the following restrictions:
10 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.
11 2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.
12 3. This notice may not be removed or altered from any source distribution.
14 //Originally written by Takahiro Harada
16 #ifndef __ADL_SOLVER_H
17 #define __ADL_SOLVER_H
19 #include "Bullet3OpenCL/ParallelPrimitives/b3OpenCLArray.h"
20 #include "b3GpuConstraint4.h"
22 #include "Bullet3Collision/NarrowPhaseCollision/shared/b3RigidBodyData.h"
23 #include "Bullet3Collision/NarrowPhaseCollision/b3Contact4.h"
25 #include "Bullet3OpenCL/ParallelPrimitives/b3PrefixScanCL.h"
26 #include "Bullet3OpenCL/ParallelPrimitives/b3RadixSort32CL.h"
27 #include "Bullet3OpenCL/ParallelPrimitives/b3BoundSearchCL.h"
29 #include "Bullet3OpenCL/Initialize/b3OpenCLUtils.h"
31 #define B3NEXTMULTIPLEOF(num, alignment) (((num) / (alignment) + (((num) % (alignment) == 0) ? 0 : 1)) * (alignment))
35 B3_SOLVER_N_SPLIT_X = 8, //16,//4,
36 B3_SOLVER_N_SPLIT_Y = 4, //16,//4,
37 B3_SOLVER_N_SPLIT_Z = 8, //,
38 B3_SOLVER_N_CELLS = B3_SOLVER_N_SPLIT_X * B3_SOLVER_N_SPLIT_Y * B3_SOLVER_N_SPLIT_Z,
39 B3_SOLVER_N_BATCHES = 8, //4,//8,//4,
40 B3_MAX_NUM_BATCHES = 128,
48 ConstraintCfg(float dt = 0.f) : m_positionDrift(0.005f), m_positionConstraintCoeff(0.2f), m_dt(dt), m_staticIdx(-1) {}
50 float m_positionDrift;
51 float m_positionConstraintCoeff;
53 bool m_enableParallelSolve;
54 float m_batchCellSize;
59 class b3Solver : public b3SolverBase
63 cl_device_id m_device;
64 cl_command_queue m_queue;
66 b3OpenCLArray<unsigned int>* m_numConstraints;
67 b3OpenCLArray<unsigned int>* m_offsets;
68 b3OpenCLArray<int> m_batchSizes;
71 cl_kernel m_batchingKernel;
72 cl_kernel m_batchingKernelNew;
73 cl_kernel m_solveContactKernel;
74 cl_kernel m_solveFrictionKernel;
75 cl_kernel m_contactToConstraintKernel;
76 cl_kernel m_setSortDataKernel;
77 cl_kernel m_reorderContactKernel;
78 cl_kernel m_copyConstraintKernel;
80 class b3RadixSort32CL* m_sort32;
81 class b3BoundSearchCL* m_search;
82 class b3PrefixScanCL* m_scan;
84 b3OpenCLArray<b3SortData>* m_sortDataBuffer;
85 b3OpenCLArray<b3Contact4>* m_contactBuffer2;
89 DYNAMIC_CONTACT_ALLOCATION_THRESHOLD = 2000000,
92 b3Solver(cl_context ctx, cl_device_id device, cl_command_queue queue, int pairCapacity);
96 void solveContactConstraint(const b3OpenCLArray<b3RigidBodyData>* bodyBuf, const b3OpenCLArray<b3InertiaData>* inertiaBuf,
97 b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n, int maxNumBatches);
99 void solveContactConstraintHost(b3OpenCLArray<b3RigidBodyData>* bodyBuf, b3OpenCLArray<b3InertiaData>* shapeBuf,
100 b3OpenCLArray<b3GpuConstraint4>* constraint, void* additionalData, int n, int maxNumBatches, b3AlignedObjectArray<int>* batchSizes);
102 void convertToConstraints(const b3OpenCLArray<b3RigidBodyData>* bodyBuf,
103 const b3OpenCLArray<b3InertiaData>* shapeBuf,
104 b3OpenCLArray<b3Contact4>* contactsIn, b3OpenCLArray<b3GpuConstraint4>* contactCOut, void* additionalData,
105 int nContacts, const ConstraintCfg& cfg);
107 void batchContacts(b3OpenCLArray<b3Contact4>* contacts, int nContacts, b3OpenCLArray<unsigned int>* n, b3OpenCLArray<unsigned int>* offsets, int staticIdx);
110 #endif //__ADL_SOLVER_H