arm_mat_add_q15.c

00001 /* ----------------------------------------------------------------------    
00002 * Copyright (C) 2010-2014 ARM Limited. All rights reserved.    
00003 *    
00004 * $Date:        12. March 2014 
00005 * $Revision:    V1.4.3
00006 *    
00007 * Project:      CMSIS DSP Library    
00008 * Title:        arm_mat_add_q15.c    
00009 *    
00010 * Description:  Q15 matrix addition    
00011 *    
00012 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
00013 *  
00014 * Redistribution and use in source and binary forms, with or without 
00015 * modification, are permitted provided that the following conditions
00016 * are met:
00017 *   - Redistributions of source code must retain the above copyright
00018 *     notice, this list of conditions and the following disclaimer.
00019 *   - Redistributions in binary form must reproduce the above copyright
00020 *     notice, this list of conditions and the following disclaimer in
00021 *     the documentation and/or other materials provided with the 
00022 *     distribution.
00023 *   - Neither the name of ARM LIMITED nor the names of its contributors
00024 *     may be used to endorse or promote products derived from this
00025 *     software without specific prior written permission.
00026 *
00027 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
00028 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
00029 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
00030 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 
00031 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
00032 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
00033 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
00034 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
00035 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
00036 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
00037 * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
00038 * POSSIBILITY OF SUCH DAMAGE.    
00039 * -------------------------------------------------------------------- */
00040 
00041 #include "arm_math.h"
00042 
00043 /**    
00044  * @ingroup groupMatrix    
00045  */
00046 
00047 /**    
00048  * @addtogroup MatrixAdd    
00049  * @{    
00050  */
00051 
00052 /**    
00053  * @brief Q15 matrix addition.    
00054  * @param[in]       *pSrcA points to the first input matrix structure    
00055  * @param[in]       *pSrcB points to the second input matrix structure    
00056  * @param[out]      *pDst points to output matrix structure    
00057  * @return          The function returns either    
00058  * <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.    
00059  *    
00060  * <b>Scaling and Overflow Behavior:</b>    
00061  * \par    
00062  * The function uses saturating arithmetic.    
00063  * Results outside of the allowable Q15 range [0x8000 0x7FFF] will be saturated.    
00064  */
00065 
00066 arm_status arm_mat_add_q15(
00067   const arm_matrix_instance_q15 * pSrcA,
00068   const arm_matrix_instance_q15 * pSrcB,
00069   arm_matrix_instance_q15 * pDst)
00070 {
00071   q15_t *pInA = pSrcA->pData;                    /* input data matrix pointer A  */
00072   q15_t *pInB = pSrcB->pData;                    /* input data matrix pointer B */
00073   q15_t *pOut = pDst->pData;                     /* output data matrix pointer */
00074   uint16_t numSamples;                           /* total number of elements in the matrix  */
00075   uint32_t blkCnt;                               /* loop counters  */
00076   arm_status status;                             /* status of matrix addition  */
00077 
00078 #ifdef ARM_MATH_MATRIX_CHECK
00079 
00080 
00081   /* Check for matrix mismatch condition */
00082   if((pSrcA->numRows != pSrcB->numRows) ||
00083      (pSrcA->numCols != pSrcB->numCols) ||
00084      (pSrcA->numRows != pDst->numRows) || (pSrcA->numCols != pDst->numCols))
00085   {
00086     /* Set status as ARM_MATH_SIZE_MISMATCH */
00087     status = ARM_MATH_SIZE_MISMATCH;
00088   }
00089   else
00090 #endif /*    #ifdef ARM_MATH_MATRIX_CHECK    */
00091 
00092   {
00093     /* Total number of samples in the input matrix */
00094     numSamples = (uint16_t) (pSrcA->numRows * pSrcA->numCols);
00095 
00096 #ifndef ARM_MATH_CM0_FAMILY
00097 
00098     /* Run the below code for Cortex-M4 and Cortex-M3 */
00099 
00100     /* Loop unrolling */
00101     blkCnt = (uint32_t) numSamples >> 2u;
00102 
00103     /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.    
00104      ** a second loop below computes the remaining 1 to 3 samples. */
00105     while(blkCnt > 0u)
00106     {
00107       /* C(m,n) = A(m,n) + B(m,n) */
00108       /* Add, Saturate and then store the results in the destination buffer. */
00109       *__SIMD32(pOut)++ = __QADD16(*__SIMD32(pInA)++, *__SIMD32(pInB)++);
00110       *__SIMD32(pOut)++ = __QADD16(*__SIMD32(pInA)++, *__SIMD32(pInB)++);
00111 
00112       /* Decrement the loop counter */
00113       blkCnt--;
00114     }
00115 
00116     /* If the blockSize is not a multiple of 4, compute any remaining output samples here.    
00117      ** No loop unrolling is used. */
00118     blkCnt = (uint32_t) numSamples % 0x4u;
00119 
00120     /* q15 pointers of input and output are initialized */
00121 
00122     while(blkCnt > 0u)
00123     {
00124       /* C(m,n) = A(m,n) + B(m,n) */
00125       /* Add, Saturate and then store the results in the destination buffer. */
00126       *pOut++ = (q15_t) __QADD16(*pInA++, *pInB++);
00127 
00128       /* Decrement the loop counter */
00129       blkCnt--;
00130     }
00131 
00132 #else
00133 
00134     /* Run the below code for Cortex-M0 */
00135 
00136     /* Initialize blkCnt with number of samples */
00137     blkCnt = (uint32_t) numSamples;
00138 
00139 
00140     /* q15 pointers of input and output are initialized */
00141     while(blkCnt > 0u)
00142     {
00143       /* C(m,n) = A(m,n) + B(m,n) */
00144       /* Add, Saturate and then store the results in the destination buffer. */
00145       *pOut++ = (q15_t) __SSAT(((q31_t) * pInA++ + *pInB++), 16);
00146 
00147       /* Decrement the loop counter */
00148       blkCnt--;
00149     }
00150 
00151 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
00152 
00153     /* set status as ARM_MATH_SUCCESS */
00154     status = ARM_MATH_SUCCESS;
00155   }
00156 
00157   /* Return to application */
00158   return (status);
00159 }
00160 
00161 /**    
00162  * @} end of MatrixAdd group    
00163  */