lib/main/CMSIS/DSP/Source/MatrixFunctions/arm_mat_add_q15.c

   1 /* ----------------------------------------------------------------------
   2  * Project:      CMSIS DSP Library
   3  * Title:        arm_mat_add_q15.c
   4  * Description:  Q15 matrix 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 groupMatrix
  33  */
  34
  35 /**
  36  * @addtogroup MatrixAdd
  37  * @{
  38  */
  39
  40 /**
  41  * @brief Q15 matrix addition.
  42  * @param[in]       *pSrcA points to the first input matrix structure
  43  * @param[in]       *pSrcB points to the second input matrix structure
  44  * @param[out]      *pDst points to output matrix structure
  45  * @return              The function returns either
  46  * <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.
  47  *
  48  * <b>Scaling and Overflow Behavior:</b>
  49  * \par
  50  * The function uses saturating arithmetic.
  51  * Results outside of the allowable Q15 range [0x8000 0x7FFF] will be saturated.
  52  */
  53
  54 arm_status arm_mat_add_q15(
  55   const arm_matrix_instance_q15 * pSrcA,
  56   const arm_matrix_instance_q15 * pSrcB,
  57   arm_matrix_instance_q15 * pDst)
  58 {
  59   q15_t *pInA = pSrcA->pData;                    /* input data matrix pointer A  */
  60   q15_t *pInB = pSrcB->pData;                    /* input data matrix pointer B */
  61   q15_t *pOut = pDst->pData;                     /* output data matrix pointer */
  62   uint16_t numSamples;                           /* total number of elements in the matrix  */
  63   uint32_t blkCnt;                               /* loop counters  */
  64   arm_status status;                             /* status of matrix addition  */
  65
  66 #ifdef ARM_MATH_MATRIX_CHECK
  67
  68
  69   /* Check for matrix mismatch condition */
  70   if ((pSrcA->numRows != pSrcB->numRows) ||
  71      (pSrcA->numCols != pSrcB->numCols) ||
  72      (pSrcA->numRows != pDst->numRows) || (pSrcA->numCols != pDst->numCols))
  73   {
  74     /* Set status as ARM_MATH_SIZE_MISMATCH */
  75     status = ARM_MATH_SIZE_MISMATCH;
  76   }
  77   else
  78 #endif /*    #ifdef ARM_MATH_MATRIX_CHECK    */
  79
  80   {
  81     /* Total number of samples in the input matrix */
  82     numSamples = (uint16_t) (pSrcA->numRows * pSrcA->numCols);
  83
  84 #if defined (ARM_MATH_DSP)
  85
  86     /* Run the below code for Cortex-M4 and Cortex-M3 */
  87
  88     /* Loop unrolling */
  89     blkCnt = (uint32_t) numSamples >> 2U;
  90
  91     /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
  92      ** a second loop below computes the remaining 1 to 3 samples. */
  93     while (blkCnt > 0U)
  94     {
  95       /* C(m,n) = A(m,n) + B(m,n) */
  96       /* Add, Saturate and then store the results in the destination buffer. */
  97       *__SIMD32(pOut)++ = __QADD16(*__SIMD32(pInA)++, *__SIMD32(pInB)++);
  98       *__SIMD32(pOut)++ = __QADD16(*__SIMD32(pInA)++, *__SIMD32(pInB)++);
  99
 100       /* Decrement the loop counter */
 101       blkCnt--;
 102     }
 103
 104     /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
 105      ** No loop unrolling is used. */
 106     blkCnt = (uint32_t) numSamples % 0x4U;
 107
 108     /* q15 pointers of input and output are initialized */
 109
 110     while (blkCnt > 0U)
 111     {
 112       /* C(m,n) = A(m,n) + B(m,n) */
 113       /* Add, Saturate and then store the results in the destination buffer. */
 114       *pOut++ = (q15_t) __QADD16(*pInA++, *pInB++);
 115
 116       /* Decrement the loop counter */
 117       blkCnt--;
 118     }
 119
 120 #else
 121
 122     /* Run the below code for Cortex-M0 */
 123
 124     /* Initialize blkCnt with number of samples */
 125     blkCnt = (uint32_t) numSamples;
 126
 127
 128     /* q15 pointers of input and output are initialized */
 129     while (blkCnt > 0U)
 130     {
 131       /* C(m,n) = A(m,n) + B(m,n) */
 132       /* Add, Saturate and then store the results in the destination buffer. */
 133       *pOut++ = (q15_t) __SSAT(((q31_t) * pInA++ + *pInB++), 16);
 134
 135       /* Decrement the loop counter */
 136       blkCnt--;
 137     }
 138
 139 #endif /* #if defined (ARM_MATH_DSP) */
 140
 141     /* set status as ARM_MATH_SUCCESS */
 142     status = ARM_MATH_SUCCESS;
 143   }
 144
 145   /* Return to application */
 146   return (status);
 147 }
 148
 149 /**
 150  * @} end of MatrixAdd group
 151  */