change lcpp Lua preprocessor, to keep #defines and comments, remove empty lines
remove duplicate data in b3Contact4 (now in btContact4Data shared between CPU/C++ and OpenCL) OpenCL kernels use #include "Bullet3Collision/NarrowPhaseCollision/shared/b3Contact4Data.h" Increase number of batches back to 250 (from 50), need to fix this hard coded number (see https://github.com/erwincoumans/bullet3/issues/12) Work towards GJK/EPA, in addition to SAT/clipping (early on)
This commit is contained in:
erwincoumans
@@ -2,36 +2,71 @@
|
||||
static const char* solverUtilsCL= \
|
||||
"/*\n"
|
||||
"Copyright (c) 2013 Advanced Micro Devices, Inc. \n"
|
||||
"\n"
|
||||
"This software is provided 'as-is', without any express or implied warranty.\n"
|
||||
"In no event will the authors be held liable for any damages arising from the use of this software.\n"
|
||||
"Permission is granted to anyone to use this software for any purpose, \n"
|
||||
"including commercial applications, and to alter it and redistribute it freely, \n"
|
||||
"subject to the following restrictions:\n"
|
||||
"\n"
|
||||
"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.\n"
|
||||
"2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software.\n"
|
||||
"3. This notice may not be removed or altered from any source distribution.\n"
|
||||
"*/\n"
|
||||
"//Originally written by Erwin Coumans\n"
|
||||
"\n"
|
||||
"#ifndef B3_CONTACT4DATA_H\n"
|
||||
"#define B3_CONTACT4DATA_H\n"
|
||||
"#ifndef B3_FLOAT4_H\n"
|
||||
"#define B3_FLOAT4_H\n"
|
||||
"#ifndef B3_PLATFORM_DEFINITIONS_H\n"
|
||||
"#define B3_PLATFORM_DEFINITIONS_H\n"
|
||||
"struct MyTest\n"
|
||||
"{\n"
|
||||
" int bla;\n"
|
||||
"};\n"
|
||||
"#endif\n"
|
||||
"#ifdef __cplusplus\n"
|
||||
"#else//bla\n"
|
||||
" typedef float4 b3Float4;\n"
|
||||
"#endif \n"
|
||||
"#endif //B3_FLOAT4_H\n"
|
||||
"typedef struct b3Contact4Data b3Contact4Data_t;\n"
|
||||
"struct b3Contact4Data\n"
|
||||
"{\n"
|
||||
" b3Float4 m_worldPos[4];\n"
|
||||
"// b3Float4 m_localPosB[4];\n"
|
||||
" b3Float4 m_worldNormal; // w: m_nPoints\n"
|
||||
" unsigned short m_restituitionCoeffCmp;\n"
|
||||
" unsigned short m_frictionCoeffCmp;\n"
|
||||
" int m_batchIdx;\n"
|
||||
" int m_bodyAPtrAndSignBit;//x:m_bodyAPtr, y:m_bodyBPtr\n"
|
||||
" int m_bodyBPtrAndSignBit;\n"
|
||||
" int m_childIndexA;\n"
|
||||
" int m_childIndexB;\n"
|
||||
" int m_unused1;\n"
|
||||
" int m_unused2;\n"
|
||||
" b3Float4 m_localPosA;\n"
|
||||
"};\n"
|
||||
"inline int b3Contact4Data_getNumPoints(const struct b3Contact4Data* contact)\n"
|
||||
"{\n"
|
||||
" return (int)contact->m_worldNormal.w;\n"
|
||||
"};\n"
|
||||
"inline void b3Contact4Data_setNumPoints(struct b3Contact4Data* contact, int numPoints)\n"
|
||||
"{\n"
|
||||
" contact->m_worldNormal.w = (float)numPoints;\n"
|
||||
"};\n"
|
||||
"#endif //B3_CONTACT4DATA_H\n"
|
||||
"#pragma OPENCL EXTENSION cl_amd_printf : enable\n"
|
||||
"#pragma OPENCL EXTENSION cl_khr_local_int32_base_atomics : enable\n"
|
||||
"#pragma OPENCL EXTENSION cl_khr_global_int32_base_atomics : enable\n"
|
||||
"#pragma OPENCL EXTENSION cl_khr_local_int32_extended_atomics : enable\n"
|
||||
"#pragma OPENCL EXTENSION cl_khr_global_int32_extended_atomics : enable\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"#ifdef cl_ext_atomic_counters_32\n"
|
||||
"#pragma OPENCL EXTENSION cl_ext_atomic_counters_32 : enable\n"
|
||||
"#else\n"
|
||||
"#define counter32_t volatile global int*\n"
|
||||
"#endif\n"
|
||||
"\n"
|
||||
"typedef unsigned int u32;\n"
|
||||
"typedef unsigned short u16;\n"
|
||||
"typedef unsigned char u8;\n"
|
||||
"\n"
|
||||
"#define GET_GROUP_IDX get_group_id(0)\n"
|
||||
"#define GET_LOCAL_IDX get_local_id(0)\n"
|
||||
