Rev 28 | Details | Compare with Previous | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
2 | mjames | 1 | /* ---------------------------------------------------------------------- |
2 | * Copyright (C) 2010-2014 ARM Limited. All rights reserved. |
||
3 | * |
||
4 | * $Date: 19. March 2015 |
||
5 | * $Revision: V.1.4.5 |
||
6 | * |
||
7 | * Project: CMSIS DSP Library |
||
8 | * Title: arm_var_q15.c |
||
9 | * |
||
10 | * Description: Variance of an array of Q15 type. |
||
11 | * |
||
12 | * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0 |
||
13 | * |
||
14 | * Redistribution and use in source and binary forms, with or without |
||
15 | * modification, are permitted provided that the following conditions |
||
16 | * are met: |
||
17 | * - Redistributions of source code must retain the above copyright |
||
18 | * notice, this list of conditions and the following disclaimer. |
||
19 | * - Redistributions in binary form must reproduce the above copyright |
||
20 | * notice, this list of conditions and the following disclaimer in |
||
21 | * the documentation and/or other materials provided with the |
||
22 | * distribution. |
||
23 | * - Neither the name of ARM LIMITED nor the names of its contributors |
||
24 | * may be used to endorse or promote products derived from this |
||
25 | * software without specific prior written permission. |
||
26 | * |
||
27 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
||
28 | * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
||
29 | * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS |
||
30 | * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE |
||
31 | * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, |
||
32 | * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, |
||
33 | * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
||
34 | * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER |
||
35 | * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
||
36 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN |
||
37 | * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
||
38 | * POSSIBILITY OF SUCH DAMAGE. |
||
39 | * -------------------------------------------------------------------- */ |
||
40 | |||
41 | #include "arm_math.h" |
||
42 | |||
43 | /** |
||
44 | * @ingroup groupStats |
||
45 | */ |
||
46 | |||
47 | /** |
||
48 | * @addtogroup variance |
||
49 | * @{ |
||
50 | */ |
||
51 | |||
52 | /** |
||
53 | * @brief Variance of the elements of a Q15 vector. |
||
54 | * @param[in] *pSrc points to the input vector |
||
55 | * @param[in] blockSize length of the input vector |
||
56 | * @param[out] *pResult variance value returned here |
||
57 | * @return none. |
||
58 | * |
||
59 | * @details |
||
60 | * <b>Scaling and Overflow Behavior:</b> |
||
61 | * |
||
62 | * \par |
||
63 | * The function is implemented using a 64-bit internal accumulator. |
||
64 | * The input is represented in 1.15 format. |
||
65 | * Intermediate multiplication yields a 2.30 format, and this |
||
66 | * result is added without saturation to a 64-bit accumulator in 34.30 format. |
||
67 | * With 33 guard bits in the accumulator, there is no risk of overflow, and the |
||
68 | * full precision of the intermediate multiplication is preserved. |
||
69 | * Finally, the 34.30 result is truncated to 34.15 format by discarding the lower |
||
70 | * 15 bits, and then saturated to yield a result in 1.15 format. |
||
71 | * |
||
72 | */ |
||
73 | |||
74 | |||
75 | void arm_var_q15( |
||
76 | q15_t * pSrc, |
||
77 | uint32_t blockSize, |
||
78 | q15_t * pResult) |
||
79 | { |
||
80 | |||
81 | q31_t sum = 0; /* Accumulator */ |
||
82 | q31_t meanOfSquares, squareOfMean; /* square of mean and mean of square */ |
||
83 | uint32_t blkCnt; /* loop counter */ |
||
84 | q63_t sumOfSquares = 0; /* Accumulator */ |
||
85 | |||
86 | #ifndef ARM_MATH_CM0_FAMILY |
||
87 | |||
