tmk_core/tool/mbed/mbed-sdk/libraries/dsp/cmsis_dsp/StatisticsFunctions/arm_var_q31.c

   1 /* ----------------------------------------------------------------------
   2 * Copyright (C) 2010-2013 ARM Limited. All rights reserved.
   3 *
   4 * $Date:        17. January 2013
   5 * $Revision:    V1.4.1
   6 *
   7 * Project:          CMSIS DSP Library
   8 * Title:                arm_var_q31.c
   9 *
  10 * Description:  Variance of an array of Q31 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 Q31 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 an internal 64-bit accumulator.
  64  * The input is represented in 1.31 format, and intermediate multiplication
  65  * yields a 2.62 format.
  66  * The accumulator maintains full precision of the intermediate multiplication results,
  67  * but provides only a single guard bit.
  68  * There is no saturation on intermediate additions.
  69  * If the accumulator overflows it wraps around and distorts the result.
  70  * In order to avoid overflows completely the input signal must be scaled down by
  71  * log2(blockSize) bits, as a total of blockSize additions are performed internally.
  72  * Finally, the 2.62 accumulator is right shifted by 31 bits to yield a 1.31 format value.
  73  *
  74  */
  75
  76
  77 void arm_var_q31(
  78   q31_t * pSrc,
  79   uint32_t blockSize,
  80   q63_t * pResult)
  81 {
  82   q63_t sum = 0, sumSquare = 0;                  /* Accumulator */
  83   q31_t meanOfSquares, squareOfMean;             /* square of mean and mean of square */
  84   q31_t mean;                                    /* mean */
  85   q31_t in;                                      /* input value */
  86   q31_t t;                                       /* Temporary variable */
  87   uint32_t blkCnt;                               /* loop counter */
  88
  89 #ifndef ARM_MATH_CM0_FAMILY
  90
  91   /* Run the below code for Cortex-M4 and Cortex-M3 */
  92   q63_t sumSquare1 = 0;                          /* Accumulator */
  93   q31_t in1, in2, in3, in4;                      /* Temporary input variables */
  94
  95   /*loop Unrolling */
  96   blkCnt = blockSize >> 2u;
  97
  98   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
  99    ** a second loop below computes the remaining 1 to 3 samples. */
 100   while(blkCnt > 0u)
 101   {
 102     /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1])  */
 103     /* Compute Sum of squares of the input samples
 104      * and then store the result in a temporary variable, sum. */
 105     /* read input samples from source buffer */
 106     in1 = pSrc[0];
 107     in2 = pSrc[1];
 108
 109     /* calculate sum of inputs */
 110     sum += in1;
 111     /* calculate sum of squares */
 112     sumSquare += ((q63_t) (in1) * (in1));
 113     in3 = pSrc[2];
 114     sum += in2;
 115     sumSquare1 += ((q63_t) (in2) * (in2));
 116     in4 = pSrc[3];
 117     sum += in3;
 118     sumSquare += ((q63_t) (in3) * (in3));
 119     sum += in4;
 120     sumSquare1 += ((q63_t) (in4) * (in4));
 121
 122     /* update input pointer to process next samples */
 123     pSrc += 4u;
 124
 125     /* Decrement the loop counter */
 126     blkCnt--;
 127   }
 128
 129   /* add two accumulators */
 130   sumSquare = sumSquare + sumSquare1;
 131
 132   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
 133    ** No loop unrolling is used. */
 134   blkCnt = blockSize % 0x4u;
 135
 136 #else
 137
 138   /* Run the below code for Cortex-M0 */
 139   blkCnt = blockSize;
 140
 141 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
 142
 143   while(blkCnt > 0u)
 144   {
 145     /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */
 146     /* Compute Sum of squares of the input samples
 147      * and then store the result in a temporary variable, sum. */
 148     in = *pSrc++;
 149     sumSquare += ((q63_t) (in) * (in));
 150     sum += in;
 151
 152     /* Decrement the loop counter */
 153     blkCnt--;
 154   }
 155
 156   t = (q31_t) ((1.0f / (float32_t) (blockSize - 1u)) * 1073741824.0f);
 157
 158   /* Compute Mean of squares of the input samples
 159    * and then store the result in a temporary variable, meanOfSquares. */
 160   sumSquare = (sumSquare >> 31);
 161   meanOfSquares = (q31_t) ((sumSquare * t) >> 30);
 162
 163   /* Compute mean of all input values */
 164   t = (q31_t) ((1.0f / (blockSize * (blockSize - 1u))) * 2147483648.0f);
 165   mean = (q31_t) (sum);
 166
 167   /* Compute square of mean */
 168   squareOfMean = (q31_t) (((q63_t) mean * mean) >> 31);
 169   squareOfMean = (q31_t) (((q63_t) squareOfMean * t) >> 31);
 170
 171   /* Compute variance and then store the result to the destination */
 172   *pResult = (q63_t) meanOfSquares - squareOfMean;
 173
 174 }
 175
 176 /**
 177  * @} end of variance group
 178  */