"#define GET_GLOBAL_IDX get_global_id(0)\n"
|
||||
@@ -45,22 +80,15 @@ static const char* solverUtilsCL= \
|
||||
"#define AtomAdd(x, value) atom_add(&(x), value)\n"
|
||||
"#define AtomCmpxhg(x, cmp, value) atom_cmpxchg( &(x), cmp, value )\n"
|
||||
"#define AtomXhg(x, value) atom_xchg ( &(x), value )\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"#define SELECT_UINT4( b, a, condition ) select( b,a,condition )\n"
|
||||
"\n"
|
||||
"#define make_float4 (float4)\n"
|
||||
"#define make_float2 (float2)\n"
|
||||
"#define make_uint4 (uint4)\n"
|
||||
"#define make_int4 (int4)\n"
|
||||
"#define make_uint2 (uint2)\n"
|
||||
"#define make_int2 (int2)\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"#define max2 max\n"
|
||||
"#define min2 min\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"///////////////////////////////////////\n"
|
||||
"// Vector\n"
|
||||
"///////////////////////////////////////\n"
|
||||
@@ -70,57 +98,47 @@ static const char* solverUtilsCL= \
|
||||
" return native_divide(numerator, denominator); \n"
|
||||
"// return numerator/denominator; \n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 fastDiv4(float4 numerator, float4 denominator)\n"
|
||||
"{\n"
|
||||
" return native_divide(numerator, denominator); \n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float fastSqrtf(float f2)\n"
|
||||
"{\n"
|
||||
" return native_sqrt(f2);\n"
|
||||
"// return sqrt(f2);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float fastRSqrt(float f2)\n"
|
||||
"{\n"
|
||||
" return native_rsqrt(f2);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float fastLength4(float4 v)\n"
|
||||
"{\n"
|
||||
" return fast_length(v);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 fastNormalize4(float4 v)\n"
|
||||
"{\n"
|
||||
" return fast_normalize(v);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float sqrtf(float a)\n"
|
||||
"{\n"
|
||||
"// return sqrt(a);\n"
|
||||
" return native_sqrt(a);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 cross3(float4 a1, float4 b1)\n"
|
||||
"{\n"
|
||||
"\n"
|
||||
" float4 a=make_float4(a1.xyz,0.f);\n"
|
||||
" float4 b=make_float4(b1.xyz,0.f);\n"
|
||||
" //float4 a=a1;\n"
|
||||
" //float4 b=b1;\n"
|
||||
" return cross(a,b);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float dot3F4(float4 a, float4 b)\n"
|
||||
"{\n"
|
||||
@@ -128,26 +146,22 @@ static const char* solverUtilsCL= \
|
||||
" float4 b1 = make_float4(b.xyz,0.f);\n"
|
||||
" return dot(a1, b1);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float length3(const float4 a)\n"
|
||||
"{\n"
|
||||
" return sqrtf(dot3F4(a,a));\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float dot4(const float4 a, const float4 b)\n"
|
||||
"{\n"
|
||||
" return dot( a, b );\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"// for height\n"
|
||||
"__inline\n"
|
||||
"float dot3w1(const float4 point, const float4 eqn)\n"
|
||||
"{\n"
|
||||
" return dot3F4(point,eqn) + eqn.w;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 normalize3(const float4 a)\n"
|
||||
"{\n"
|
||||
@@ -156,14 +170,12 @@ static const char* solverUtilsCL= \
|
||||
"// float length = sqrtf(dot3F4(a, a));\n"
|
||||
"// return 1.f/length * a;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 normalize4(const float4 a)\n"
|
||||
"{\n"
|
||||
" float length = sqrtf(dot4(a, a));\n"
|
||||
" return 1.f/length * a;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 createEquation(const float4 a, const float4 b, const float4 c)\n"
|
||||
"{\n"
|
||||
@@ -174,34 +186,25 @@ static const char* solverUtilsCL= \
|
||||
" eqn.w = -dot3F4(eqn,a);\n"
|
||||
" return eqn;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"///////////////////////////////////////\n"
