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
|
/* ----------------------------------------------------------------------
* Project: CMSIS DSP Library
* Title: arm_scale_f32.c
* Description: Multiplies a floating-point vector by a scalar
*
* $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/basic_math_functions.h"
/**
@ingroup groupMath
*/
/**
@defgroup BasicScale Vector Scale
Multiply a vector by a scalar value. For floating-point data, the algorithm used is:
<pre>
pDst[n] = pSrc[n] * scale, 0 <= n < blockSize.
</pre>
In the fixed-point Q7, Q15, and Q31 functions, <code>scale</code> is represented by
a fractional multiplication <code>scaleFract</code> and an arithmetic shift <code>shift</code>.
The shift allows the gain of the scaling operation to exceed 1.0.
The algorithm used with fixed-point data is:
<pre>
pDst[n] = (pSrc[n] * scaleFract) << shift, 0 <= n < blockSize.
</pre>
The overall scale factor applied to the fixed-point data is
<pre>
scale = scaleFract * 2^shift.
</pre>
The functions support in-place computation allowing the source and destination
pointers to reference the same memory buffer.
*/
/**
@addtogroup BasicScale
@{
*/
/**
@brief Multiplies a floating-point vector by a scalar.
@param[in] pSrc points to the input vector
@param[in] scale scale factor to be applied
@param[out] pDst points to the output vector
@param[in] blockSize number of samples in each vector
@return none
*/
#if defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE)
#include "arm_helium_utils.h"
void arm_scale_f32(
const float32_t * pSrc,
float32_t scale,
float32_t * pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* Loop counter */
f32x4_t vec1;
f32x4_t res;
/* Compute 4 outputs at a time */
blkCnt = blockSize >> 2U;
while (blkCnt > 0U)
{
/* C = A + offset */
/* Add offset and then store the results in the destination buffer. */
vec1 = vld1q(pSrc);
res = vmulq(vec1,scale);
vst1q(pDst, res);
/* Increment pointers */
pSrc += 4;
pDst += 4;
/* Decrement the loop counter */
blkCnt--;
}
/* Tail */
blkCnt = blockSize & 0x3;
if (blkCnt > 0U)
{
mve_pred16_t p0 = vctp32q(blkCnt);
vec1 = vld1q((float32_t const *) pSrc);
vstrwq_p(pDst, vmulq(vec1, scale), p0);
}
}
#else
void arm_scale_f32(
const float32_t *pSrc,
float32_t scale,
float32_t *pDst,
uint32_t blockSize)
{
uint32_t blkCnt; /* Loop counter */
#if defined(ARM_MATH_NEON_EXPERIMENTAL)
f32x4_t vec1;
f32x4_t res;
/* Compute 4 outputs at a time */
blkCnt = blockSize >> 2U;
while (blkCnt > 0U)
{
/* C = A * scale */
/* Scale the input and then store the results in the destination buffer. */
vec1 = vld1q_f32(pSrc);
res = vmulq_f32(vec1, vdupq_n_f32(scale));
vst1q_f32(pDst, res);
/* Increment pointers */
pSrc += 4;
pDst += 4;
/* Decrement the loop counter */
blkCnt--;
}
/* Tail */
blkCnt = blockSize & 0x3;
#else
#if defined (ARM_MATH_LOOPUNROLL)
/* Loop unrolling: Compute 4 outputs at a time */
blkCnt = blockSize >> 2U;
while (blkCnt > 0U)
{
float32_t in1, in2, in3, in4;
/* C = A * scale */
/* Scale input and store result in destination buffer. */
in1 = (*pSrc++) * scale;
in2 = (*pSrc++) * scale;
in3 = (*pSrc++) * scale;
in4 = (*pSrc++) * scale;
*pDst++ = in1;
*pDst++ = in2;
*pDst++ = in3;
*pDst++ = in4;
/* Decrement loop counter */
blkCnt--;
}
/* Loop unrolling: Compute remaining outputs */
blkCnt = blockSize % 0x4U;
#else
/* Initialize blkCnt with number of samples */
blkCnt = blockSize;
#endif /* #if defined (ARM_MATH_LOOPUNROLL) */
#endif /* #if defined(ARM_MATH_NEON_EXPERIMENTAL) */
while (blkCnt > 0U)
{
/* C = A * scale */
/* Scale input and store result in destination buffer. */
*pDst++ = (*pSrc++) * scale;
/* Decrement loop counter */
blkCnt--;
}
}
#endif /* defined(ARM_MATH_MVEF) && !defined(ARM_MATH_AUTOVECTORIZE) */
/**
@} end of BasicScale group
*/
|
