lib/main/CMSIS/DSP/Source/StatisticsFunctions/arm_rms_q31.c

   1 /* ----------------------------------------------------------------------
   2  * Project:      CMSIS DSP Library
   3  * Title:        arm_rms_q31.c
   4  * Description:  Root Mean Square of the elements of a Q31 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 /**
  38  * @brief Root Mean Square of the elements of a Q31 vector.
  39  * @param[in]       *pSrc points to the input vector
  40  * @param[in]       blockSize length of the input vector
  41  * @param[out]      *pResult rms value returned here
  42  * @return none.
  43  *
  44  * @details
  45  * <b>Scaling and Overflow Behavior:</b>
  46  *
  47  *\par
  48  * The function is implemented using an internal 64-bit accumulator.
  49  * The input is represented in 1.31 format, and intermediate multiplication
  50  * yields a 2.62 format.
  51  * The accumulator maintains full precision of the intermediate multiplication results,
  52  * but provides only a single guard bit.
  53  * There is no saturation on intermediate additions.
  54  * If the accumulator overflows, it wraps around and distorts the result.
  55  * In order to avoid overflows completely, the input signal must be scaled down by
  56  * log2(blockSize) bits, as a total of blockSize additions are performed internally.
  57  * Finally, the 2.62 accumulator is right shifted by 31 bits to yield a 1.31 format value.
  58  *
  59  */
  60
  61 void arm_rms_q31(
  62   q31_t * pSrc,
  63   uint32_t blockSize,
  64   q31_t * pResult)
  65 {
  66   q63_t sum = 0;                                 /* accumulator */
  67   q31_t in;                                      /* Temporary variable to store the input */
  68   uint32_t blkCnt;                               /* loop counter */
  69
  70 #if defined (ARM_MATH_DSP)
  71   /* Run the below code for Cortex-M4 and Cortex-M3 */
  72
  73   q31_t in1, in2, in3, in4;                      /* Temporary input variables */
  74
  75   /*loop Unrolling */
  76   blkCnt = blockSize >> 2U;
  77
  78   /* First part of the processing with loop unrolling.  Compute 8 outputs at a time.
  79    ** a second loop below computes the remaining 1 to 7 samples. */
  80   while (blkCnt > 0U)
  81   {
  82     /* C = A[0] * A[0] + A[1] * A[1] + A[2] * A[2] + ... + A[blockSize-1] * A[blockSize-1] */
  83     /* Compute sum of the squares and then store the result in a temporary variable, sum */
  84     /* read two samples from source buffer */
  85     in1 = pSrc[0];
  86     in2 = pSrc[1];
  87
  88     /* calculate power and accumulate to accumulator */
  89     sum += (q63_t) in1 *in1;
  90     sum += (q63_t) in2 *in2;
  91
  92     /* read two samples from source buffer */
  93     in3 = pSrc[2];
  94     in4 = pSrc[3];
  95
  96     /* calculate power and accumulate to accumulator */
  97     sum += (q63_t) in3 *in3;
  98     sum += (q63_t) in4 *in4;
  99
 100
 101     /* update source buffer to process next samples */
 102     pSrc += 4U;
 103
 104     /* Decrement the loop counter */
 105     blkCnt--;
 106   }
 107
 108   /* If the blockSize is not a multiple of 8, compute any remaining output samples here.
 109    ** No loop unrolling is used. */
 110   blkCnt = blockSize % 0x4U;
 111
 112 #else
 113   /* Run the below code for Cortex-M0 */
 114
 115   blkCnt = blockSize;
 116
 117 #endif /* #if defined (ARM_MATH_DSP) */
 118
 119   while (blkCnt > 0U)
 120   {
 121     /* C = A[0] * A[0] + A[1] * A[1] + A[2] * A[2] + ... + A[blockSize-1] * A[blockSize-1] */
 122     /* Compute sum of the squares and then store the results in a temporary variable, sum */
 123     in = *pSrc++;
 124     sum += (q63_t) in *in;
 125
 126     /* Decrement the loop counter */
 127     blkCnt--;
 128   }
 129
 130   /* Convert data in 2.62 to 1.31 by 31 right shifts and saturate */
 131   /* Compute Rms and store the result in the destination vector */
 132   arm_sqrt_q31(clip_q63_to_q31((sum / (q63_t) blockSize) >> 31), pResult);
 133 }
 134
 135 /**
 136  * @} end of RMS group
 137  */