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
|
/*
* Copyright (c) 2016-2019 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to
* deal in the Software without restriction, including without limitation the
* rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
* sell copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#ifndef __ARM_COMPUTE_NEMATH_H__
#define __ARM_COMPUTE_NEMATH_H__
#include <arm_neon.h>
namespace arm_compute
{
/** Calculate floor of a vector.
*
* @param[in] val Input vector value in F32 format.
*
* @return The calculated floor vector.
*/
float32x4_t vfloorq_f32(float32x4_t val);
/** Calculate round value of a vector to nearest with ties to even.
*
* @param[in] val Input vector value in F32 format.
*
* @return The calculated round vector.
*/
float32x4_t vroundq_rte_f32(float32x4_t val);
/** Calculate inverse square root.
*
* @param[in] x Input value.
*
* @return The calculated inverse square root.
*/
float32x2_t vinvsqrt_f32(float32x2_t x);
/** Calculate inverse square root.
*
* @param[in] x Input value.
*
* @return The calculated inverse square root.
*/
float32x4_t vinvsqrtq_f32(float32x4_t x);
/** Calculate reciprocal.
*
* @param[in] x Input value.
*
* @return The calculated reciprocal.
*/
float32x2_t vinv_f32(float32x2_t x);
/** Calculate reciprocal.
*
* @param[in] x Input value.
*
* @return The calculated reciprocal.
*/
float32x4_t vinvq_f32(float32x4_t x);
/** Perform a 7th degree polynomial approximation using Estrin's method.
*
* @param[in] x Input vector value in F32 format.
* @param[in] coeffs Polynomial coefficients table.
*
* @return The calculated approximation.
*/
float32x4_t vtaylor_polyq_f32(float32x4_t x, const std::array<float32x4_t, 8> &coeffs);
/** Calculate exponential
*
* @param[in] x Input vector value in F32 format.
*
* @return The calculated exponent.
*/
float32x4_t vexpq_f32(float32x4_t x);
/** Calculate logarithm
*
* @param[in] x Input vector value in F32 format.
*
* @return The calculated logarithm.
*/
float32x4_t vlogq_f32(float32x4_t x);
/** Calculate hyperbolic tangent.
*
* tanh(x) = (e^2x - 1)/(e^2x + 1)
*
* @note We clamp x to [-5,5] to avoid overflowing issues.
*
* @param[in] val Input vector value in F32 format.
*
* @return The calculated Hyperbolic Tangent.
*/
float32x4_t vtanhq_f32(float32x4_t val);
/** Calculate n power of a number.
*
* pow(x,n) = e^(n*log(x))
*
* @param[in] val Input vector value in F32 format.
* @param[in] n Powers to raise the input to.
*
* @return The calculated power.
*/
float32x4_t vpowq_f32(float32x4_t val, float32x4_t n);
/** Round to the nearest division by a power-of-two using exponent
*
* @note This function calculates the following expression: (x + 2^n -1 ) / 2^n where n = exponent
*
* @param[in] x Vector of 4 elements
* @param[in] exponent Integer value used to round to nearest division by a power-of-two
*
* @return the nearest division by a power-of-two using exponent
*/
int32x4_t rounding_divide_by_pow2(int32x4_t x, int exponent);
/** Round to the nearest division by a power-of-two using exponent
*
* @note This function calculates the following expression: (x + 2^n -1 ) / 2^n where n = exponent
*
* @param[in] x Element to divide.
* @param[in] exponent Integer value used to round to nearest division by a power-of-two
*
* @return the nearest division by a power-of-two using exponent
*/
int32_t rounding_divide_by_pow2(int32_t x, int exponent);
/** Calculate sine.
*
* @param[in] val Input vector value in radians, F32 format.
*
* @return The calculated sine.
*/
float32x4_t vsinq_f32(float32x4_t val);
/** Calculate sine.
*
* @param[in] val Input vector value in radians, F32 format.
*
* @return The calculated sine.
*/
float32x2_t vsin_f32(float32x2_t val);
#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
/** Calculate hyperbolic tangent.
*
* tanh(x) = (e^2x - 1)/(e^2x + 1)
*
* @note We clamp x to [-5,5] to avoid overflowing issues.
*
* @param[in] val Input vector value in F16 format.
*
* @return The calculated Hyperbolic Tangent.
*/
float16x8_t vtanhq_f16(float16x8_t val);
/** Calculate round value of a vector to nearest with ties to even.
*
* @param[in] val Input vector value in F16 format.
*
* @return The calculated round vector.
*/
float16x8_t vroundq_rte_f16(float16x8_t val);
/** Calculate reciprocal.
*
* @param[in] x Input value.
*
* @return The calculated reciprocal.
*/
float16x4_t vinv_f16(float16x4_t x);
/** Calculate reciprocal.
*
* @param[in] x Input value.
*
* @return The calculated reciprocal.
*/
float16x8_t vinvq_f16(float16x8_t x);
/** Calculate inverse square root.
*
* @param[in] x Input value.
*
* @return The calculated inverse square root.
*/
float16x4_t vinvsqrt_f16(float16x4_t x);
/** Calculate inverse square root.
*
* @param[in] x Input value.
*
* @return The calculated inverse square root.
*/
float16x8_t vinvsqrtq_f16(float16x8_t x);
/** Calculate exponential
*
* @param[in] x Input vector value in F16 format.
*
* @return The calculated exponent.
*/
float16x8_t vexpq_f16(float16x8_t x);
/** Calculate n power of a number.
*
* pow(x,n) = e^(n*log(x))
*
* @param[in] val Input vector value in F16 format.
* @param[in] n Powers to raise the input to.
*
* @return The calculated power.
*/
float16x8_t vpowq_f16(float16x8_t val, float16x8_t n);
/** Calculate sine.
*
* @param[in] val Input vector value in radians, F16 format.
*
* @return The calculated sine.
*/
float16x8_t vsinq_f16(float16x8_t val);
#endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
} // namespace arm_compute
#include "arm_compute/core/NEON/NEMath.inl"
#endif /* __ARM_COMPUTE_NEMATH_H__ */
|
