1  /* ----------------------------------------------------------------------
  2   * Project:      CMSIS DSP Library
  3   * Title:        arm_add_q15.c
  4   * Description:  Q15 vector addition
  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/basic_math_functions.h"
 30  
 31  /**
 32    @ingroup groupMath
 33   */
 34  
 35  /**
 36    @addtogroup BasicAdd
 37    @{
 38   */
 39  
 40  /**
 41    @brief         Q15 vector addition.
 42    @param[in]     pSrcA      points to the first input vector
 43    @param[in]     pSrcB      points to the second input vector
 44    @param[out]    pDst       points to the output vector
 45    @param[in]     blockSize  number of samples in each vector
 46    @return        none
 47  
 48    @par           Scaling and Overflow Behavior
 49                     The function uses saturating arithmetic.
 50                     Results outside of the allowable Q15 range [0x8000 0x7FFF] are saturated.
 51   */
 52  
 53  #if defined(ARM_MATH_MVEI) && !defined(ARM_MATH_AUTOVECTORIZE)
 54  
 55  #include "arm_helium_utils.h"
 56  
 57  void arm_add_q15(
 58      const q15_t * pSrcA,
 59      const q15_t * pSrcB,
 60      q15_t * pDst,
 61      uint32_t blockSize)
 62  {
 63      uint32_t  blkCnt;           /* loop counters */
 64      q15x8_t vecA;
 65      q15x8_t vecB;
 66  
 67      /* Compute 8 outputs at a time */
 68      blkCnt = blockSize >> 3;
 69      while (blkCnt > 0U)
 70      {
 71          /*
 72           * C = A + B
 73           * Add and then store the results in the destination buffer.
 74           */
 75          vecA = vld1q(pSrcA);
 76          vecB = vld1q(pSrcB);
 77          vst1q(pDst, vqaddq(vecA, vecB));
 78          /*
 79           * Decrement the blockSize loop counter
 80           */
 81          blkCnt--;
 82          /*
 83           * advance vector source and destination pointers
 84           */
 85          pSrcA  += 8;
 86          pSrcB  += 8;
 87          pDst   += 8;
 88      }
 89      /*
 90       * tail
 91       */
 92      blkCnt = blockSize & 7;
 93      if (blkCnt > 0U)
 94      {
 95          mve_pred16_t p0 = vctp16q(blkCnt);
 96          vecA = vld1q(pSrcA);
 97          vecB = vld1q(pSrcB);
 98          vstrhq_p(pDst, vqaddq(vecA, vecB), p0);
 99      }
100  }
101  
102  #else
103  void arm_add_q15(
104    const q15_t * pSrcA,
105    const q15_t * pSrcB,
106          q15_t * pDst,
107          uint32_t blockSize)
108  {
109          uint32_t blkCnt;                               /* Loop counter */
110  
111  #if defined (ARM_MATH_LOOPUNROLL)
112  
113  #if defined (ARM_MATH_DSP)
114    q31_t inA1, inA2;
115    q31_t inB1, inB2;
116  #endif
117  
118    /* Loop unrolling: Compute 4 outputs at a time */
119    blkCnt = blockSize >> 2U;
120  
121    while (blkCnt > 0U)
122    {
123      /* C = A + B */
124  
125  #if defined (ARM_MATH_DSP)
126      /* read 2 times 2 samples at a time from sourceA */
127      inA1 = read_q15x2_ia (&pSrcA);
128      inA2 = read_q15x2_ia (&pSrcA);
129      /* read 2 times 2 samples at a time from sourceB */
130      inB1 = read_q15x2_ia (&pSrcB);
131      inB2 = read_q15x2_ia (&pSrcB);
132  
133      /* Add and store 2 times 2 samples at a time */
134      write_q15x2_ia (&pDst, __QADD16(inA1, inB1));
135      write_q15x2_ia (&pDst, __QADD16(inA2, inB2));
136  #else
137      *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
138      *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
139      *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
140      *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
141  #endif
142  
143      /* Decrement loop counter */
144      blkCnt--;
145    }
146  
147    /* Loop unrolling: Compute remaining outputs */
148    blkCnt = blockSize % 0x4U;
149  
150  #else
151  
152    /* Initialize blkCnt with number of samples */
153    blkCnt = blockSize;
154  
155  #endif /* #if defined (ARM_MATH_LOOPUNROLL) */
156  
157    while (blkCnt > 0U)
158    {
159      /* C = A + B */
160  
161      /* Add and store result in destination buffer. */
162  #if defined (ARM_MATH_DSP)
163      *pDst++ = (q15_t) __QADD16(*pSrcA++, *pSrcB++);
164  #else
165      *pDst++ = (q15_t) __SSAT(((q31_t) *pSrcA++ + *pSrcB++), 16);
166  #endif
167  
168      /* Decrement loop counter */
169      blkCnt--;
170    }
171  
172  }
173  #endif /* defined(ARM_MATH_MVEI) */
174  /**
175    @} end of BasicAdd group
176   */