blob: b70f1cfe696d6e03f4553030f9617552673a9053 (
plain) (
blame)
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
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
|
//===----------------------------------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
// SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES.
//
//===----------------------------------------------------------------------===//
#ifndef _CUDA_STD_NUMBERS
#define _CUDA_STD_NUMBERS
#include <cuda/std/detail/__config>
#if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
# pragma GCC system_header
#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_CLANG)
# pragma clang system_header
#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_MSVC)
# pragma system_header
#endif // no system header
#include <cuda/std/__floating_point/nvfp_types.h>
#include <cuda/std/__floating_point/storage.h>
#include <cuda/std/__type_traits/always_false.h>
#include <cuda/std/version>
#include <cuda/std/__cccl/prologue.h>
_LIBCUDACXX_BEGIN_NAMESPACE_STD
template <class _Tp>
struct __numbers
{
static_assert(__always_false_v<_Tp>,
"[math.constants] A program that instantiates a primary template of a mathematical constant variable "
"template is ill-formed.");
};
#define _CCCL_STD_NUMBERS_SPECIALATION_IMPL(_TYPE, _SUFFIX) \
template <> \
struct __numbers<_TYPE> \
{ \
static _CCCL_API constexpr _TYPE __e() noexcept \
{ \
return 0x1.5bf0a8b1457695355fb8ac404e7ap+1##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __log2e() noexcept \
{ \
return 0x1.71547652b82fe1777d0ffda0d23ap+0##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __log10e() noexcept \
{ \
return 0x1.bcb7b1526e50e32a6ab7555f5a68p-2##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __pi() noexcept \
{ \
return 0x1.921fb54442d18469898cc51701b8p+1##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __inv_pi() noexcept \
{ \
return 0x1.45f306dc9c882a53f84eafa3ea6ap-2##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __inv_sqrtpi() noexcept \
{ \
return 0x1.20dd750429b6d11ae3a914fed7fep-1##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __ln2() noexcept \
{ \
return 0x1.62e42fefa39ef35793c7673007e6p-1##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __ln10() noexcept \
{ \
return 0x1.26bb1bbb5551582dd4adac5705a6p+1##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __sqrt2() noexcept \
{ \
return 0x1.6a09e667f3bcc908b2fb1366ea95p+0##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __sqrt3() noexcept \
{ \
return 0x1.bb67ae8584caa73b25742d7078b8p+0##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __inv_sqrt3() noexcept \
{ \
return 0x1.279a74590331c4d218f81e4afb25p-1##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __egamma() noexcept \
{ \
return 0x1.2788cfc6fb618f49a37c7f0202a6p-1##_SUFFIX; \
} \
static _CCCL_API constexpr _TYPE __phi() noexcept \
{ \
return 0x1.9e3779b97f4a7c15f39cc0605ceep+0##_SUFFIX; \
} \
};
_CCCL_BEGIN_NV_DIAG_SUPPRESS(1046) // floating-point value cannot be represented exactly
_CCCL_STD_NUMBERS_SPECIALATION_IMPL(float, f);
_CCCL_STD_NUMBERS_SPECIALATION_IMPL(double, );
#if _CCCL_HAS_LONG_DOUBLE()
_CCCL_STD_NUMBERS_SPECIALATION_IMPL(long double, l);
#endif // _CCCL_HAS_LONG_DOUBLE()
#if _CCCL_HAS_FLOAT128()
_CCCL_STD_NUMBERS_SPECIALATION_IMPL(__float128, q);
#endif // _CCCL_HAS_FLOAT128()
_CCCL_END_NV_DIAG_SUPPRESS()
#undef _CCCL_STD_NUMBERS_SPECIALATION_IMPL
#if _CCCL_HAS_NVFP16()
template <>
struct __numbers<__half>
{
static _CCCL_API constexpr __half __e() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x4170u});
}
static _CCCL_API constexpr __half __log2e() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x3dc5u});
}
static _CCCL_API constexpr __half __log10e() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x36f3u});
}
static _CCCL_API constexpr __half __pi() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x4248u});
