lib/main/CMSIS/DSP/Source/ComplexMathFunctions/arm_cmplx_dot_prod_f32.c

   1 /* ----------------------------------------------------------------------
   2  * Project:      CMSIS DSP Library
   3  * Title:        arm_cmplx_dot_prod_f32.c
   4  * Description:  Floating-point complex dot product
   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 groupCmplxMath
  33  */
  34
  35 /**
  36  * @defgroup cmplx_dot_prod Complex Dot Product
  37  *
  38  * Computes the dot product of two complex vectors.
  39  * The vectors are multiplied element-by-element and then summed.
  40  *
  41  * The <code>pSrcA</code> points to the first complex input vector and
  42  * <code>pSrcB</code> points to the second complex input vector.
  43  * <code>numSamples</code> specifies the number of complex samples
  44  * and the data in each array is stored in an interleaved fashion
  45  * (real, imag, real, imag, ...).
  46  * Each array has a total of <code>2*numSamples</code> values.
  47  *
  48  * The underlying algorithm is used:
  49  * <pre>
  50  * realResult=0;
  51  * imagResult=0;
  52  * for(n=0; n<numSamples; n++) {
  53  *     realResult += pSrcA[(2*n)+0]*pSrcB[(2*n)+0] - pSrcA[(2*n)+1]*pSrcB[(2*n)+1];
  54  *     imagResult += pSrcA[(2*n)+0]*pSrcB[(2*n)+1] + pSrcA[(2*n)+1]*pSrcB[(2*n)+0];
  55  * }
  56  * </pre>
  57  *
  58  * There are separate functions for floating-point, Q15, and Q31 data types.
  59  */
  60
  61 /**
  62  * @addtogroup cmplx_dot_prod
  63  * @{
  64  */
  65
  66 /**
  67  * @brief  Floating-point complex dot product
  68  * @param  *pSrcA points to the first input vector
  69  * @param  *pSrcB points to the second input vector
  70  * @param  numSamples number of complex samples in each vector
  71  * @param  *realResult real part of the result returned here
  72  * @param  *imagResult imaginary part of the result returned here
  73  * @return none.
  74  */
  75
  76 void arm_cmplx_dot_prod_f32(
  77   float32_t * pSrcA,
  78   float32_t * pSrcB,
  79   uint32_t numSamples,
  80   float32_t * realResult,
  81   float32_t * imagResult)
  82 {
  83   float32_t real_sum = 0.0f, imag_sum = 0.0f;    /* Temporary result storage */
  84   float32_t a0,b0,c0,d0;
  85
  86 #if defined (ARM_MATH_DSP)
  87
  88   /* Run the below code for Cortex-M4 and Cortex-M3 */
  89   uint32_t blkCnt;                               /* loop counter */
  90
  91   /*loop Unrolling */
  92   blkCnt = numSamples >> 2U;
  93
  94   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
  95    ** a second loop below computes the remaining 1 to 3 samples. */
  96   while (blkCnt > 0U)
  97   {
  98       a0 = *pSrcA++;
  99       b0 = *pSrcA++;
 100       c0 = *pSrcB++;
 101       d0 = *pSrcB++;
 102
 103       real_sum += a0 * c0;
 104       imag_sum += a0 * d0;
 105       real_sum -= b0 * d0;
 106       imag_sum += b0 * c0;
 107
 108       a0 = *pSrcA++;
 109       b0 = *pSrcA++;
 110       c0 = *pSrcB++;
 111       d0 = *pSrcB++;
 112
 113       real_sum += a0 * c0;
 114       imag_sum += a0 * d0;
 115       real_sum -= b0 * d0;
 116       imag_sum += b0 * c0;
 117
 118       a0 = *pSrcA++;
 119       b0 = *pSrcA++;
 120       c0 = *pSrcB++;
 121       d0 = *pSrcB++;
 122
 123       real_sum += a0 * c0;
 124       imag_sum += a0 * d0;
 125       real_sum -= b0 * d0;
 126       imag_sum += b0 * c0;
 127
 128       a0 = *pSrcA++;
 129       b0 = *pSrcA++;
 130       c0 = *pSrcB++;
 131       d0 = *pSrcB++;
 132
 133       real_sum += a0 * c0;
 134       imag_sum += a0 * d0;
 135       real_sum -= b0 * d0;
 136       imag_sum += b0 * c0;
 137
 138       /* Decrement the loop counter */
 139       blkCnt--;
 140   }
 141
 142   /* If the numSamples is not a multiple of 4, compute any remaining output samples here.
 143    ** No loop unrolling is used. */
 144   blkCnt = numSamples & 0x3U;
 145
 146   while (blkCnt > 0U)
 147   {
 148       a0 = *pSrcA++;
 149       b0 = *pSrcA++;
 150       c0 = *pSrcB++;
 151       d0 = *pSrcB++;
 152
 153       real_sum += a0 * c0;
 154       imag_sum += a0 * d0;
 155       real_sum -= b0 * d0;
 156       imag_sum += b0 * c0;
 157
 158       /* Decrement the loop counter */
 159       blkCnt--;
 160   }
 161
 162 #else
 163
 164   /* Run the below code for Cortex-M0 */
 165
 166   while (numSamples > 0U)
 167   {
 168       a0 = *pSrcA++;
 169       b0 = *pSrcA++;
 170       c0 = *pSrcB++;
 171       d0 = *pSrcB++;
 172
 173       real_sum += a0 * c0;
 174       imag_sum += a0 * d0;
 175       real_sum -= b0 * d0;
 176       imag_sum += b0 * c0;
 177
 178       /* Decrement the loop counter */
 179       numSamples--;
 180   }
 181
 182 #endif /* #if defined (ARM_MATH_DSP) */
 183
 184   /* Store the real and imaginary results in the destination buffers */
 185   *realResult = real_sum;
 186   *imagResult = imag_sum;
 187 }
 188
 189 /**
 190  * @} end of cmplx_dot_prod group
 191  */