lib/main/CMSIS/DSP/Source/MatrixFunctions/arm_mat_sub_q15.c

   1 /* ----------------------------------------------------------------------
   2  * Project:      CMSIS DSP Library
   3  * Title:        arm_mat_sub_q15.c
   4  * Description:  Q15 Matrix subtraction
   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 MatrixSub
  37  * @{
  38  */
  39
  40 /**
  41  * @brief Q15 matrix subtraction.
  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_sub_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   uint32_t numSamples;                           /* total number of elements in the matrix */
  63   uint32_t blkCnt;                               /* loop counters  */
  64   arm_status status;                             /* status of matrix subtraction  */
  65
  66
  67 #ifdef ARM_MATH_MATRIX_CHECK
  68
  69
  70   /* Check for matrix mismatch condition */
  71   if ((pSrcA->numRows != pSrcB->numRows) ||
  72      (pSrcA->numCols != pSrcB->numCols) ||
  73      (pSrcA->numRows != pDst->numRows) || (pSrcA->numCols != pDst->numCols))
  74   {
  75     /* Set status as ARM_MATH_SIZE_MISMATCH */
  76     status = ARM_MATH_SIZE_MISMATCH;
  77   }
  78   else
  79 #endif /*    #ifdef ARM_MATH_MATRIX_CHECK    */
  80
  81   {
  82     /* Total number of samples in the input matrix */
  83     numSamples = (uint32_t) pSrcA->numRows * pSrcA->numCols;
  84
  85 #if defined (ARM_MATH_DSP)
  86
  87     /* Run the below code for Cortex-M4 and Cortex-M3 */
  88
  89     /* Apply loop unrolling */
  90     blkCnt = numSamples >> 2U;
  91
  92     /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
  93      ** a second loop below computes the remaining 1 to 3 samples. */
  94     while (blkCnt > 0U)
  95     {
  96       /* C(m,n) = A(m,n) - B(m,n) */
  97       /* Subtract, Saturate and then store the results in the destination buffer. */
  98       *__SIMD32(pOut)++ = __QSUB16(*__SIMD32(pInA)++, *__SIMD32(pInB)++);
  99       *__SIMD32(pOut)++ = __QSUB16(*__SIMD32(pInA)++, *__SIMD32(pInB)++);
 100
 101       /* Decrement the loop counter */
 102       blkCnt--;
 103     }
 104
 105     /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
 106      ** No loop unrolling is used. */
 107     blkCnt = numSamples % 0x4U;
 108
 109     while (blkCnt > 0U)
 110     {
 111       /* C(m,n) = A(m,n) - B(m,n) */
 112       /* Subtract and then store the results in the destination buffer. */
 113       *pOut++ = (q15_t) __QSUB16(*pInA++, *pInB++);
 114
 115       /* Decrement the loop counter */
 116       blkCnt--;
 117     }
 118
 119 #else
 120
 121     /* Run the below code for Cortex-M0 */
 122
 123     /* Initialize blkCnt with number of samples */
 124     blkCnt = numSamples;
 125
 126     while (blkCnt > 0U)
 127     {
 128       /* C(m,n) = A(m,n) - B(m,n) */
 129       /* Subtract and then store the results in the destination buffer. */
 130       *pOut++ = (q15_t) __SSAT(((q31_t) * pInA++ - *pInB++), 16);
 131
 132       /* Decrement the loop counter */
 133       blkCnt--;
 134     }
 135
 136 #endif /* #if defined (ARM_MATH_DSP) */
 137
 138     /* Set status as ARM_MATH_SUCCESS */
 139     status = ARM_MATH_SUCCESS;
 140   }
 141
 142   /* Return to application */
 143   return (status);
 144 }
 145
 146 /**
 147  * @} end of MatrixSub group
 148  */