1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
|
// This code contains NVIDIA Confidential Information and is disclosed to you
// under a form of NVIDIA software license agreement provided separately to you.
//
// Notice
// NVIDIA Corporation and its licensors retain all intellectual property and
// proprietary rights in and to this software and related documentation and
// any modifications thereto. Any use, reproduction, disclosure, or
// distribution of this software and related documentation without an express
// license agreement from NVIDIA Corporation is strictly prohibited.
//
// ALL NVIDIA DESIGN SPECIFICATIONS, CODE ARE PROVIDED "AS IS.". NVIDIA MAKES
// NO WARRANTIES, EXPRESSED, IMPLIED, STATUTORY, OR OTHERWISE WITH RESPECT TO
// THE MATERIALS, AND EXPRESSLY DISCLAIMS ALL IMPLIED WARRANTIES OF NONINFRINGEMENT,
// MERCHANTABILITY, AND FITNESS FOR A PARTICULAR PURPOSE.
//
// Information and code furnished is believed to be accurate and reliable.
// However, NVIDIA Corporation assumes no responsibility for the consequences of use of such
// information or for any infringement of patents or other rights of third parties that may
// result from its use. No license is granted by implication or otherwise under any patent
// or patent rights of NVIDIA Corporation. Details are subject to change without notice.
// This code supersedes and replaces all information previously supplied.
// NVIDIA Corporation products are not authorized for use as critical
// components in life support devices or systems without express written approval of
// NVIDIA Corporation.
//
// Copyright (c) 2008-2016 NVIDIA Corporation. All rights reserved.
// Copyright (c) 2004-2008 AGEIA Technologies, Inc. All rights reserved.
// Copyright (c) 2001-2004 NovodeX AG. All rights reserved.
#ifndef DY_SOLVERCOREGENERAL_H
#define DY_SOLVERCOREGENERAL_H
#include "DySolverCore.h"
#include "DySolverConstraintDesc.h"
namespace physx
{
namespace Dy
{
struct FsData;
inline void BusyWaitState(volatile PxU32* state, const PxU32 requiredState)
{
while(requiredState != *state );
}
inline void WaitBodyRequiredState(PxU32* state, const PxU32 requiredState)
{
if(*state != requiredState)
{
BusyWaitState(state, requiredState);
}
}
inline void BusyWaitStates(volatile PxU32* stateA, volatile PxU32* stateB, const PxU32 requiredStateA, const PxU32 requiredStateB)
{
while(*stateA != requiredStateA);
while(*stateB != requiredStateB);
}
PX_FORCE_INLINE void WaitBodyABodyBRequiredState(const PxSolverConstraintDesc& desc, const PxI32 iterationA, const PxI32 iterationB)
{
PxSolverBody* PX_RESTRICT pBodyA = desc.bodyA;
PxSolverBody* PX_RESTRICT pBodyB = desc.bodyB;
const PxU32 requiredProgressA=(desc.bodyASolverProgress == 0xFFFF) ? 0xFFFF : PxU32(desc.bodyASolverProgress + iterationA * pBodyA->maxSolverNormalProgress + iterationB * pBodyA->maxSolverFrictionProgress);
const PxU32 requiredProgressB=(desc.bodyBSolverProgress == 0xFFFF) ? 0xFFFF : PxU32(desc.bodyBSolverProgress + iterationA * pBodyB->maxSolverNormalProgress + iterationB * pBodyB->maxSolverFrictionProgress);
PX_ASSERT(requiredProgressA!=0xFFFFFFFF || requiredProgressB!=0xFFFFFFFF);
const PxU32 solverProgressA = pBodyA->solverProgress;
const PxU32 solverProgressB = pBodyB->solverProgress;
if(solverProgressA != requiredProgressA || solverProgressB != requiredProgressB)
{
BusyWaitStates(&pBodyA->solverProgress, &pBodyB->solverProgress, requiredProgressA, requiredProgressB);
}
}
PX_FORCE_INLINE void IncrementBodyProgress(const PxSolverConstraintDesc& desc)
{
PxSolverBody* PX_RESTRICT pBodyA = desc.bodyA;
PxSolverBody* PX_RESTRICT pBodyB = desc.bodyB;
