• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Copyright (c) 2016-2020 Arm Limited.
3  *
4  * SPDX-License-Identifier: MIT
5  *
6  * Permission is hereby granted, free of charge, to any person obtaining a copy
7  * of this software and associated documentation files (the "Software"), to
8  * deal in the Software without restriction, including without limitation the
9  * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
10  * sell copies of the Software, and to permit persons to whom the Software is
11  * furnished to do so, subject to the following conditions:
12  *
13  * The above copyright notice and this permission notice shall be included in all
14  * copies or substantial portions of the Software.
15  *
16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
22  * SOFTWARE.
23  */
24 #ifndef ARM_COMPUTE_NEMATH_H
25 #define ARM_COMPUTE_NEMATH_H
26 
27 #include <arm_neon.h>
28 #include <array>
29 
30 namespace arm_compute
31 {
32 /** Calculate floor of a vector.
33  *
34  * @param[in] val Input vector value in F32 format.
35  *
36  * @return The calculated floor vector.
37  */
38 float32x4_t vfloorq_f32(float32x4_t val);
39 
40 /** Calculate round value of a vector to nearest with ties to even.
41  *
42  * @param[in] val Input vector value in F32 format.
43  *
44  * @return The calculated round vector.
45  */
46 float32x4_t vroundq_rte_f32(float32x4_t val);
47 
48 /** Calculate inverse square root.
49  *
50  * @param[in] x Input value.
51  *
52  * @return The calculated inverse square root.
53  */
54 float32x2_t vinvsqrt_f32(float32x2_t x);
55 
56 /** Calculate inverse square root.
57  *
58  * @param[in] x Input value.
59  *
60  * @return The calculated inverse square root.
61  */
62 float32x4_t vinvsqrtq_f32(float32x4_t x);
63 
64 /** Calculate reciprocal.
65  *
66  * @param[in] x Input value.
67  *
68  * @return The calculated reciprocal.
69  */
70 float32x2_t vinv_f32(float32x2_t x);
71 
72 /** Calculate reciprocal.
73  *
74  * @param[in] x Input value.
75  *
76  * @return The calculated reciprocal.
77  */
78 float32x4_t vinvq_f32(float32x4_t x);
79 
80 /** Perform a 7th degree polynomial approximation using Estrin's method.
81  *
82  * @param[in] x      Input vector value in F32 format.
83  * @param[in] coeffs Polynomial coefficients table.
84  *
85  * @return The calculated approximation.
86  */
87 float32x4_t vtaylor_polyq_f32(float32x4_t x, const std::array<float32x4_t, 8> &coeffs);
88 
89 /** Calculate exponential
90  *
91  * @param[in] x Input vector value in F32 format.
92  *
93  * @return The calculated exponent.
94  */
95 float32x4_t vexpq_f32(float32x4_t x);
96 
97 /** Calculate logarithm
98  *
99  * @param[in] x Input vector value in F32 format.
100  *
101  * @return The calculated logarithm.
102  */
103 float32x4_t vlogq_f32(float32x4_t x);
104 
105 /** Calculate hyperbolic tangent.
106  *
107  * tanh(x) = (e^2x - 1)/(e^2x + 1)
108  *
109  * @note We clamp x to [-5,5] to avoid overflowing issues.
110  *
111  * @param[in] val Input vector value in F32 format.
112  *
113  * @return The calculated Hyperbolic Tangent.
114  */
115 float32x4_t vtanhq_f32(float32x4_t val);
116 
117 /** Calculate n power of a number.
118  *
119  * pow(x,n) = e^(n*log(x))
120  *
121  * @param[in] val Input vector value in F32 format.
122  * @param[in] n   Powers to raise the input to.
123  *
124  * @return The calculated power.
125  */
126 float32x4_t vpowq_f32(float32x4_t val, float32x4_t n);
127 
128 /** Round to the nearest division by a power-of-two using exponent
129  *
130  * @note This function calculates the following expression: (x + 2^n -1 ) / 2^n where n = exponent
131  *
132  * @param[in] x        Vector of 4 elements
133  * @param[in] exponent Vector of 4 elements with integer value used to round to nearest division by a power-of-two
134  *
135  * @return the nearest division by a power-of-two using exponent
136  */
137 int32x4_t rounding_divide_by_pow2(int32x4_t x, int32x4_t exponent);
138 
139 /** Round to the nearest division by a power-of-two using exponent
140  *
141  * @note This function calculates the following expression: (x + 2^n -1 ) / 2^n where n = exponent
142  *
143  * @param[in] x        Vector of 4 elements
144  * @param[in] exponent Integer value used to round to nearest division by a power-of-two
145  *
146  * @return the nearest division by a power-of-two using exponent
147  */
148 int32x4_t rounding_divide_by_pow2(int32x4_t x, int exponent);
149 
150 /** Round to the nearest division by a power-of-two using exponent
151  *
152  * @note This function calculates the following expression: (x + 2^n -1 ) / 2^n where n = exponent
153  *
154  * @param[in] x        Element to divide.
