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
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
|
/*
* 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_FIXEDPOINT_H__
#define __ARM_COMPUTE_FIXEDPOINT_H__
#include <cstdint>
namespace arm_compute
{
using qint8_t = int8_t; /**< 8 bit fixed point scalar value */
using qint16_t = int16_t; /**< 16 bit fixed point scalar value */
using qint32_t = int32_t; /**< 32 bit fixed point scalar value */
using qint64_t = int64_t; /**< 64 bit fixed point scalar value */
/** 8 bit fixed point scalar saturating shift left
*
* @param[in] a First 8 bit fixed point input
* @param[in] shift Shift amount (positive only values)
*
* @return The result of the 8 bit fixed point shift. The result is saturated in case of overflow
*/
qint8_t sqshl_qs8(qint8_t a, int shift);
/** 8 bit fixed point scalar shift right
*
* @param[in] a First 8 bit fixed point input
* @param[in] shift Shift amount (positive only values)
*
* @return The result of the 8 bit fixed point shift
*/
qint8_t sshr_qs8(qint8_t a, int shift);
/** 16 bit fixed point scalar shift right
*
* @param[in] a First 16 bit fixed point input
* @param[in] shift Shift amount (positive only values)
*
* @return The result of the 16 bit fixed point shift
*/
qint16_t sshr_qs16(qint16_t a, int shift);
/** 16 bit fixed point scalar saturating shift left
*
* @param[in] a First 16 bit fixed point input
* @param[in] shift Shift amount (positive only values)
*
* @return The result of the 16 bit fixed point shift. The result is saturated in case of overflow
*/
qint16_t sqshl_qs16(qint16_t a, int shift);
/** 8 bit fixed point scalar absolute value
*
* @param[in] a 8 bit fixed point input
*
* @return The result of the 8 bit fixed point absolute value
*/
qint8_t sabs_qs8(qint8_t a);
/** 16 bit fixed point scalar absolute value
*
* @param[in] a 16 bit fixed point input
*
* @return The result of the 16 bit fixed point absolute value
*/
qint16_t sabs_qs16(qint16_t a);
/** 8 bit fixed point scalar add
*
* @param[in] a First 8 bit fixed point input
* @param[in] b Second 8 bit fixed point input
*
* @return The result of the 8 bit fixed point addition
*/
qint8_t sadd_qs8(qint8_t a, qint8_t b);
/** 16 bit fixed point scalar add
*
* @param[in] a First 16 bit fixed point input
* @param[in] b Second 16 bit fixed point input
*
* @return The result of the 16 bit fixed point addition
*/
qint16_t sadd_qs16(qint16_t a, qint16_t b);
/** 8 bit fixed point scalar saturating add
*
* @param[in] a First 8 bit fixed point input
* @param[in] b Second 8 bit fixed point input
*
* @return The result of the 8 bit fixed point addition. The result is saturated in case of overflow
*/
qint8_t sqadd_qs8(qint8_t a, qint8_t b);
/** 16 bit fixed point scalar saturating add
*
* @param[in] a First 16 bit fixed point input
* @param[in] b Second 16 bit fixed point input
*
* @return The result of the 16 bit fixed point addition. The result is saturated in case of overflow
*/
qint16_t sqadd_qs16(qint16_t a, qint16_t b);
/** 32 bit fixed point scalar saturating add
*
* @param[in] a First 32 bit fixed point input
* @param[in] b Second 32 bit fixed point input
*
* @return The result of the 32 bit fixed point addition. The result is saturated in case of overflow
*/
qint32_t sqadd_qs32(qint32_t a, qint32_t b);
/** 8 bit fixed point scalar subtraction
*
* @param[in] a First 8 bit fixed point input
* @param[in] b Second 8 bit fixed point input
*
* @return The result of the 8 bit fixed point subtraction
*/
qint8_t ssub_qs8(qint8_t a, qint8_t b);
/** 16 bit fixed point scalar subtraction
*
* @param[in] a First 16 bit fixed point input
* @param[in] b Second 16 bit fixed point input
*
* @return The result of the 16 bit fixed point subtraction
*/
qint16_t ssub_qs16(qint16_t a, qint16_t b);
/** 8 bit fixed point scalar saturating subtraction
*
* @param[in] a First 8 bit fixed point input
* @param[in] b Second 8 bit fixed point input
*
* @return The result of the 8 bit fixed point subtraction. The result is saturated in case of overflow
*/
qint8_t sqsub_qs8(qint8_t a, qint8_t b);
/** 16 bit fixed point scalar saturating subtraction
*
* @param[in] a First 16 bit fixed point input
* @param[in] b Second 16 bit fixed point input
*
* @return The result of the 16 bit fixed point subtraction. The result is saturated in case of overflow
*/
qint16_t sqsub_qs16(qint16_t a, qint16_t b);
/** 8 bit fixed point scalar multiply
*
* @param[in] a First 8 bit fixed point input
* @param[in] b Second 8 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 8 bit fixed point multiplication.
*/
qint8_t smul_qs8(qint8_t a, qint8_t b, int fixed_point_position);
/** 16 bit fixed point scalar multiply
*
* @param[in] a First 16 bit fixed point input
* @param[in] b Second 16 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 16 bit fixed point multiplication.
