lib/main/CMSIS/DSP/Source/StatisticsFunctions/arm_rms_q15.c

   1 /* ----------------------------------------------------------------------
   2  * Project:      CMSIS DSP Library
   3  * Title:        arm_rms_q15.c
   4  * Description:  Root Mean Square of the elements 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  * @addtogroup RMS
  33  * @{
  34  */
  35
  36 /**
  37  * @brief Root Mean Square of the elements of a Q15 vector.
  38  * @param[in]       *pSrc points to the input vector
  39  * @param[in]       blockSize length of the input vector
  40  * @param[out]      *pResult rms value returned here
  41  * @return none.
  42  *
  43  * @details
  44  * <b>Scaling and Overflow Behavior:</b>
  45  *
  46  * \par
  47  * The function is implemented using a 64-bit internal accumulator.
  48  * The input is represented in 1.15 format.
  49  * Intermediate multiplication yields a 2.30 format, and this
  50  * result is added without saturation to a 64-bit accumulator in 34.30 format.
  51  * With 33 guard bits in the accumulator, there is no risk of overflow, and the
  52  * full precision of the intermediate multiplication is preserved.
  53  * Finally, the 34.30 result is truncated to 34.15 format by discarding the lower
  54  * 15 bits, and then saturated to yield a result in 1.15 format.
  55  *
  56  */
  57
  58 void arm_rms_q15(
  59   q15_t * pSrc,
  60   uint32_t blockSize,
  61   q15_t * pResult)
  62 {
  63   q63_t sum = 0;                                 /* accumulator */
  64
  65 #if defined (ARM_MATH_DSP)
  66   /* Run the below code for Cortex-M4 and Cortex-M3 */
  67
  68   q31_t in;                                      /* temporary variable to store the input value */
  69   q15_t in1;                                     /* temporary variable to store the input value */
  70   uint32_t blkCnt;                               /* loop counter */
  71
  72   /* loop Unrolling */
  73   blkCnt = blockSize >> 2U;
  74
  75   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
  76    ** a second loop below computes the remaining 1 to 3 samples. */
  77   while (blkCnt > 0U)
  78   {
  79     /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */
  80     /* Compute sum of the squares and then store the results in a temporary variable, sum */
  81     in = *__SIMD32(pSrc)++;
  82     sum = __SMLALD(in, in, sum);
  83     in = *__SIMD32(pSrc)++;
  84     sum = __SMLALD(in, in, sum);
  85
  86     /* Decrement the loop counter */
  87     blkCnt--;
  88   }
  89
  90   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
  91    ** No loop unrolling is used. */
  92   blkCnt = blockSize % 0x4U;
  93
  94   while (blkCnt > 0U)
  95   {
  96     /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */
  97     /* Compute sum of the squares and then store the results in a temporary variable, sum */
  98     in1 = *pSrc++;
  99     sum = __SMLALD(in1, in1, sum);
 100
 101     /* Decrement the loop counter */
 102     blkCnt--;
 103   }
 104
 105   /* Truncating and saturating the accumulator to 1.15 format */
 106   /* Store the result in the destination */
 107   arm_sqrt_q15(__SSAT((sum / (q63_t)blockSize) >> 15, 16), pResult);
 108
 109 #else
 110   /* Run the below code for Cortex-M0 */
 111
 112   q15_t in;                                      /* temporary variable to store the input value */
 113   uint32_t blkCnt;                               /* loop counter */
 114
 115   /* Loop over blockSize number of values */
 116   blkCnt = blockSize;
 117
 118   while (blkCnt > 0U)
 119   {
 120     /* C = (A[0] * A[0] + A[1] * A[1] + ... + A[blockSize-1] * A[blockSize-1]) */
 121     /* Compute sum of the squares and then store the results in a temporary variable, sum */
 122     in = *pSrc++;
 123     sum += ((q31_t) in * in);
 124
 125     /* Decrement the loop counter */
 126     blkCnt--;
 127   }
 128
 129   /* Truncating and saturating the accumulator to 1.15 format */
 130   /* Store the result in the destination */
 131   arm_sqrt_q15(__SSAT((sum / (q63_t)blockSize) >> 15, 16), pResult);
 132
 133 #endif /* #if defined (ARM_MATH_DSP) */
 134
 135 }
 136
 137 /**
 138  * @} end of RMS group
 139  */