|
||||
"// Matrix3x3\n"
|
||||
"///////////////////////////////////////\n"
|
||||
"\n"
|
||||
"typedef struct\n"
|
||||
"{\n"
|
||||
" float4 m_row[3];\n"
|
||||
"}Matrix3x3;\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Matrix3x3 mtZero();\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Matrix3x3 mtIdentity();\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Matrix3x3 mtTranspose(Matrix3x3 m);\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Matrix3x3 mtMul(Matrix3x3 a, Matrix3x3 b);\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 mtMul1(Matrix3x3 a, float4 b);\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 mtMul3(float4 a, Matrix3x3 b);\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Matrix3x3 mtZero()\n"
|
||||
"{\n"
|
||||
@@ -211,7 +214,6 @@ static const char* solverUtilsCL= \
|
||||
" m.m_row[2] = (float4)(0.f);\n"
|
||||
" return m;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Matrix3x3 mtIdentity()\n"
|
||||
"{\n"
|
||||
@@ -221,7 +223,6 @@ static const char* solverUtilsCL= \
|
||||
" m.m_row[2] = (float4)(0,0,1,0);\n"
|
||||
" return m;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Matrix3x3 mtTranspose(Matrix3x3 m)\n"
|
||||
"{\n"
|
||||
@@ -231,7 +232,6 @@ static const char* solverUtilsCL= \
|
||||
" out.m_row[2] = (float4)(m.m_row[0].z, m.m_row[1].z, m.m_row[2].z, 0.f);\n"
|
||||
" return out;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Matrix3x3 mtMul(Matrix3x3 a, Matrix3x3 b)\n"
|
||||
"{\n"
|
||||
@@ -252,7 +252,6 @@ static const char* solverUtilsCL= \
|
||||
" }\n"
|
||||
" return ans;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 mtMul1(Matrix3x3 a, float4 b)\n"
|
||||
"{\n"
|
||||
@@ -263,43 +262,30 @@ static const char* solverUtilsCL= \
|
||||
" ans.w = 0.f;\n"
|
||||
" return ans;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 mtMul3(float4 a, Matrix3x3 b)\n"
|
||||
"{\n"
|
||||
" float4 colx = make_float4(b.m_row[0].x, b.m_row[1].x, b.m_row[2].x, 0);\n"
|
||||
" float4 coly = make_float4(b.m_row[0].y, b.m_row[1].y, b.m_row[2].y, 0);\n"
|
||||
" float4 colz = make_float4(b.m_row[0].z, b.m_row[1].z, b.m_row[2].z, 0);\n"
|
||||
"\n"
|
||||
" float4 ans;\n"
|
||||
" ans.x = dot3F4( a, colx );\n"
|
||||
" ans.y = dot3F4( a, coly );\n"
|
||||
" ans.z = dot3F4( a, colz );\n"
|
||||
" return ans;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"///////////////////////////////////////\n"
|
||||
"// Quaternion\n"
|
||||
"///////////////////////////////////////\n"
|
||||
"\n"
|
||||
"typedef float4 Quaternion;\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Quaternion qtMul(Quaternion a, Quaternion b);\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Quaternion qtNormalize(Quaternion in);\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 qtRotate(Quaternion q, float4 vec);\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Quaternion qtInvert(Quaternion q);\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Quaternion qtMul(Quaternion a, Quaternion b)\n"
|
||||
"{\n"
|
||||
@@ -310,7 +296,6 @@ static const char* solverUtilsCL= \
|
||||
" ans.w = a.w*b.w - dot3F4(a, b);\n"
|
||||
" return ans;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Quaternion qtNormalize(Quaternion in)\n"
|
||||
"{\n"
|
||||
@@ -327,43 +312,33 @@ static const char* solverUtilsCL= \
|
||||
" float4 out = qtMul(qtMul(q,vcpy),qInv);\n"
|
||||
" return out;\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"Quaternion qtInvert(Quaternion q)\n"
|
||||
"{\n"
|
||||
" return (Quaternion)(-q.xyz, q.w);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__inline\n"
|
||||
"float4 qtInvRotate(const Quaternion q, float4 vec)\n"
|
||||
"{\n"
|
||||