*/
qint16_t smul_qs16(qint16_t a, qint16_t b, int fixed_point_position);
/** 8 bit fixed point scalar saturating multiply
*
* @param[in] a First 8 bit fixed point input
* @param[in] b Second 8 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 8 bit fixed point multiplication. The result is saturated in case of overflow
*/
qint8_t sqmul_qs8(qint8_t a, qint8_t b, int fixed_point_position);
/** 16 bit fixed point scalar saturating multiply
*
* @param[in] a First 16 bit fixed point input
* @param[in] b Second 16 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 16 bit fixed point multiplication. The result is saturated in case of overflow
*/
qint16_t sqmul_qs16(qint16_t a, qint16_t b, int fixed_point_position);
/** 8 bit fixed point scalar multiply long
*
* @param[in] a First 8 bit fixed point input
* @param[in] b Second 8 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 8 bit fixed point multiplication long. The result is saturated in case of overflow
*/
qint16_t sqmull_qs8(qint8_t a, qint8_t b, int fixed_point_position);
/** 16 bit fixed point scalar multiply long
*
* @param[in] a First 16 bit fixed point input
* @param[in] b Second 16 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 16 bit fixed point multiplication long. The result is saturated in case of overflow
*/
qint32_t sqmull_qs16(qint16_t a, qint16_t b, int fixed_point_position);
/** 16 bit fixed point scalar saturating multiply
*
* @param[in] a First 16 bit fixed point input
* @param[in] b Second 16 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 16 bit fixed point multiplication. The result is saturated in case of overflow
*/
qint16_t sqmul_qs16(qint16_t a, qint16_t b, int fixed_point_position);
/** 8 bit fixed point scalar inverse square root
*
* @param[in] a 8 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 8 bit fixed point inverse square root.
*/
qint8_t sinvsqrt_qs8(qint8_t a, int fixed_point_position);
/** 16 bit fixed point scalar inverse square root
*
* @param[in] a 16 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 16 bit fixed point inverse square root.
*/
qint16_t sinvsqrt_qs16(qint16_t a, int fixed_point_position);
/** 8 bit fixed point scalar division
*
* @param[in] a First 8 bit fixed point input
* @param[in] b Second 8 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 8 bit fixed point division.
*/
qint8_t sdiv_qs8(qint8_t a, qint8_t b, int fixed_point_position);
/** 16 bit fixed point scalar division
*
* @param[in] a First 16 bit fixed point input
* @param[in] b Second 16 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 16 bit fixed point division.
*/
qint16_t sdiv_qs16(qint16_t a, qint16_t b, int fixed_point_position);
/** 8 bit fixed point scalar exponential
*
* @param[in] a 8 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 8 bit fixed point exponential.
*/
qint8_t sqexp_qs8(qint8_t a, int fixed_point_position);
/** 16 bit fixed point scalar exponential
*
* @param[in] a 16 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 16 bit fixed point exponential.
*/
qint16_t sqexp_qs16(qint16_t a, int fixed_point_position);
/** 16 bit fixed point scalar exponential
*
* @param[in] a 16 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 16 bit fixed point exponential.
*/
qint16_t sexp_qs16(qint16_t a, int fixed_point_position);
/** 8 bit fixed point scalar logarithm
*
* @param[in] a 8 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 8 bit fixed point logarithm.
*/
qint8_t slog_qs8(qint8_t a, int fixed_point_position);
/** 16 bit fixed point scalar logarithm
*
* @param[in] a 16 bit fixed point input
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the 16 bit fixed point logarithm.
*/
qint16_t slog_qs16(qint16_t a, int fixed_point_position);
/** Convert an 8 bit fixed point to float
*
* @param[in] a Input to convert
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the conversion 8 bit fixed point -> float
*/
float scvt_f32_qs8(qint8_t a, int fixed_point_position);
/** Convert a float to 8 bit fixed point
*
* @param[in] a Input to convert
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the conversion float -> 8 bit fixed point
*/
qint8_t sqcvt_qs8_f32(float a, int fixed_point_position);
/** Convert a 16 bit fixed point to float
*
* @param[in] a Input to convert
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the conversion 16 bit fixed point -> float
*/
float scvt_f32_qs16(qint16_t a, int fixed_point_position);
/** Convert a float to 16 bit fixed point
*
* @param[in] a Input to convert
* @param[in] fixed_point_position Fixed point position that expresses the number of bits for the fractional part of the number
*
* @return The result of the conversion float -> 16 bit fixed point
*/
qint16_t sqcvt_qs16_f32(float a, int fixed_point_position);
/** Scalar saturating move and narrow.
*
* @param[in] a Input to convert to 8 bit fixed point
*
* @return The narrowing conversion to 8 bit
*/
qint8_t sqmovn_qs16(qint16_t a);
/** Scalar saturating move and narrow.
*
* @param[in] a Input to convert to 16 bit fixed point
*
* @return The narrowing conversion to 16 bit
*/
qint16_t sqmovn_qs32(qint32_t a);
}
#include "arm_compute/core/FixedPoint.inl"
#endif /* __ARM_COMPUTE_FIXEDPOINT_H__ */
|
