1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_mean_q7.c
4 * Description: Mean value of a Q7 vector
5 *
6 * $Date: 23 April 2021
7 * $Revision: V1.9.0
8 *
9 * Target Processor: Cortex-M and Cortex-A cores
10 * -------------------------------------------------------------------- */
11 /*
12 * Copyright (C) 2010-2021 ARM Limited or its affiliates. All rights reserved.
13 *
14 * SPDX-License-Identifier: Apache-2.0
15 *
16 * Licensed under the Apache License, Version 2.0 (the License); you may
17 * not use this file except in compliance with the License.
18 * You may obtain a copy of the License at
19 *
20 * www.apache.org/licenses/LICENSE-2.0
21 *
22 * Unless required by applicable law or agreed to in writing, software
23 * distributed under the License is distributed on an AS IS BASIS, WITHOUT
24 * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
25 * See the License for the specific language governing permissions and
26 * limitations under the License.
27 */
28
29 #include "dsp/statistics_functions.h"
30
31 /**
32 @ingroup groupStats
33 */
34
35 /**
36 @addtogroup mean
37 @{
38 */
39
40 /**
41 @brief Mean value of a Q7 vector.
42 @param[in] pSrc points to the input vector
43 @param[in] blockSize number of samples in input vector
44 @param[out] pResult mean value returned here
45
46 @par Scaling and Overflow Behavior
47 The function is implemented using a 32-bit internal accumulator.
48 The input is represented in 1.7 format and is accumulated in a 32-bit
49 accumulator in 25.7 format.
50 There is no risk of internal overflow with this approach, and the
51 full precision of intermediate result is preserved.
52 Finally, the accumulator is truncated to yield a result of 1.7 format.
53 */
54
55 #if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
56
arm_mean_q7(const q7_t * pSrc,uint32_t blockSize,q7_t * pResult)57 ARM_DSP_ATTRIBUTE void arm_mean_q7(
58 const q7_t * pSrc,
59 uint32_t blockSize,
60 q7_t * pResult)
61 {
62 uint32_t blkCnt; /* loop counters */
63 q7x16_t vecSrc;
64 q31_t sum = 0L;
65
66
67 blkCnt = blockSize >> 4;
68 while (blkCnt > 0U)
69 {
70 vecSrc = vldrbq_s8(pSrc);
71 /*
72 * sum lanes
73 */
74 sum = vaddvaq(sum, vecSrc);
75
76 blkCnt--;
77 pSrc += 16;
78 }
79
80 blkCnt = blockSize & 0xF;
81 while (blkCnt > 0U)
82 {
83 /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
84 sum += *pSrc++;
85
86 /* Decrement loop counter */
87 blkCnt--;
88 }
89
90 /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) / blockSize */
91 /* Store the result to the destination */
92 *pResult = (q7_t) (sum / (int32_t) blockSize);
93 }
94 #else
arm_mean_q7(const q7_t * pSrc,uint32_t blockSize,q7_t * pResult)95 ARM_DSP_ATTRIBUTE void arm_mean_q7(
96 const q7_t * pSrc,
97 uint32_t blockSize,
98 q7_t * pResult)
99 {
100 uint32_t blkCnt; /* Loop counter */
101 q31_t sum = 0; /* Temporary result storage */
102
103 #if defined (ARM_MATH_LOOPUNROLL)
104 q31_t in;
105 #endif
106
107 #if defined (ARM_MATH_LOOPUNROLL)
108
109 /* Loop unrolling: Compute 4 outputs at a time */
110 blkCnt = blockSize >> 2U;
111
112 while (blkCnt > 0U)
113 {
114 /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
115 in = read_q7x4_ia (&pSrc);
116 sum += ((in << 24U) >> 24U);
117 sum += ((in << 16U) >> 24U);
118 sum += ((in << 8U) >> 24U);
119 sum += (in >> 24U);
120
121 /* Decrement the loop counter */
122 blkCnt--;
123 }
124
125 /* Loop unrolling: Compute remaining outputs */
126 blkCnt = blockSize % 0x4U;
127
128 #else
129
130 /* Initialize blkCnt with number of samples */
131 blkCnt = blockSize;
132
133 #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
134
135 while (blkCnt > 0U)
136 {
137 /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
138 sum += *pSrc++;
139
140 /* Decrement loop counter */
141 blkCnt--;
142 }
143
144 /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) / blockSize */
145 /* Store result to destination */
146 *pResult = (q7_t) (sum / (int32_t) blockSize);
147 }
148 #endif /* defined(ARM_MATH_MVEI) */
149
150 /**
151 @} end of mean group
152 */
153