" return qtRotate( qtInvert( q ), vec );\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"#define WG_SIZE 64\n"
|
||||
"\n"
|
||||
"typedef struct\n"
|
||||
"{\n"
|
||||
" float4 m_pos;\n"
|
||||
" Quaternion m_quat;\n"
|
||||
" float4 m_linVel;\n"
|
||||
" float4 m_angVel;\n"
|
||||
"\n"
|
||||
" u32 m_shapeIdx;\n"
|
||||
" float m_invMass;\n"
|
||||
" float m_restituitionCoeff;\n"
|
||||
" float m_frictionCoeff;\n"
|
||||
"} Body;\n"
|
||||
"\n"
|
||||
"typedef struct\n"
|
||||
"{\n"
|
||||
" Matrix3x3 m_invInertia;\n"
|
||||
" Matrix3x3 m_initInvInertia;\n"
|
||||
"} Shape;\n"
|
||||
"\n"
|
||||
"typedef struct\n"
|
||||
"{\n"
|
||||
" float4 m_linear;\n"
|
||||
@@ -372,35 +347,14 @@ static const char* solverUtilsCL= \
|
||||
" float m_jacCoeffInv[4];\n"
|
||||
" float m_b[4];\n"
|
||||
" float m_appliedRambdaDt[4];\n"
|
||||
"\n"
|
||||
" float m_fJacCoeffInv[2]; \n"
|
||||
" float m_fAppliedRambdaDt[2]; \n"
|
||||
"\n"
|
||||
" u32 m_bodyA;\n"
|
||||
" u32 m_bodyB;\n"
|
||||
" int m_batchIdx;\n"
|
||||
" u32 m_paddings;\n"
|
||||
"} Constraint4;\n"
|
||||
"\n"
|
||||
"typedef struct\n"
|
||||
"{\n"
|
||||
" float4 m_worldPos[4];\n"
|
||||
" float4 m_worldNormal;\n"
|
||||
" u32 m_coeffs;\n"
|
||||
" int m_batchIdx;\n"
|
||||
"\n"
|
||||
" int m_bodyAPtrAndSignBit;\n"
|
||||
" int m_bodyBPtrAndSignBit;\n"
|
||||
"\n"
|
||||
" int m_childIndexA;\n"
|
||||
" int m_childIndexB;\n"
|
||||
" int m_unused1;\n"
|
||||
" int m_unused2;\n"
|
||||
"\n"
|
||||
"} Contact4;\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"__kernel void CountBodiesKernel(__global Contact4* manifoldPtr, __global unsigned int* bodyCount, __global int2* contactConstraintOffsets, int numContactManifolds, int fixedBodyIndex)\n"
|
||||
"__kernel void CountBodiesKernel(__global struct b3Contact4Data* manifoldPtr, __global unsigned int* bodyCount, __global int2* contactConstraintOffsets, int numContactManifolds, int fixedBodyIndex)\n"
|
||||
"{\n"
|
||||
" int i = GET_GLOBAL_IDX;\n"
|
||||
" \n"
|
||||
@@ -423,7 +377,6 @@ static const char* solverUtilsCL= \
|
||||
" } \n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"__kernel void ClearVelocitiesKernel(__global float4* linearVelocities,__global float4* angularVelocities, int numSplitBodies)\n"
|
||||
"{\n"
|
||||
" int i = GET_GLOBAL_IDX;\n"
|
||||
@@ -434,8 +387,6 @@ static const char* solverUtilsCL= \
|
||||
" angularVelocities[i] = make_float4(0);\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"__kernel void AverageVelocitiesKernel(__global Body* gBodies,__global int* offsetSplitBodies,__global const unsigned int* bodyCount,\n"
|
||||
"__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities, int numBodies)\n"
|
||||
"{\n"
|
||||
@@ -465,23 +416,16 @@ static const char* solverUtilsCL= \
|
||||
" }//bodies[i].m_invMass\n"
|
||||
" }//i<numBodies\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"void setLinearAndAngular( float4 n, float4 r0, float4 r1, float4* linear, float4* angular0, float4* angular1)\n"
|
||||
"{\n"
|
||||
" *linear = make_float4(-n.xyz,0.f);\n"
|
||||
" *angular0 = -cross3(r0, n);\n"
|
||||
" *angular1 = cross3(r1, n);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"float calcRelVel( float4 l0, float4 l1, float4 a0, float4 a1, float4 linVel0, float4 angVel0, float4 linVel1, float4 angVel1 )\n"
|
||||
"{\n"
|
||||
" return dot3F4(l0, linVel0) + dot3F4(a0, angVel0) + dot3F4(l1, linVel1) + dot3F4(a1, angVel1);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"float calcJacCoeff(const float4 linear0, const float4 linear1, const float4 angular0, const float4 angular1,\n"
|
||||
" float invMass0, const Matrix3x3* invInertia0, float invMass1, const Matrix3x3* invInertia1, float countA, float countB)\n"