155  * @param[in] exponent Integer value used to round to nearest division by a power-of-two
156  *
157  * @return the nearest division by a power-of-two using exponent
158  */
159 int32_t rounding_divide_by_pow2(int32_t x, int exponent);
160 
161 /** Converts from uint8x16 to float32x4x4_t
162  *
163  * @param[in] in Vector of uint8 to be converted
164  *
165  * @return Converted vector of float
166  */
167 float32x4x4_t convert_uint8x16_to_float32x4x4(const uint8x16_t &in);
168 
169 /** Converts from int8x16 to float32x4x4_t
170  *
171  * @param[in] in Vector of int8 to be converted
172  *
173  * @return Converted vector of float
174  */
175 float32x4x4_t convert_int8x16_to_float32x4x4(const int8x16_t &in);
176 
177 /** Converts to float32x4x4_t from the specified templated 16 elements vectors
178  *
179  * @param[in] in Vector of float to be converted
180  *
181  * @return Converted vector of float
182  */
183 template <typename T>
184 float32x4x4_t convert_to_float32x4x4(const T &in);
185 
186 /** Converts from two float32x4x3_t to just one uint8x8x3_t
187  *
188  * @param[in]  in1 First input vector of float to be converted
189  * @param[in]  in2 Second input vector of float to be converted
190  * @param[out] out Converted output vector uint8 to store the result
191  */
192 void convert_float32x4x3_to_uint8x8x3(const float32x4x3_t &in1, const float32x4x3_t &in2, uint8x8x3_t &out);
193 
194 /** Converts from two float32x4x4_t to just one uint8x16_t
195  *
196  * @param[in]  in  Vector of float to be converted
197  * @param[out] out Converted vector of uint8 to store the result
198  */
199 void convert_float32x4x4_to_uint8x16(const float32x4x4_t &in, uint8x16_t &out);
200 
201 /** Converts from float32x4x4_t to just one int8x16_t
202  *
203  * @param[in]  in  Vector of float to be converted
204  * @param[out] out Converted vector of uint8 to store the result
205  */
206 void convert_float32x4x4_to_int8x16(const float32x4x4_t &in, int8x16_t &out);
207 
208 /** Calculate sine.
209  *
210  * @param[in] val Input vector value in radians, F32 format.
211  *
212  * @return The calculated sine.
213  */
214 float32x4_t vsinq_f32(float32x4_t val);
215 
216 /** Calculate sine.
217  *
218  * @param[in] val Input vector value in radians, F32 format.
219  *
220  * @return The calculated sine.
221  */
222 float32x2_t vsin_f32(float32x2_t val);
223 
224 #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
225 /** Calculate hyperbolic tangent.
226  *
227  * tanh(x) = (e^2x - 1)/(e^2x + 1)
228  *
229  * @note We clamp x to [-5,5] to avoid overflowing issues.
230  *
231  * @param[in] val Input vector value in F16 format.
232  *
233  * @return The calculated Hyperbolic Tangent.
234  */
235 float16x8_t vtanhq_f16(float16x8_t val);
236 
237 /** Calculate round value of a vector to nearest with ties to even.
238  *
239  * @param[in] val Input vector value in F16 format.
240  *
241  * @return The calculated round vector.
242  */
243 float16x8_t vroundq_rte_f16(float16x8_t val);
244 
245 /** Calculate reciprocal.
246  *
247  * @param[in] x Input value.
248  *
249  * @return The calculated reciprocal.
250  */
251 float16x4_t vinv_f16(float16x4_t x);
252 
253 /** Calculate reciprocal.
254  *
255  * @param[in] x Input value.
256  *
257  * @return The calculated reciprocal.
258  */
259 float16x8_t vinvq_f16(float16x8_t x);
260 
261 /** Calculate inverse square root.
262  *
263  * @param[in] x Input value.
264  *
265  * @return The calculated inverse square root.
266  */
267 float16x4_t vinvsqrt_f16(float16x4_t x);
268 
269 /** Calculate inverse square root.
270  *
271  * @param[in] x Input value.
272  *
273  * @return The calculated inverse square root.
274  */
275 float16x8_t vinvsqrtq_f16(float16x8_t x);
276 
277 /** Calculate exponential
278  *
279  * @param[in] x Input vector value in F16 format.
280  *
281  * @return The calculated exponent.
282  */
283 float16x8_t vexpq_f16(float16x8_t x);
284 
285 /** Calculate n power of a number.
286  *
287  * pow(x,n) = e^(n*log(x))
288  *
289  * @param[in] val Input vector value in F16 format.
290  * @param[in] n   Powers to raise the input to.
291  *
292  * @return The calculated power.
293  */
294 float16x8_t vpowq_f16(float16x8_t val, float16x8_t n);
295 
296 /** Calculate sine.
297  *
298  * @param[in] val Input vector value in radians, F16 format.
299  *
300  * @return The calculated sine.
301  */
302 float16x8_t vsinq_f16(float16x8_t val);
303 
304 #endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
305 } // namespace arm_compute
306 #include "src/core/NEON/NEMath.inl"
307 #endif /* ARM_COMPUTE_NEMATH_H */
308