}
static _CCCL_API constexpr __half __inv_pi() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x3518u});
}
static _CCCL_API constexpr __half __inv_sqrtpi() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x3883u});
}
static _CCCL_API constexpr __half __ln2() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x398cu});
}
static _CCCL_API constexpr __half __ln10() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x409bu});
}
static _CCCL_API constexpr __half __sqrt2() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x3da8u});
}
static _CCCL_API constexpr __half __sqrt3() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x3eeeu});
}
static _CCCL_API constexpr __half __inv_sqrt3() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x389eu});
}
static _CCCL_API constexpr __half __egamma() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x389eu});
}
static _CCCL_API constexpr __half __phi() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__half>(__fp_storage_of_t<__half>{0x3e79u});
}
};
#endif // _CCCL_HAS_NVFP16()
#if _CCCL_HAS_NVBF16()
template <>
struct __numbers<__nv_bfloat16>
{
static _CCCL_API constexpr __nv_bfloat16 __e() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x402eu});
}
static _CCCL_API constexpr __nv_bfloat16 __log2e() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x3fb9u});
}
static _CCCL_API constexpr __nv_bfloat16 __log10e() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x3edeu});
}
static _CCCL_API constexpr __nv_bfloat16 __pi() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x4049u});
}
static _CCCL_API constexpr __nv_bfloat16 __inv_pi() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x3ea3u});
}
static _CCCL_API constexpr __nv_bfloat16 __inv_sqrtpi() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x3f10u});
}
static _CCCL_API constexpr __nv_bfloat16 __ln2() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x3f31u});
}
static _CCCL_API constexpr __nv_bfloat16 __ln10() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x4013u});
}
static _CCCL_API constexpr __nv_bfloat16 __sqrt2() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x3fb5u});
}
static _CCCL_API constexpr __nv_bfloat16 __sqrt3() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x3fdeu});
}
static _CCCL_API constexpr __nv_bfloat16 __inv_sqrt3() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x3f14u});
}
static _CCCL_API constexpr __nv_bfloat16 __egamma() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x3f14u});
}
static _CCCL_API constexpr __nv_bfloat16 __phi() noexcept
{
return _CUDA_VSTD::__fp_from_storage<__nv_bfloat16>(__fp_storage_of_t<__nv_bfloat16>{0x3fcfu});
}
};
#endif // _CCCL_HAS_NVBF16()
namespace numbers
{
template <class _Tp>
inline constexpr _Tp e_v = __numbers<_Tp>::__e();
template <class _Tp>
inline constexpr _Tp log2e_v = __numbers<_Tp>::__log2e();
template <class _Tp>
inline constexpr _Tp log10e_v = __numbers<_Tp>::__log10e();
template <class _Tp>
inline constexpr _Tp pi_v = __numbers<_Tp>::__pi();
template <class _Tp>
inline constexpr _Tp inv_pi_v = __numbers<_Tp>::__inv_pi();
template <class _Tp>
inline constexpr _Tp inv_sqrtpi_v = __numbers<_Tp>::__inv_sqrtpi();
template <class _Tp>
inline constexpr _Tp ln2_v = __numbers<_Tp>::__ln2();
template <class _Tp>
inline constexpr _Tp ln10_v = __numbers<_Tp>::__ln10();
template <class _Tp>
inline constexpr _Tp sqrt2_v = __numbers<_Tp>::__sqrt2();
template <class _Tp>
inline constexpr _Tp sqrt3_v = __numbers<_Tp>::__sqrt3();
template <class _Tp>
inline constexpr _Tp inv_sqrt3_v = __numbers<_Tp>::__inv_sqrt3();
template <class _Tp>
inline constexpr _Tp egamma_v = __numbers<_Tp>::__egamma();
template <class _Tp>
inline constexpr _Tp phi_v = __numbers<_Tp>::__phi();
#if !_CCCL_COMPILER(MSVC)
// MSVC errors here because of "error: A __device__ variable template cannot have a const qualified type on Windows"
