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
|
// 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) 2008-2018 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 PSFOUNDATION_PSWINDOWSINTRINSICS_H
#define PSFOUNDATION_PSWINDOWSINTRINSICS_H
#include "Ps.h"
#include "foundation/PxAssert.h"
// this file is for internal intrinsics - that is, intrinsics that are used in
// cross platform code but do not appear in the API
#if !PX_WINDOWS_FAMILY
#error "This file should only be included by Windows builds!!"
#endif
#pragma warning(push)
//'symbol' is not defined as a preprocessor macro, replacing with '0' for 'directives'
#pragma warning(disable : 4668)
#if PX_VC == 10
#pragma warning(disable : 4987) // nonstandard extension used: 'throw (...)'
#endif
#include <intrin.h>
#pragma warning(pop)
#pragma warning(push)
#pragma warning(disable : 4985) // 'symbol name': attributes not present on previous declaration
#include <math.h>
#pragma warning(pop)
#include <float.h>
#include <mmintrin.h>
#pragma intrinsic(_BitScanForward)
#pragma intrinsic(_BitScanReverse)
namespace physx
{
namespace shdfnd
{
/*
* Implements a memory barrier
*/
PX_FORCE_INLINE void memoryBarrier()
{
_ReadWriteBarrier();
/* long Barrier;
__asm {
xchg Barrier, eax
}*/
}
/*!
Returns the index of the highest set bit. Not valid for zero arg.
*/
PX_FORCE_INLINE uint32_t highestSetBitUnsafe(uint32_t v)
{
unsigned long retval;
_BitScanReverse(&retval, v);
return retval;
}
/*!
Returns the index of the highest set bit. Undefined for zero arg.
*/
PX_FORCE_INLINE uint32_t lowestSetBitUnsafe(uint32_t v)
{
unsigned long retval;
_BitScanForward(&retval, v);
return retval;
}
/*!
Returns the number of leading zeros in v. Returns 32 for v=0.
*/
PX_FORCE_INLINE uint32_t countLeadingZeros(uint32_t v)
{
if(v)
{
unsigned long bsr = (unsigned long)-1;
_BitScanReverse(&bsr, v);
return 31 - bsr;
}
else
return 32;
}
/*!
Prefetch aligned cache size around \c ptr+offset.
*/
#if !PX_ARM
PX_FORCE_INLINE void prefetchLine(const void* ptr, uint32_t offset = 0)
{
// cache line on X86/X64 is 64-bytes so a 128-byte prefetch would require 2 prefetches.
// However, we can only dispatch a limited number of prefetch instructions so we opt to prefetch just 1 cache line
/*_mm_prefetch(((const char*)ptr + offset), _MM_HINT_T0);*/
// We get slightly better performance prefetching to non-temporal addresses instead of all cache levels
_mm_prefetch(((const char*)ptr + offset), _MM_HINT_NTA);
}
#else
PX_FORCE_INLINE void prefetchLine(const void* ptr, uint32_t offset = 0)
{
// arm does have 32b cache line size
__prefetch(((const char*)ptr + offset));
}
#endif
/*!
Prefetch \c count bytes starting at \c ptr.
*/
#if !PX_ARM
PX_FORCE_INLINE void prefetch(const void* ptr, uint32_t count = 1)
{
const char* cp = (char*)ptr;
uint64_t p = size_t(ptr);
uint64_t startLine = p >> 6, endLine = (p + count - 1) >> 6;
uint64_t lines = endLine - startLine + 1;
do
{
prefetchLine(cp);
cp += 64;
} while(--lines);
}
#else
PX_FORCE_INLINE void prefetch(const void* ptr, uint32_t count = 1)
{
const char* cp = (char*)ptr;
uint32_t p = size_t(ptr);
uint32_t startLine = p >> 5, endLine = (p + count - 1) >> 5;
uint32_t lines = endLine - startLine + 1;
do
{
prefetchLine(cp);
cp += 32;
} while(--lines);
}
#endif
//! \brief platform-specific reciprocal
PX_CUDA_CALLABLE PX_FORCE_INLINE float recipFast(float a)
{
return 1.0f / a;
}
//! \brief platform-specific fast reciprocal square root
PX_CUDA_CALLABLE PX_FORCE_INLINE float recipSqrtFast(float a)
{
return 1.0f / ::sqrtf(a);
}
//! \brief platform-specific floor
PX_CUDA_CALLABLE PX_FORCE_INLINE float floatFloor(float x)
{
return ::floorf(x);
}
#define NS_EXPECT_TRUE(x) x
#define NS_EXPECT_FALSE(x) x
} // namespace shdfnd
} // namespace physx
#endif // #ifndef PSFOUNDATION_PSWINDOWSINTRINSICS_H
|