|
||||
"{\n"
|
||||
@@ -492,8 +436,6 @@ static const char* solverUtilsCL= \
|
||||
" float jmj3 = dot3F4(mtMul3(angular1,*invInertia1), angular1);\n"
|
||||
" return -1.f/((jmj0+jmj1)*countA+(jmj2+jmj3)*countB);\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"void btPlaneSpace1 (float4 n, float4* p, float4* q);\n"
|
||||
" void btPlaneSpace1 (float4 n, float4* p, float4* q)\n"
|
||||
"{\n"
|
||||
@@ -522,11 +464,6 @@ static const char* solverUtilsCL= \
|
||||
" q[0].z = a*k;\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"void solveContact(__global Constraint4* cs,\n"
|
||||
" float4 posA, float4* linVelA, float4* angVelA, float invMassA, Matrix3x3 invInertiaA,\n"
|
||||
" float4 posB, float4* linVelB, float4* angVelB, float invMassB, Matrix3x3 invInertiaB,\n"
|
||||
@@ -534,22 +471,17 @@ static const char* solverUtilsCL= \
|
||||
"{\n"
|
||||
" float minRambdaDt = 0;\n"
|
||||
" float maxRambdaDt = FLT_MAX;\n"
|
||||
"\n"
|
||||
" for(int ic=0; ic<4; ic++)\n"
|
||||
" {\n"
|
||||
" if( cs->m_jacCoeffInv[ic] == 0.f ) continue;\n"
|
||||
"\n"
|
||||
" float4 angular0, angular1, linear;\n"
|
||||
" float4 r0 = cs->m_worldPos[ic] - posA;\n"
|
||||
" float4 r1 = cs->m_worldPos[ic] - posB;\n"
|
||||
" setLinearAndAngular( -cs->m_linear, r0, r1, &linear, &angular0, &angular1 );\n"
|
||||
" \n"
|
||||
"\n"
|
||||
"\n"
|
||||
" float rambdaDt = calcRelVel( cs->m_linear, -cs->m_linear, angular0, angular1, \n"
|
||||
" *linVelA+*dLinVelA, *angVelA+*dAngVelA, *linVelB+*dLinVelB, *angVelB+*dAngVelB ) + cs->m_b[ic];\n"
|
||||
" rambdaDt *= cs->m_jacCoeffInv[ic];\n"
|
||||
"\n"
|
||||
" \n"
|
||||
" {\n"
|
||||
" float prevSum = cs->m_appliedRambdaDt[ic];\n"
|
||||
@@ -560,13 +492,11 @@ static const char* solverUtilsCL= \
|
||||
" rambdaDt = updated - prevSum;\n"
|
||||
" cs->m_appliedRambdaDt[ic] = updated;\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
" \n"
|
||||
" float4 linImp0 = invMassA*linear*rambdaDt;\n"
|
||||
" float4 linImp1 = invMassB*(-linear)*rambdaDt;\n"
|
||||
" float4 angImp0 = mtMul1(invInertiaA, angular0)*rambdaDt;\n"
|
||||
" float4 angImp1 = mtMul1(invInertiaB, angular1)*rambdaDt;\n"
|
||||
"\n"
|
||||
" \n"
|
||||
" if (invMassA)\n"
|
||||
" {\n"
|
||||
@@ -580,32 +510,24 @@ static const char* solverUtilsCL= \
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"// solveContactConstraint( gBodies, gShapes, &gConstraints[i] ,contactConstraintOffsets,offsetSplitBodies, deltaLinearVelocities, deltaAngularVelocities);\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"void solveContactConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs, \n"
|
||||
"__global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,\n"
|
||||
"__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities)\n"
|
||||
"{\n"
|
||||
"\n"
|
||||
" //float frictionCoeff = ldsCs[0].m_linear.w;\n"
|
||||
" int aIdx = ldsCs[0].m_bodyA;\n"
|
||||
" int bIdx = ldsCs[0].m_bodyB;\n"
|
||||
"\n"
|
||||
" float4 posA = gBodies[aIdx].m_pos;\n"
|
||||
" float4 linVelA = gBodies[aIdx].m_linVel;\n"
|
||||
" float4 angVelA = gBodies[aIdx].m_angVel;\n"
|
||||
" float invMassA = gBodies[aIdx].m_invMass;\n"
|
||||
" Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
|
||||
"\n"
|
||||
" float4 posB = gBodies[bIdx].m_pos;\n"
|
||||
" float4 linVelB = gBodies[bIdx].m_linVel;\n"
|
||||
" float4 angVelB = gBodies[bIdx].m_angVel;\n"
|
||||
" float invMassB = gBodies[bIdx].m_invMass;\n"
|
||||
" Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
|
||||
"\n"
|
||||
" \n"
|
||||
" float4 dLinVelA = make_float4(0,0,0,0);\n"
|
||||
" float4 dAngVelA = make_float4(0,0,0,0);\n"
|
||||
@@ -621,20 +543,16 @@ static const char* solverUtilsCL= \
|
||||