# if _CCCL_HAS_NVFP16()
template <>
_CCCL_GLOBAL_CONSTANT __half e_v<__half> = __numbers<__half>::__e();
template <>
_CCCL_GLOBAL_CONSTANT __half log2e_v<__half> = __numbers<__half>::__log2e();
template <>
_CCCL_GLOBAL_CONSTANT __half log10e_v<__half> = __numbers<__half>::__log10e();
template <>
_CCCL_GLOBAL_CONSTANT __half pi_v<__half> = __numbers<__half>::__pi();
template <>
_CCCL_GLOBAL_CONSTANT __half inv_pi_v<__half> = __numbers<__half>::__inv_pi();
template <>
_CCCL_GLOBAL_CONSTANT __half inv_sqrtpi_v<__half> = __numbers<__half>::__inv_sqrtpi();
template <>
_CCCL_GLOBAL_CONSTANT __half ln2_v<__half> = __numbers<__half>::__ln2();
template <>
_CCCL_GLOBAL_CONSTANT __half ln10_v<__half> = __numbers<__half>::__ln10();
template <>
_CCCL_GLOBAL_CONSTANT __half sqrt2_v<__half> = __numbers<__half>::__sqrt2();
template <>
_CCCL_GLOBAL_CONSTANT __half sqrt3_v<__half> = __numbers<__half>::__sqrt3();
template <>
_CCCL_GLOBAL_CONSTANT __half inv_sqrt3_v<__half> = __numbers<__half>::__inv_sqrt3();
template <>
_CCCL_GLOBAL_CONSTANT __half egamma_v<__half> = __numbers<__half>::__egamma();
template <>
_CCCL_GLOBAL_CONSTANT __half phi_v<__half> = __numbers<__half>::__phi();
# endif // _CCCL_HAS_NVFP16()
# if _CCCL_HAS_NVBF16()
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 e_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__e();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 log2e_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__log2e();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 log10e_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__log10e();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 pi_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__pi();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 inv_pi_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__inv_pi();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 inv_sqrtpi_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__inv_sqrtpi();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 ln2_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__ln2();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 ln10_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__ln10();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 sqrt2_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__sqrt2();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 sqrt3_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__sqrt3();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 inv_sqrt3_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__inv_sqrt3();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 egamma_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__egamma();
template <>
_CCCL_GLOBAL_CONSTANT __nv_bfloat16 phi_v<__nv_bfloat16> = __numbers<__nv_bfloat16>::__phi();
# endif // _CCCL_HAS_NVBF16()
#endif // !_CCCL_COMPILER(MSVC)
inline constexpr double e = __numbers<double>::__e();
inline constexpr double log2e = __numbers<double>::__log2e();
inline constexpr double log10e = __numbers<double>::__log10e();
inline constexpr double pi = __numbers<double>::__pi();
inline constexpr double inv_pi = __numbers<double>::__inv_pi();
inline constexpr double inv_sqrtpi = __numbers<double>::__inv_sqrtpi();
inline constexpr double ln2 = __numbers<double>::__ln2();
inline constexpr double ln10 = __numbers<double>::__ln10();
inline constexpr double sqrt2 = __numbers<double>::__sqrt2();
inline constexpr double sqrt3 = __numbers<double>::__sqrt3();
inline constexpr double inv_sqrt3 = __numbers<double>::__inv_sqrt3();
inline constexpr double egamma = __numbers<double>::__egamma();
inline constexpr double phi = __numbers<double>::__phi();
} // namespace numbers
_LIBCUDACXX_END_NAMESPACE_STD
#include <cuda/std/__cccl/epilogue.h>
#endif // _CUDA_STD_NUMBERS
|
