1  /* ----------------------------------------------------------------------
  2   * Project:      CMSIS DSP Library
  3   * Title:        arm_cmplx_mag_squared_q31.c
  4   * Description:  Q31 complex magnitude squared
  5   *
  6   * $Date:        23 April 2021
  7   * $Revision:    V1.9.0
  8   *
  9   * Target Processor: Cortex-M and Cortex-A cores
 10   * -------------------------------------------------------------------- */
 11  /*
 12   * Copyright (C) 2010-2021 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 "dsp/complex_math_functions.h"
 30  
 31  /**
 32    @ingroup groupCmplxMath
 33   */
 34  
 35  /**
 36    @addtogroup cmplx_mag_squared
 37    @{
 38   */
 39  
 40  /**
 41    @brief         Q31 complex magnitude squared.
 42    @param[in]     pSrc        points to input vector
 43    @param[out]    pDst        points to output vector
 44    @param[in]     numSamples  number of samples in each vector
 45    @return        none
 46  
 47    @par           Scaling and Overflow Behavior
 48                     The function implements 1.31 by 1.31 multiplications and finally output is converted into 3.29 format.
 49                     Input down scaling is not required.
 50   */
 51  
 52  #if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
 53  
 54  void arm_cmplx_mag_squared_q31(
 55    const q31_t * pSrc,
 56          q31_t * pDst,
 57          uint32_t numSamples)
 58  {
 59      int32_t blockSize = numSamples;  /* loop counters */
 60      uint32_t  blkCnt;           /* loop counters */
 61      q31x4x2_t vecSrc;
 62      q31x4_t vReal, vImag;
 63      q31x4_t vMagSq;
 64      q31_t real, imag;                              /* Temporary input variables */
 65      q31_t acc0, acc1;                              /* Accumulators */
 66  
 67      /* Compute 4 complex samples at a time */
 68      blkCnt = blockSize >> 2;
 69      while (blkCnt > 0U)
 70      {
 71          vecSrc = vld2q(pSrc);
 72          vReal = vmulhq(vecSrc.val[0], vecSrc.val[0]);
 73          vImag = vmulhq(vecSrc.val[1], vecSrc.val[1]);
 74          vMagSq = vqaddq(vReal, vImag);
 75          vMagSq = vshrq(vMagSq, 1);
 76  
 77          vst1q(pDst, vMagSq);
 78  
 79          pSrc += 8;
 80          pDst += 4;
 81          /*
 82           * Decrement the blkCnt loop counter
 83           * Advance vector source and destination pointers
 84           */
 85          blkCnt --;
 86      } 
 87  
 88      /* Tail */
 89      blkCnt = blockSize & 3;
 90      while (blkCnt > 0U)
 91      {
 92        /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
 93    
 94        real = *pSrc++;
 95        imag = *pSrc++;
 96        acc0 = (q31_t) (((q63_t) real * real) >> 33);
 97        acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
 98    
 99        /* store result in 3.29 format in destination buffer. */
100        *pDst++ = acc0 + acc1;
101    
102        /* Decrement loop counter */
103        blkCnt--;
104      }
105  }
106  
107  #else
108  void arm_cmplx_mag_squared_q31(
109    const q31_t * pSrc,
110          q31_t * pDst,
111          uint32_t numSamples)
112  {
113          uint32_t blkCnt;                               /* Loop counter */
114          q31_t real, imag;                              /* Temporary input variables */
115          q31_t acc0, acc1;                              /* Accumulators */
116  
117  #if defined (ARM_MATH_LOOPUNROLL)
118  
119    /* Loop unrolling: Compute 4 outputs at a time */
120    blkCnt = numSamples >> 2U;
121  
122    while (blkCnt > 0U)
123    {
124      /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
125  
126      real = *pSrc++;
127      imag = *pSrc++;
128      acc0 = (q31_t) (((q63_t) real * real) >> 33);
129      acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
130      /* store the result in 3.29 format in the destination buffer. */
131      *pDst++ = acc0 + acc1;
132  
133      real = *pSrc++;
134      imag = *pSrc++;
135      acc0 = (q31_t) (((q63_t) real * real) >> 33);
136      acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
137      *pDst++ = acc0 + acc1;
138  
139      real = *pSrc++;
140      imag = *pSrc++;
141      acc0 = (q31_t) (((q63_t) real * real) >> 33);
142      acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
143      *pDst++ = acc0 + acc1;
144  
145      real = *pSrc++;
146      imag = *pSrc++;
147      acc0 = (q31_t) (((q63_t) real * real) >> 33);
148      acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
149      *pDst++ = acc0 + acc1;
150  
151      /* Decrement loop counter */
152      blkCnt--;
153    }
154  
155    /* Loop unrolling: Compute remaining outputs */
156    blkCnt = numSamples % 0x4U;
157  
158  #else
159  
160    /* Initialize blkCnt with number of samples */
161    blkCnt = numSamples;
162  
163  #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
164  
165    while (blkCnt > 0U)
166    {
167      /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
168  
169      real = *pSrc++;
170      imag = *pSrc++;
171      acc0 = (q31_t) (((q63_t) real * real) >> 33);
172      acc1 = (q31_t) (((q63_t) imag * imag) >> 33);
173  
174      /* store result in 3.29 format in destination buffer. */
175      *pDst++ = acc0 + acc1;
176  
177      /* Decrement loop counter */
178      blkCnt--;
179    }
180  
181  }
182  
183  #endif /* defined(ARM_MATH_MVEI) */
184  
185  /**
186    @} end of cmplx_mag_squared group
187   */