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_add_q31.c
  4.  * Description:  Q31 vector addition
  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.  * @addtogroup BasicAdd
  37.  * @{
  38.  */
  39.  
  40.  
  41. /**
  42.  * @brief Q31 vector addition.
  43.  * @param[in]       *pSrcA points to the first input vector
  44.  * @param[in]       *pSrcB points to the second input vector
  45.  * @param[out]      *pDst points to the output vector
  46.  * @param[in]       blockSize number of samples in each vector
  47.  * @return none.
  48.  *
  49.  * <b>Scaling and Overflow Behavior:</b>
  50.  * \par
  51.  * The function uses saturating arithmetic.
  52.  * Results outside of the allowable Q31 range[0x80000000 0x7FFFFFFF] will be saturated.
  53.  */
  54.  
  55. void arm_add_q31(
  56.   q31_t * pSrcA,
  57.   q31_t * pSrcB,
  58.   q31_t * pDst,
  59.   uint32_t blockSize)
  60. {
  61.   uint32_t blkCnt;                               /* loop counter */
  62.  
  63. #if defined (ARM_MATH_DSP)
  64.  
  65. /* Run the below code for Cortex-M4 and Cortex-M3 */
  66.   q31_t inA1, inA2, inA3, inA4;
  67.   q31_t inB1, inB2, inB3, inB4;
  68.  
  69.   /*loop Unrolling */
  70.   blkCnt = blockSize >> 2U;
  71.  
  72.   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
  73.    ** a second loop below computes the remaining 1 to 3 samples. */
  74.   while (blkCnt > 0U)
  75.   {
  76.     /* C = A + B */
  77.     /* Add and then store the results in the destination buffer. */
  78.     inA1 = *pSrcA++;
  79.     inA2 = *pSrcA++;
  80.     inB1 = *pSrcB++;
  81.     inB2 = *pSrcB++;
  82.  
  83.     inA3 = *pSrcA++;
  84.     inA4 = *pSrcA++;
  85.     inB3 = *pSrcB++;
  86.     inB4 = *pSrcB++;
  87.  
  88.     *pDst++ = __QADD(inA1, inB1);
  89.     *pDst++ = __QADD(inA2, inB2);
  90.     *pDst++ = __QADD(inA3, inB3);
  91.     *pDst++ = __QADD(inA4, inB4);
  92.  
  93.     /* Decrement the loop counter */
  94.     blkCnt--;
  95.   }
  96.  
  97.   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
  98.    ** No loop unrolling is used. */
  99.   blkCnt = blockSize % 0x4U;
  100.  
  101.   while (blkCnt > 0U)
  102.   {
  103.     /* C = A + B */
  104.     /* Add and then store the results in the destination buffer. */
  105.     *pDst++ = __QADD(*pSrcA++, *pSrcB++);
  106.  
  107.     /* Decrement the loop counter */
  108.     blkCnt--;
  109.   }
  110.  
  111. #else
  112.  
  113.   /* Run the below code for Cortex-M0 */
  114.  
  115.  
  116.  
  117.   /* Initialize blkCnt with number of samples */
  118.   blkCnt = blockSize;
  119.  
  120.   while (blkCnt > 0U)
  121.   {
  122.     /* C = A + B */
  123.     /* Add and then store the results in the destination buffer. */
  124.     *pDst++ = (q31_t) clip_q63_to_q31((q63_t) * pSrcA++ + *pSrcB++);
  125.  
  126.     /* Decrement the loop counter */
  127.     blkCnt--;
  128.   }
  129.  
  130. #endif /* #if defined (ARM_MATH_DSP) */
  131.  
  132. }
  133.  
  134. /**
  135.  * @} end of BasicAdd group
  136.  */
  137.