Subversion Repositories dashGPS

Rev

Rev 2 | Blame | Compare with Previous | Last modification | View Log | Download | RSS feed

  1. /* ----------------------------------------------------------------------
  2.  * Project:      CMSIS DSP Library
  3.  * Title:        arm_dot_prod_f32.c
  4.  * Description:  Floating-point 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 groupMath
  33.  */
  34.  
  35. /**
  36.  * @defgroup dot_prod Vector Dot Product
  37.  *
  38.  * Computes the dot product of two vectors.
  39.  * The vectors are multiplied element-by-element and then summed.
  40.  *
  41.  * <pre>
  42.  *     sum = pSrcA[0]*pSrcB[0] + pSrcA[1]*pSrcB[1] + ... + pSrcA[blockSize-1]*pSrcB[blockSize-1]
  43.  * </pre>
  44.  *
  45.  * There are separate functions for floating-point, Q7, Q15, and Q31 data types.
  46.  */
  47.  
  48. /**
  49.  * @addtogroup dot_prod
  50.  * @{
  51.  */
  52.  
  53. /**
  54.  * @brief Dot product of floating-point vectors.
  55.  * @param[in]       *pSrcA points to the first input vector
  56.  * @param[in]       *pSrcB points to the second input vector
  57.  * @param[in]       blockSize number of samples in each vector
  58.  * @param[out]      *result output result returned here
  59.  * @return none.
  60.  */
  61.  
  62.  
  63. void arm_dot_prod_f32(
  64.   float32_t * pSrcA,
  65.   float32_t * pSrcB,
  66.   uint32_t blockSize,
  67.   float32_t * result)
  68. {
  69.   float32_t sum = 0.0f;                          /* Temporary result storage */
  70.   uint32_t blkCnt;                               /* loop counter */
  71.  
  72.  
  73. #if defined (ARM_MATH_DSP)
  74.  
  75. /* Run the below code for Cortex-M4 and Cortex-M3 */
  76.   /*loop Unrolling */
  77.   blkCnt = blockSize >> 2U;
  78.  
  79.   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
  80.    ** a second loop below computes the remaining 1 to 3 samples. */
  81.   while (blkCnt > 0U)
  82.   {
  83.     /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
  84.     /* Calculate dot product and then store the result in a temporary buffer */
  85.     sum += (*pSrcA++) * (*pSrcB++);
  86.     sum += (*pSrcA++) * (*pSrcB++);
  87.     sum += (*pSrcA++) * (*pSrcB++);
  88.     sum += (*pSrcA++) * (*pSrcB++);
  89.  
  90.     /* Decrement the loop counter */
  91.     blkCnt--;
  92.   }
  93.  
  94.   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
  95.    ** No loop unrolling is used. */
  96.   blkCnt = blockSize % 0x4U;
  97.  
  98. #else
  99.  
  100.   /* Run the below code for Cortex-M0 */
  101.  
  102.   /* Initialize blkCnt with number of samples */
  103.   blkCnt = blockSize;
  104.  
  105. #endif /* #if defined (ARM_MATH_DSP) */
  106.  
  107.  
  108.   while (blkCnt > 0U)
  109.   {
  110.     /* C = A[0]* B[0] + A[1]* B[1] + A[2]* B[2] + .....+ A[blockSize-1]* B[blockSize-1] */
  111.     /* Calculate dot product and then store the result in a temporary buffer. */
  112.     sum += (*pSrcA++) * (*pSrcB++);
  113.  
  114.     /* Decrement the loop counter */
  115.     blkCnt--;
  116.   }
  117.   /* Store the result back in the destination buffer */
  118.   *result = sum;
  119. }
  120.  
  121. /**
  122.  * @} end of dot_prod group
  123.  */
  124.