" dLinVelA = deltaLinearVelocities[splitIndexA];\n"
|
||||
" dAngVelA = deltaAngularVelocities[splitIndexA];\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
" int bodyOffsetB = offsetSplitBodies[bIdx];\n"
|
||||
" int constraintOffsetB = contactConstraintOffsets[0].y;\n"
|
||||
" int splitIndexB= bodyOffsetB+constraintOffsetB;\n"
|
||||
"\n"
|
||||
" if (invMassB)\n"
|
||||
" {\n"
|
||||
" dLinVelB = deltaLinearVelocities[splitIndexB];\n"
|
||||
" dAngVelB = deltaAngularVelocities[splitIndexB];\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
" solveContact( ldsCs, posA, &linVelA, &angVelA, invMassA, invInertiaA,\n"
|
||||
" posB, &linVelB, &angVelB, invMassB, invInertiaB ,&dLinVelA, &dAngVelA, &dLinVelB, &dAngVelB);\n"
|
||||
"\n"
|
||||
" if (invMassA)\n"
|
||||
" {\n"
|
||||
" deltaLinearVelocities[splitIndexA] = dLinVelA;\n"
|
||||
@@ -645,10 +563,7 @@ static const char* solverUtilsCL= \
|
||||
" deltaLinearVelocities[splitIndexB] = dLinVelB;\n"
|
||||
" deltaAngularVelocities[splitIndexB] = dAngVelB;\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"__kernel void SolveContactJacobiKernel(__global Constraint4* gConstraints, __global Body* gBodies, __global Shape* gShapes ,\n"
|
||||
"__global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,__global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities,\n"
|
||||
"float deltaTime, float positionDrift, float positionConstraintCoeff, int fixedBodyIndex, int numManifolds\n"
|
||||
@@ -660,10 +575,6 @@ static const char* solverUtilsCL= \
|
||||
" solveContactConstraint( gBodies, gShapes, &gConstraints[i] ,&contactConstraintOffsets[i],offsetSplitBodies, deltaLinearVelocities, deltaAngularVelocities);\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"void solveFrictionConstraint(__global Body* gBodies, __global Shape* gShapes, __global Constraint4* ldsCs,\n"
|
||||
" __global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,\n"
|
||||
" __global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities)\n"
|
||||
@@ -671,21 +582,17 @@ static const char* solverUtilsCL= \
|
||||
" float frictionCoeff = 0.7f;//ldsCs[0].m_linear.w;\n"
|
||||
" int aIdx = ldsCs[0].m_bodyA;\n"
|
||||
" int bIdx = ldsCs[0].m_bodyB;\n"
|
||||
"\n"
|
||||
"\n"
|
||||
" float4 posA = gBodies[aIdx].m_pos;\n"
|
||||
" float4 linVelA = gBodies[aIdx].m_linVel;\n"
|
||||
" float4 angVelA = gBodies[aIdx].m_angVel;\n"
|
||||
" float invMassA = gBodies[aIdx].m_invMass;\n"
|
||||
" Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
|
||||
"\n"
|
||||
" float4 posB = gBodies[bIdx].m_pos;\n"
|
||||
" float4 linVelB = gBodies[bIdx].m_linVel;\n"
|
||||
" float4 angVelB = gBodies[bIdx].m_angVel;\n"
|
||||
" float invMassB = gBodies[bIdx].m_invMass;\n"
|
||||
" Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
|
||||
" \n"
|
||||
"\n"
|
||||
" float4 dLinVelA = make_float4(0,0,0,0);\n"
|
||||
" float4 dAngVelA = make_float4(0,0,0,0);\n"
|
||||
" float4 dLinVelB = make_float4(0,0,0,0);\n"
|
||||
@@ -700,24 +607,17 @@ static const char* solverUtilsCL= \
|
||||
" dLinVelA = deltaLinearVelocities[splitIndexA];\n"
|
||||
" dAngVelA = deltaAngularVelocities[splitIndexA];\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
" int bodyOffsetB = offsetSplitBodies[bIdx];\n"
|
||||
" int constraintOffsetB = contactConstraintOffsets[0].y;\n"
|
||||
" int splitIndexB= bodyOffsetB+constraintOffsetB;\n"
|
||||
"\n"
|
||||
" if (invMassB)\n"
|
||||
" {\n"
|
||||
" dLinVelB = deltaLinearVelocities[splitIndexB];\n"
|
||||
" dAngVelB = deltaAngularVelocities[splitIndexB];\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
" {\n"
|
||||
" float maxRambdaDt[4] = {FLT_MAX,FLT_MAX,FLT_MAX,FLT_MAX};\n"
|
||||
" float minRambdaDt[4] = {0.f,0.f,0.f,0.f};\n"
