arm_mat_add_q31.c

00001 /* ----------------------------------------------------------------------    
00002 * Copyright (C) 2010-2014 ARM Limited. All rights reserved.    
00003 *    
00004 * $Date:        19. March 2015
00005 * $Revision:    V.1.4.5
00006 *    
00007 * Project:      CMSIS DSP Library    
00008 * Title:        arm_mat_add_q31.c    
00009 *    
00010 * Description:  Q31 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 Q31 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 Q31 range [0x80000000 0x7FFFFFFF] will be saturated.      
00064  */
00065 
00066 arm_status arm_mat_add_q31(
00067   const arm_matrix_instance_q31 * pSrcA,
00068   const arm_matrix_instance_q31 * pSrcB,
00069   arm_matrix_instance_q31 * pDst)
00070 {
00071   q31_t *pIn1 = pSrcA->pData;                    /* input data matrix pointer A */
00072   q31_t *pIn2 = pSrcB->pData;                    /* input data matrix pointer B */
00073   q31_t *pOut = pDst->pData;                     /* output data matrix pointer */
00074   q31_t inA1, inB1;                              /* temporary variables */
00075 
00076 #ifndef ARM_MATH_CM0_FAMILY
00077 
00078   q31_t inA2, inB2;                              /* temporary variables */
00079   q31_t out1, out2;                              /* temporary variables */
00080 
00081 #endif //      #ifndef ARM_MATH_CM0_FAMILY
00082 
00083   uint32_t numSamples;                           /* total number of elements in the matrix  */
00084   uint32_t blkCnt;                               /* loop counters */
00085   arm_status status;                             /* status of matrix addition */
00086 
00087 #ifdef ARM_MATH_MATRIX_CHECK
00088   /* Check for matrix mismatch condition */
00089   if((pSrcA->numRows != pSrcB->numRows) ||
00090      (pSrcA->numCols != pSrcB->numCols) ||
00091      (pSrcA->numRows != pDst->numRows) || (pSrcA->numCols != pDst->numCols))
00092   {
00093     /* Set status as ARM_MATH_SIZE_MISMATCH */
00094     status = ARM_MATH_SIZE_MISMATCH;
00095   }
00096   else
00097 #endif
00098   {
00099     /* Total number of samples in the input matrix */
00100     numSamples = (uint32_t) pSrcA->numRows * pSrcA->numCols;
00101 
00102 #ifndef ARM_MATH_CM0_FAMILY
00103 
00104     /* Run the below code for Cortex-M4 and Cortex-M3 */
00105 
00106     /* Loop Unrolling */
00107     blkCnt = numSamples >> 2u;
00108 
00109 
00110     /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.    
00111      ** a second loop below computes the remaining 1 to 3 samples. */
00112     while(blkCnt > 0u)
00113     {
00114       /* C(m,n) = A(m,n) + B(m,n) */
00115       /* Add, saturate and then store the results in the destination buffer. */
00116       /* Read values from source A */
00117       inA1 = pIn1[0];
00118 
00119       /* Read values from source B */
00120       inB1 = pIn2[0];
00121 
00122       /* Read values from source A */
00123       inA2 = pIn1[1];
00124 
00125       /* Add and saturate */
00126       out1 = __QADD(inA1, inB1);
00127 
00128       /* Read values from source B */
00129       inB2 = pIn2[1];
00130 
00131       /* Read values from source A */
00132       inA1 = pIn1[2];
00133 
00134       /* Add and saturate */
00135       out2 = __QADD(inA2, inB2);
00136 
00137       /* Read values from source B */
00138       inB1 = pIn2[2];
00139 
00140       /* Store result in destination */
00141       pOut[0] = out1;
00142       pOut[1] = out2;
00143 
00144       /* Read values from source A */
00145       inA2 = pIn1[3];
00146 
00147       /* Read values from source B */
00148       inB2 = pIn2[3];
00149 
00150       /* Add and saturate */
00151       out1 = __QADD(inA1, inB1);
00152       out2 = __QADD(inA2, inB2);
00153 
00154       /* Store result in destination */
00155       pOut[2] = out1;
00156       pOut[3] = out2;
00157 
00158       /* update pointers to process next sampels */
00159       pIn1 += 4u;
00160       pIn2 += 4u;
00161       pOut += 4u;
00162 
00163       /* Decrement the loop counter */
00164       blkCnt--;
00165     }
00166 
00167     /* If the numSamples is not a multiple of 4, compute any remaining output samples here.      
00168      ** No loop unrolling is used. */
00169     blkCnt = numSamples % 0x4u;
00170 
00171 #else
00172 
00173     /* Run the below code for Cortex-M0 */
00174 
00175     /* Initialize blkCnt with number of samples */
00176     blkCnt = numSamples;
00177 
00178 
00179 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
00180 
00181     while(blkCnt > 0u)
00182     {
00183       /* C(m,n) = A(m,n) + B(m,n) */
00184       /* Add, saturate and then store the results in the destination buffer. */
00185       inA1 = *pIn1++;
00186       inB1 = *pIn2++;
00187 
00188       inA1 = __QADD(inA1, inB1);
00189 
00190       /* Decrement the loop counter */
00191       blkCnt--;
00192 
00193       *pOut++ = inA1;
00194 
00195     }
00196 
00197     /* set status as ARM_MATH_SUCCESS */
00198     status = ARM_MATH_SUCCESS;
00199   }
00200 
00201   /* Return to application */
00202   return (status);
00203 }
00204 
00205 /**      
00206  * @} end of MatrixAdd group      
00207  */