const PxU32 maxProgressA = pBodyA->maxSolverNormalProgress;
const PxU32 maxProgressB = pBodyB->maxSolverNormalProgress;
//NB - this approach removes the need for an imul (which is a non-pipeline instruction on PPC chips)
const PxU32 requiredProgressA=(maxProgressA == 0xFFFF) ? 0xFFFF : pBodyA->solverProgress + 1;
const PxU32 requiredProgressB=(maxProgressB == 0xFFFF) ? 0xFFFF : pBodyB->solverProgress + 1;
volatile PxU32* solveProgressA = &pBodyA->solverProgress;
volatile PxU32* solveProgressB = &pBodyB->solverProgress;
*solveProgressA=requiredProgressA;
*solveProgressB=requiredProgressB;
}
class BatchIterator
{
public:
PxConstraintBatchHeader* constraintBatchHeaders;
PxU32 mSize;
PxU32 mCurrentIndex;
BatchIterator(PxConstraintBatchHeader* _constraintBatchHeaders, PxU32 size) : constraintBatchHeaders(_constraintBatchHeaders),
mSize(size), mCurrentIndex(0)
{
}
PX_FORCE_INLINE const PxConstraintBatchHeader& GetCurrentHeader(const PxU32 constraintIndex)
{
PxU32 currentIndex = mCurrentIndex;
while((constraintIndex - constraintBatchHeaders[currentIndex].mStartIndex) >= constraintBatchHeaders[currentIndex].mStride)
currentIndex = (currentIndex + 1)%mSize;
Ps::prefetchLine(&constraintBatchHeaders[currentIndex], 128);
mCurrentIndex = currentIndex;
return constraintBatchHeaders[currentIndex];
}
private:
BatchIterator& operator=(const BatchIterator&);
};
template<bool bWaitIncrement>
void SolveBlockParallel (PxSolverConstraintDesc* PX_RESTRICT constraintList, const PxI32 batchCount, const PxI32 index,
const PxI32 headerCount, SolverContext& cache, BatchIterator& iterator,
SolveBlockMethod solveTable[], const PxI32 normalIteration, const PxI32 frictionIteration,
const PxI32 iteration
)
{
const PxI32 indA = index - (iteration * headerCount);
const PxConstraintBatchHeader* PX_RESTRICT headers = iterator.constraintBatchHeaders;
const PxI32 endIndex = indA + batchCount;
for(PxI32 i = indA; i < endIndex; ++i)
{
const PxConstraintBatchHeader& header = headers[i];
const PxI32 numToGrab = header.mStride;
PxSolverConstraintDesc* PX_RESTRICT block = &constraintList[header.mStartIndex];
Ps::prefetch(block[0].constraint, 384);
for(PxI32 b = 0; b < numToGrab; ++b)
{
Ps::prefetchLine(block[b].bodyA);
Ps::prefetchLine(block[b].bodyB);
if(bWaitIncrement)
WaitBodyABodyBRequiredState(block[b], normalIteration, frictionIteration);
}
//OK. We have a number of constraints to run...
solveTable[header.mConstraintType](block, PxU32(numToGrab), cache);
//Increment body progresses
if(bWaitIncrement)
{
Ps::memoryBarrier();
for(PxI32 j = 0; j < numToGrab; ++j)
{
IncrementBodyProgress(block[j]);
}
}
}
}
class SolverCoreGeneral : public SolverCore
{
public:
static SolverCoreGeneral* create();
// Implements SolverCore
virtual void destroyV();
virtual PxI32 solveVParallelAndWriteBack
(SolverIslandParams& params) const;
virtual void solveV_Blocks
(SolverIslandParams& params) const;
virtual void writeBackV
(const PxSolverConstraintDesc* PX_RESTRICT constraintList, const PxU32 constraintListSize, PxConstraintBatchHeader* contactConstraintBatches, const PxU32 numBatches,
ThresholdStreamElement* PX_RESTRICT thresholdStream, const PxU32 thresholdStreamLength, PxU32& outThresholdPairs,
PxSolverBodyData* atomListData, WriteBackBlockMethod writeBackTable[]) const;
private:
//~Implements SolverCore
};
#define SOLVEV_BLOCK_METHOD_ARGS \
SolverCore* solverCore, \
SolverIslandParams& params
void solveVBlock(SOLVEV_BLOCK_METHOD_ARGS);
SolveBlockMethod* getSolveBlockTable();
SolveBlockMethod* getSolverConcludeBlockTable();
SolveWriteBackBlockMethod* getSolveWritebackBlockTable();
}
}
#endif //DY_SOLVERCOREGENERAL_H
|