|
||||
"\n"
|
||||
" float sum = 0;\n"
|
||||
" for(int j=0; j<4; j++)\n"
|
||||
" {\n"
|
||||
@@ -729,7 +629,6 @@ static const char* solverUtilsCL= \
|
||||
" maxRambdaDt[j] = frictionCoeff*sum;\n"
|
||||
" minRambdaDt[j] = -maxRambdaDt[j];\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
" \n"
|
||||
"// solveFriction( ldsCs, posA, &linVelA, &angVelA, invMassA, invInertiaA,\n"
|
||||
"// posB, &linVelB, &angVelB, invMassB, invInertiaB, maxRambdaDt, minRambdaDt );\n"
|
||||
@@ -789,11 +688,9 @@ static const char* solverUtilsCL= \
|
||||
" }\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
" \n"
|
||||
" \n"
|
||||
" }\n"
|
||||
"\n"
|
||||
" if (invMassA)\n"
|
||||
" {\n"
|
||||
" deltaLinearVelocities[splitIndexA] = dLinVelA;\n"
|
||||
@@ -805,10 +702,7 @@ static const char* solverUtilsCL= \
|
||||
" deltaAngularVelocities[splitIndexB] = dAngVelB;\n"
|
||||
" }\n"
|
||||
" \n"
|
||||
"\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"__kernel void SolveFrictionJacobiKernel(__global Constraint4* gConstraints, __global Body* gBodies, __global Shape* gShapes ,\n"
|
||||
" __global int2* contactConstraintOffsets,__global unsigned int* offsetSplitBodies,\n"
|
||||
" __global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities,\n"
|
||||
@@ -821,8 +715,6 @@ static const char* solverUtilsCL= \
|
||||
" solveFrictionConstraint( gBodies, gShapes, &gConstraints[i] ,&contactConstraintOffsets[i],offsetSplitBodies, deltaLinearVelocities, deltaAngularVelocities);\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"__kernel void UpdateBodyVelocitiesKernel(__global Body* gBodies,__global int* offsetSplitBodies,__global const unsigned int* bodyCount,\n"
|
||||
" __global float4* deltaLinearVelocities, __global float4* deltaAngularVelocities, int numBodies)\n"
|
||||
"{\n"
|
||||
@@ -841,85 +733,68 @@ static const char* solverUtilsCL= \
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"void setConstraint4( const float4 posA, const float4 linVelA, const float4 angVelA, float invMassA, const Matrix3x3 invInertiaA,\n"
|
||||
" const float4 posB, const float4 linVelB, const float4 angVelB, float invMassB, const Matrix3x3 invInertiaB, \n"
|
||||
" __global Contact4* src, float dt, float positionDrift, float positionConstraintCoeff,float countA, float countB,\n"
|
||||
" __global struct b3Contact4Data* src, float dt, float positionDrift, float positionConstraintCoeff,float countA, float countB,\n"
|
||||
" Constraint4* dstC )\n"
|
||||
"{\n"
|
||||
" dstC->m_bodyA = abs(src->m_bodyAPtrAndSignBit);\n"
|
||||
" dstC->m_bodyB = abs(src->m_bodyBPtrAndSignBit);\n"
|
||||
"\n"
|
||||
" float dtInv = 1.f/dt;\n"
|
||||
" for(int ic=0; ic<4; ic++)\n"
|
||||
" {\n"
|
||||
" dstC->m_appliedRambdaDt[ic] = 0.f;\n"
|
||||
" }\n"
|
||||
" dstC->m_fJacCoeffInv[0] = dstC->m_fJacCoeffInv[1] = 0.f;\n"
|
||||
"\n"
|
||||
"\n"
|
||||
" dstC->m_linear = -src->m_worldNormal;\n"
|
||||
" dstC->m_linear.w = 0.7f ;//src->getFrictionCoeff() );\n"
|
||||
" for(int ic=0; ic<4; ic++)\n"
|
||||
" {\n"
|
||||
" float4 r0 = src->m_worldPos[ic] - posA;\n"
|
||||
" float4 r1 = src->m_worldPos[ic] - posB;\n"
|
||||
"\n"
|
||||
" if( ic >= src->m_worldNormal.w )//npoints\n"
|
||||
" {\n"
|
||||
" dstC->m_jacCoeffInv[ic] = 0.f;\n"
|
||||
" continue;\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
" float relVelN;\n"
|
||||
" {\n"
|
||||
" float4 linear, angular0, angular1;\n"
|
||||
" setLinearAndAngular(src->m_worldNormal, r0, r1, &linear, &angular0, &angular1);\n"
|
||||
"\n"
|
||||
" dstC->m_jacCoeffInv[ic] = calcJacCoeff(linear, -linear, angular0, angular1,\n"
|
||||
" invMassA, &invInertiaA, invMassB, &invInertiaB , countA, countB);\n"
|
||||
"\n"
|
||||