88 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
||
89 | |||
90 | q31_t in; /* input value */ |
||
91 | q15_t in1; /* input value */ |
||
92 | |||
93 | if(blockSize == 1) |
||
94 | { |
||
95 | *pResult = 0; |
||
96 | return; |
||
97 | } |
||
98 | |||
99 | /*loop Unrolling */ |
||
100 | blkCnt = blockSize >> 2u; |
||
101 | |||
102 | /* First part of the processing with loop unrolling. Compute 4 outputs at a time. |
||
103 | ** a second loop below computes the remaining 1 to 3 samples. */ |
||
104 | while(blkCnt > 0u) |
||
105 | { |
||
106 | /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */ |
||
107 | /* Compute Sum of squares of the input samples |
||
108 | * and then store the result in a temporary variable, sum. */ |
||
109 | in = *__SIMD32(pSrc)++; |
||
110 | sum += ((in << 16) >> 16); |
||
111 | sum += (in >> 16); |
||
112 | sumOfSquares = __SMLALD(in, in, sumOfSquares); |
||
113 | in = *__SIMD32(pSrc)++; |
||
114 | sum += ((in << 16) >> 16); |
||
115 | sum += (in >> 16); |
||
116 | sumOfSquares = __SMLALD(in, in, sumOfSquares); |
||
117 | |||
118 | /* Decrement the loop counter */ |
||
119 | blkCnt--; |
||
120 | } |
||
121 | |||
122 | /* If the blockSize is not a multiple of 4, compute any remaining output samples here. |
||
123 | ** No loop unrolling is used. */ |
||
124 | blkCnt = blockSize % 0x4u; |
||
125 | |||
126 | while(blkCnt > 0u) |
||
127 | { |
||
128 | /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */ |
||
129 | /* Compute Sum of squares of the input samples |
||
130 | * and then store the result in a temporary variable, sum. */ |
||
131 | in1 = *pSrc++; |
||
132 | sumOfSquares = __SMLALD(in1, in1, sumOfSquares); |
||
133 | sum += in1; |
||
134 | |||
135 | /* Decrement the loop counter */ |
||
136 | blkCnt--; |
||
137 | } |
||
138 | |||
139 | /* Compute Mean of squares of the input samples |
||
140 | * and then store the result in a temporary variable, meanOfSquares. */ |
||
141 | meanOfSquares = (q31_t) (sumOfSquares / (q63_t)(blockSize - 1)); |
||
142 | |||
143 | /* Compute square of mean */ |
||
144 | squareOfMean = (q31_t)((q63_t)sum * sum / (q63_t)(blockSize * (blockSize - 1))); |
||
145 | |||
146 | /* mean of the squares minus the square of the mean. */ |
||
147 | *pResult = (meanOfSquares - squareOfMean) >> 15; |
||
148 | |||
149 | #else |
||
150 | |||
151 | /* Run the below code for Cortex-M0 */ |
||
152 | q15_t in; /* input value */ |
||
153 | |||
154 | if(blockSize == 1) |
||
155 | { |
||
156 | *pResult = 0; |
||
157 | return; |
||
158 | } |
||
159 | |||
160 | /* Loop over blockSize number of values */ |
||
161 | blkCnt = blockSize; |
||
162 | |||
163 | while(blkCnt > 0u) |
||
164 | { |
||
165 | /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */ |
||
166 | /* Compute Sum of squares of the input samples |
||
167 | * and then store the result in a temporary variable, sumOfSquares. */ |
||
168 | in = *pSrc++; |
||
169 | sumOfSquares += (in * in); |
||
170 | |||
171 | /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */ |
||
172 | /* Compute sum of all input values and then store the result in a temporary variable, sum. */ |
||
173 | sum += in; |
||
174 | |||
175 | /* Decrement the loop counter */ |
||
176 | blkCnt--; |
||
177 | } |
||
178 | |||
179 | /* Compute Mean of squares of the input samples |
||
180 | * and then store the result in a temporary variable, meanOfSquares. */ |
||
181 | meanOfSquares = (q31_t) (sumOfSquares / (q63_t)(blockSize - 1)); |
||
182 | |||
183 | /* Compute square of mean */ |
||
184 | squareOfMean = (q31_t)((q63_t)sum * sum / (q63_t)(blockSize * (blockSize - 1))); |
||
185 | |||
186 | /* mean of the squares minus the square of the mean. */ |
||
187 | *pResult = (meanOfSquares - squareOfMean) >> 15; |
||
188 | |||
189 | #endif /* #ifndef ARM_MATH_CM0_FAMILY */ |
||
190 | |||
191 | } |
||
192 | |||
193 | /** |
||
194 | * @} end of variance group |
||
195 | */ |