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
|
/* ----------------------------------------------------------------------
* Project: CMSIS DSP Library
* Title: arm_canberra_distance_f32.c
* Description: Canberra distance between two vectors
*
* $Date: 23 April 2021
* $Revision: V1.9.0
*
* Target Processor: Cortex-M and Cortex-A cores
* -------------------------------------------------------------------- */
/*
* Copyright (C) 2010-2021 ARM Limited or its affiliates. All rights reserved.
*
* SPDX-License-Identifier: Apache-2.0
*
* Licensed under the Apache License, Version 2.0 (the License); you may
* not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an AS IS BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "dsp/distance_functions.h"
#include <limits.h>
#include <math.h>
/**
@addtogroup Canberra
@{
*/
/**
* @brief Canberra distance between two vectors
*
* This function may divide by zero when samples pA[i] and pB[i] are both zero.
* The result of the computation will be correct. So the division per zero may be
* ignored.
*
* @param[in] pA First vector
* @param[in] pB Second vector
* @param[in] blockSize vector length
* @return distance
*
*/
#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
#include "arm_helium_utils.h"
#include "arm_vec_math.h"
float32_t arm_canberra_distance_f32(const float32_t *pA,const float32_t *pB, uint32_t blockSize)
{
float32_t accum = 0.0f;
uint32_t blkCnt;
f32x4_t a, b, c, accumV;
accumV = vdupq_n_f32(0.0f);
blkCnt = blockSize >> 2;
while (blkCnt > 0) {
a = vld1q(pA);
b = vld1q(pB);
c = vabdq(a, b);
a = vabsq(a);
b = vabsq(b);
a = vaddq(a, b);
/*
* May divide by zero when a and b have both the same lane at zero.
*/
a = vrecip_medprec_f32(a);
/*
* Force result of a division by 0 to 0. It the behavior of the
* sklearn canberra function.
*/
a = vdupq_m_n_f32(a, 0.0f, vcmpeqq(a, 0.0f));
c = vmulq(c, a);
accumV = vaddq(accumV, c);
pA += 4;
pB += 4;
blkCnt--;
}
blkCnt = blockSize & 3;
if (blkCnt > 0U) {
mve_pred16_t p0 = vctp32q(blkCnt);
a = vldrwq_z_f32(pA, p0);
b = vldrwq_z_f32(pB, p0);
c = vabdq(a, b);
a = vabsq(a);
b = vabsq(b);
a = vaddq(a, b);
/*
* May divide by zero when a and b have both the same lane at zero.
*/
a = vrecip_medprec_f32(a);
/*
* Force result of a division by 0 to 0. It the behavior of the
* sklearn canberra function.
*/
a = vdupq_m_n_f32(a, 0.0f, vcmpeqq(a, 0.0f));
c = vmulq(c, a);
accumV = vaddq_m(accumV, accumV, c, p0);
}
accum = vecAddAcrossF32Mve(accumV);
return (accum);
}
#else
#if defined(ARM_MATH_NEON)
#include "NEMath.h"
float32_t arm_canberra_distance_f32(const float32_t *pA,const float32_t *pB, uint32_t blockSize)
{
float32_t accum=0.0f, tmpA, tmpB,diff,sum;
uint32_t blkCnt;
float32x4_t a,b,c,accumV;
float32x2_t accumV2;
uint32x4_t isZeroV;
float32x4_t zeroV = vdupq_n_f32(0.0f);
accumV = vdupq_n_f32(0.0f);
blkCnt = blockSize >> 2;
while(blkCnt > 0)
{
a = vld1q_f32(pA);
b = vld1q_f32(pB);
c = vabdq_f32(a,b);
a = vabsq_f32(a);
b = vabsq_f32(b);
a = vaddq_f32(a,b);
isZeroV = vceqq_f32(a,zeroV);
/*
* May divide by zero when a and b have both the same lane at zero.
*/
a = vinvq_f32(a);
/*
* Force result of a division by 0 to 0. It the behavior of the
* sklearn canberra function.
*/
a = vreinterpretq_f32_s32(vbicq_s32(vreinterpretq_s32_f32(a),vreinterpretq_s32_u32(isZeroV)));
c = vmulq_f32(c,a);
accumV = vaddq_f32(accumV,c);
pA += 4;
pB += 4;
blkCnt --;
}
accumV2 = vpadd_f32(vget_low_f32(accumV),vget_high_f32(accumV));
accum = vget_lane_f32(accumV2, 0) + vget_lane_f32(accumV2, 1);
blkCnt = blockSize & 3;
while(blkCnt > 0)
{
tmpA = *pA++;
tmpB = *pB++;
diff = fabsf(tmpA - tmpB);
sum = fabsf(tmpA) + fabsf(tmpB);
if ((tmpA != 0.0f) || (tmpB != 0.0f))
{
accum += (diff / sum);
}
blkCnt --;
}
return(accum);
}
#else
float32_t arm_canberra_distance_f32(const float32_t *pA,const float32_t *pB, uint32_t blockSize)
{
float32_t accum=0.0f, tmpA, tmpB,diff,sum;
while(blockSize > 0)
{
tmpA = *pA++;
tmpB = *pB++;
diff = fabsf(tmpA - tmpB);
sum = fabsf(tmpA) + fabsf(tmpB);
if ((tmpA != 0.0f) || (tmpB != 0.0f))
{
accum += (diff / sum);
}
blockSize --;
}
return(accum);
}
#endif
#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
/**
* @} end of Canberra group
*/
|
