arm_mat_sub_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_sub_q31.c    
00009 *    
00010 * Description:  Q31 matrix subtraction    
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 MatrixSub        
00049  * @{        
00050  */
00051 
00052 /**        
00053  * @brief Q31 matrix subtraction.        
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 
00067 arm_status arm_mat_sub_q31(
00068   const arm_matrix_instance_q31 * pSrcA,
00069   const arm_matrix_instance_q31 * pSrcB,
00070   arm_matrix_instance_q31 * pDst)
00071 {
00072   q31_t *pIn1 = pSrcA->pData;                    /* input data matrix pointer A */
00073   q31_t *pIn2 = pSrcB->pData;                    /* input data matrix pointer B */
00074   q31_t *pOut = pDst->pData;                     /* output data matrix pointer */
00075   q31_t inA1, inB1;                              /* temporary variables */
00076 
00077 #ifndef ARM_MATH_CM0_FAMILY
00078 
00079   q31_t inA2, inB2;                              /* temporary variables */
00080   q31_t out1, out2;                              /* temporary variables */
00081 
00082 #endif //      #ifndef ARM_MATH_CM0_FAMILY
00083 
00084   uint32_t numSamples;                           /* total number of elements in the matrix  */
00085   uint32_t blkCnt;                               /* loop counters */
00086   arm_status status;                             /* status of matrix subtraction */
00087 
00088 
00089 #ifdef ARM_MATH_MATRIX_CHECK
00090   /* Check for matrix mismatch condition  */
00091   if((pSrcA->numRows != pSrcB->numRows) ||
00092      (pSrcA->numCols != pSrcB->numCols) ||
00093      (pSrcA->numRows != pDst->numRows) || (pSrcA->numCols != pDst->numCols))
00094   {
00095     /* Set status as ARM_MATH_SIZE_MISMATCH */
00096     status = ARM_MATH_SIZE_MISMATCH;
00097   }
00098   else
00099 #endif
00100   {
00101     /* Total number of samples in the input matrix */
00102     numSamples = (uint32_t) pSrcA->numRows * pSrcA->numCols;
00103 
00104 #ifndef ARM_MATH_CM0_FAMILY
00105 
00106     /* Run the below code for Cortex-M4 and Cortex-M3 */
00107 
00108     /* Loop Unrolling */
00109     blkCnt = numSamples >> 2u;
00110 
00111     /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.    
00112      ** a second loop below computes the remaining 1 to 3 samples. */
00113     while(blkCnt > 0u)
00114     {
00115       /* C(m,n) = A(m,n) - B(m,n) */
00116       /* Subtract, saturate and then store the results in the destination buffer. */
00117       /* Read values from source A */
00118       inA1 = pIn1[0];
00119 
00120       /* Read values from source B */
00121       inB1 = pIn2[0];
00122 
00123       /* Read values from source A */
00124       inA2 = pIn1[1];
00125 
00126       /* Subtract and saturate */
00127       out1 = __QSUB(inA1, inB1);
00128 
00129       /* Read values from source B */
00130       inB2 = pIn2[1];
00131 
00132       /* Read values from source A */
00133       inA1 = pIn1[2];
00134 
00135       /* Subtract and saturate */
00136       out2 = __QSUB(inA2, inB2);
00137 
00138       /* Read values from source B */
00139       inB1 = pIn2[2];
00140 
00141       /* Store result in destination */
00142       pOut[0] = out1;
00143       pOut[1] = out2;
00144 
00145       /* Read values from source A */
00146       inA2 = pIn1[3];
00147 
00148       /* Read values from source B */
00149       inB2 = pIn2[3];
00150 
00151       /* Subtract and saturate */
00152       out1 = __QSUB(inA1, inB1);
00153 
00154       /* Subtract and saturate */
00155       out2 = __QSUB(inA2, inB2);
00156 
00157       /* Store result in destination */
00158       pOut[2] = out1;
00159       pOut[3] = out2;
00160 
00161       /* update pointers to process next samples */
00162       pIn1 += 4u;
00163       pIn2 += 4u;
00164       pOut += 4u;
00165 
00166       /* Decrement the loop counter */
00167       blkCnt--;
00168     }
00169 
00170     /* If the numSamples is not a multiple of 4, compute any remaining output samples here.        
00171      ** No loop unrolling is used. */
00172     blkCnt = numSamples % 0x4u;
00173 
00174 #else
00175 
00176     /* Run the below code for Cortex-M0 */
00177 
00178     /* Initialize blkCnt with number of samples */
00179     blkCnt = numSamples;
00180 
00181 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
00182 
00183     while(blkCnt > 0u)
00184     {
00185       /* C(m,n) = A(m,n) - B(m,n) */
00186       /* Subtract, saturate and then store the results in the destination buffer. */
00187       inA1 = *pIn1++;
00188       inB1 = *pIn2++;
00189 
00190       inA1 = __QSUB(inA1, inB1);
00191 
00192       *pOut++ = inA1;
00193 
00194       /* Decrement the loop counter */
00195       blkCnt--;
00196     }
00197 
00198     /* Set status as ARM_MATH_SUCCESS */
00199     status = ARM_MATH_SUCCESS;
00200   }
00201 
00202   /* Return to application */
00203   return (status);
00204 }
00205 
00206 /**        
00207  * @} end of MatrixSub group        
00208  */