lib/main/CMSIS/DSP/Source/StatisticsFunctions/arm_mean_q15.c

   1 /* ----------------------------------------------------------------------
   2  * Project:      CMSIS DSP Library
   3  * Title:        arm_mean_q15.c
   4  * Description:  Mean value of a Q15 vector
   5  *
   6  * $Date:        27. January 2017
   7  * $Revision:    V.1.5.1
   8  *
   9  * Target Processor: Cortex-M cores
  10  * -------------------------------------------------------------------- */
  11 /*
  12  * Copyright (C) 2010-2017 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 "arm_math.h"
  30
  31 /**
  32  * @ingroup groupStats
  33  */
  34
  35 /**
  36  * @addtogroup mean
  37  * @{
  38  */
  39
  40
  41 /**
  42  * @brief Mean value of a Q15 vector.
  43  * @param[in]       *pSrc points to the input vector
  44  * @param[in]       blockSize length of the input vector
  45  * @param[out]      *pResult mean value returned here
  46  * @return none.
  47  *
  48  * @details
  49  * <b>Scaling and Overflow Behavior:</b>
  50  * \par
  51  * The function is implemented using a 32-bit internal accumulator.
  52  * The input is represented in 1.15 format and is accumulated in a 32-bit
  53  * accumulator in 17.15 format.
  54  * There is no risk of internal overflow with this approach, and the
  55  * full precision of intermediate result is preserved.
  56  * Finally, the accumulator is saturated and truncated to yield a result of 1.15 format.
  57  *
  58  */
  59
  60 void arm_mean_q15(
  61   q15_t * pSrc,
  62   uint32_t blockSize,
  63   q15_t * pResult)
  64 {
  65   q31_t sum = 0;                                 /* Temporary result storage */
  66   uint32_t blkCnt;                               /* loop counter */
  67
  68 #if defined (ARM_MATH_DSP)
  69   /* Run the below code for Cortex-M4 and Cortex-M3 */
  70
  71   q31_t in;
  72
  73   /*loop Unrolling */
  74   blkCnt = blockSize >> 2U;
  75
  76   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
  77    ** a second loop below computes the remaining 1 to 3 samples. */
  78   while (blkCnt > 0U)
  79   {
  80     /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
  81     in = *__SIMD32(pSrc)++;
  82     sum += ((in << 16U) >> 16U);
  83     sum +=  (in >> 16U);
  84     in = *__SIMD32(pSrc)++;
  85     sum += ((in << 16U) >> 16U);
  86     sum +=  (in >> 16U);
  87
  88     /* Decrement the loop counter */
  89     blkCnt--;
  90   }
  91
  92   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
  93    ** No loop unrolling is used. */
  94   blkCnt = blockSize % 0x4U;
  95
  96 #else
  97   /* Run the below code for Cortex-M0 */
  98
  99   /* Loop over blockSize number of values */
 100   blkCnt = blockSize;
 101
 102 #endif /* #if defined (ARM_MATH_DSP) */
 103
 104   while (blkCnt > 0U)
 105   {
 106     /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) */
 107     sum += *pSrc++;
 108
 109     /* Decrement the loop counter */
 110     blkCnt--;
 111   }
 112
 113   /* C = (A[0] + A[1] + A[2] + ... + A[blockSize-1]) / blockSize  */
 114   /* Store the result to the destination */
 115   *pResult = (q15_t) (sum / (q31_t)blockSize);
 116 }
 117
 118 /**
 119  * @} end of mean group
 120  */