" relVelN = calcRelVel(linear, -linear, angular0, angular1,\n"
|
||||
" linVelA, angVelA, linVelB, angVelB);\n"
|
||||
"\n"
|
||||
" float e = 0.f;//src->getRestituitionCoeff();\n"
|
||||
" if( relVelN*relVelN < 0.004f ) e = 0.f;\n"
|
||||
"\n"
|
||||
" dstC->m_b[ic] = e*relVelN;\n"
|
||||
" //float penetration = src->m_worldPos[ic].w;\n"
|
||||
" dstC->m_b[ic] += (src->m_worldPos[ic].w + positionDrift)*positionConstraintCoeff*dtInv;\n"
|
||||
" dstC->m_appliedRambdaDt[ic] = 0.f;\n"
|
||||
" }\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
" if( src->m_worldNormal.w > 0 )//npoints\n"
|
||||
" { // prepare friction\n"
|
||||
" float4 center = make_float4(0.f);\n"
|
||||
" for(int i=0; i<src->m_worldNormal.w; i++) \n"
|
||||
" center += src->m_worldPos[i];\n"
|
||||
" center /= (float)src->m_worldNormal.w;\n"
|
||||
"\n"
|
||||
" float4 tangent[2];\n"
|
||||
" btPlaneSpace1(src->m_worldNormal,&tangent[0],&tangent[1]);\n"
|
||||
" \n"
|
||||
" float4 r[2];\n"
|
||||
" r[0] = center - posA;\n"
|
||||
" r[1] = center - posB;\n"
|
||||
"\n"
|
||||
" for(int i=0; i<2; i++)\n"
|
||||
" {\n"
|
||||
" float4 linear, angular0, angular1;\n"
|
||||
" setLinearAndAngular(tangent[i], r[0], r[1], &linear, &angular0, &angular1);\n"
|
||||
"\n"
|
||||
" dstC->m_fJacCoeffInv[i] = calcJacCoeff(linear, -linear, angular0, angular1,\n"
|
||||
" invMassA, &invInertiaA, invMassB, &invInertiaB ,countA, countB);\n"
|
||||
" dstC->m_fAppliedRambdaDt[i] = 0.f;\n"
|
||||
" }\n"
|
||||
" dstC->m_center = center;\n"
|
||||
" }\n"
|
||||
"\n"
|
||||
" for(int i=0; i<4; i++)\n"
|
||||
" {\n"
|
||||
" if( i<src->m_worldNormal.w )\n"
|
||||
@@ -932,11 +807,9 @@ static const char* solverUtilsCL= \
|
||||
" }\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
"\n"
|
||||
"\n"
|
||||
"__kernel\n"
|
||||
"__attribute__((reqd_work_group_size(WG_SIZE,1,1)))\n"
|
||||
"void ContactToConstraintSplitKernel(__global const Contact4* gContact, __global const Body* gBodies, __global const Shape* gShapes, __global Constraint4* gConstraintOut, \n"
|
||||
"void ContactToConstraintSplitKernel(__global const struct b3Contact4Data* gContact, __global const Body* gBodies, __global const Shape* gShapes, __global Constraint4* gConstraintOut, \n"
|
||||
"__global const unsigned int* bodyCount,\n"
|
||||
"int nContacts,\n"
|
||||
"float dt,\n"
|
||||
@@ -950,30 +823,24 @@ static const char* solverUtilsCL= \
|
||||
" {\n"
|
||||
" int aIdx = abs(gContact[gIdx].m_bodyAPtrAndSignBit);\n"
|
||||
" int bIdx = abs(gContact[gIdx].m_bodyBPtrAndSignBit);\n"
|
||||
"\n"
|
||||
" float4 posA = gBodies[aIdx].m_pos;\n"
|
||||
" float4 linVelA = gBodies[aIdx].m_linVel;\n"
|
||||
" float4 angVelA = gBodies[aIdx].m_angVel;\n"
|
||||
" float invMassA = gBodies[aIdx].m_invMass;\n"
|
||||
" Matrix3x3 invInertiaA = gShapes[aIdx].m_invInertia;\n"
|
||||
"\n"
|
||||
" float4 posB = gBodies[bIdx].m_pos;\n"
|
||||
" float4 linVelB = gBodies[bIdx].m_linVel;\n"
|
||||
" float4 angVelB = gBodies[bIdx].m_angVel;\n"
|
||||
" float invMassB = gBodies[bIdx].m_invMass;\n"
|
||||
" Matrix3x3 invInertiaB = gShapes[bIdx].m_invInertia;\n"
|
||||
"\n"
|
||||
" Constraint4 cs;\n"
|
||||
"\n"
|
||||
" float countA = invMassA ? (float)bodyCount[aIdx] : 1;\n"
|
||||
" float countB = invMassB ? (float)bodyCount[bIdx] : 1;\n"
|
||||
"\n"
|
||||
" setConstraint4( posA, linVelA, angVelA, invMassA, invInertiaA, posB, linVelB, angVelB, invMassB, invInertiaB,\n"
|
||||
" &gContact[gIdx], dt, positionDrift, positionConstraintCoeff,countA,countB,\n"
|
||||
" &cs );\n"
|
||||
" \n"
|
||||
" cs.m_batchIdx = gContact[gIdx].m_batchIdx;\n"
|
||||
"\n"
|
||||
" gConstraintOut[gIdx] = cs;\n"
|
||||
" }\n"
|
||||
"}\n"
|
||||
|
||||
Reference in New Issue
Block a user