arm_cmplx_dot_prod_q31.c

00001 /* ----------------------------------------------------------------------    
00002 * Copyright (C) 2010-2013 ARM Limited. All rights reserved.    
00003 *    
00004 * $Date:        17. January 2013
00005 * $Revision:    V1.4.1
00006 *    
00007 * Project:      CMSIS DSP Library    
00008 * Title:        arm_cmplx_dot_prod_q31.c    
00009 *    
00010 * Description:  Q31 complex dot product    
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 groupCmplxMath    
00045  */
00046 
00047 /**    
00048  * @addtogroup cmplx_dot_prod    
00049  * @{    
00050  */
00051 
00052 /**    
00053  * @brief  Q31 complex dot product    
00054  * @param  *pSrcA points to the first input vector    
00055  * @param  *pSrcB points to the second input vector    
00056  * @param  numSamples number of complex samples in each vector    
00057  * @param  *realResult real part of the result returned here    
00058  * @param  *imagResult imaginary part of the result returned here    
00059  * @return none.    
00060  *    
00061  * <b>Scaling and Overflow Behavior:</b>    
00062  * \par    
00063  * The function is implemented using an internal 64-bit accumulator.    
00064  * The intermediate 1.31 by 1.31 multiplications are performed with 64-bit precision and then shifted to 16.48 format.    
00065  * The internal real and imaginary accumulators are in 16.48 format and provide 15 guard bits.    
00066  * Additions are nonsaturating and no overflow will occur as long as <code>numSamples</code> is less than 32768.    
00067  * The return results <code>realResult</code> and <code>imagResult</code> are in 16.48 format.    
00068  * Input down scaling is not required.    
00069  */
00070 
00071 void arm_cmplx_dot_prod_q31(
00072   q31_t * pSrcA,
00073   q31_t * pSrcB,
00074   uint32_t numSamples,
00075   q63_t * realResult,
00076   q63_t * imagResult)
00077 {
00078   q63_t real_sum = 0, imag_sum = 0;              /* Temporary result storage */
00079 
00080 #ifndef ARM_MATH_CM0_FAMILY
00081 
00082   /* Run the below code for Cortex-M4 and Cortex-M3 */
00083   uint32_t blkCnt;                               /* loop counter */
00084 
00085 
00086   /*loop Unrolling */
00087   blkCnt = numSamples >> 2u;
00088 
00089   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.    
00090    ** a second loop below computes the remaining 1 to 3 samples. */
00091   while(blkCnt > 0u)
00092   {
00093     /* CReal = A[0]* B[0] + A[2]* B[2] + A[4]* B[4] + .....+ A[numSamples-2]* B[numSamples-2] */
00094     /* Convert real data in 2.62 to 16.48 by 14 right shifts */
00095     real_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00096     /* CImag = A[1]* B[1] + A[3]* B[3] + A[5]* B[5] + .....+ A[numSamples-1]* B[numSamples-1] */
00097     /* Convert imag data in 2.62 to 16.48 by 14 right shifts */
00098     imag_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00099 
00100     real_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00101     imag_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00102 
00103     real_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00104     imag_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00105 
00106     real_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00107     imag_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00108 
00109 
00110     /* Decrement the loop counter */
00111     blkCnt--;
00112   }
00113 
00114   /* If the numSamples  is not a multiple of 4, compute any remaining output samples here.    
00115    ** No loop unrolling is used. */
00116   blkCnt = numSamples % 0x4u;
00117 
00118   while(blkCnt > 0u)
00119   {
00120     /* CReal = A[0]* B[0] + A[2]* B[2] + A[4]* B[4] + .....+ A[numSamples-2]* B[numSamples-2] */
00121     real_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00122     /* CImag = A[1]* B[1] + A[3]* B[3] + A[5]* B[5] + .....+ A[numSamples-1]* B[numSamples-1] */
00123     imag_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00124 
00125     /* Decrement the loop counter */
00126     blkCnt--;
00127   }
00128 
00129 #else
00130 
00131   /* Run the below code for Cortex-M0 */
00132 
00133   while(numSamples > 0u)
00134   {
00135     /* outReal = realA[0]* realB[0] + realA[2]* realB[2] + realA[4]* realB[4] + .....+ realA[numSamples-2]* realB[numSamples-2] */
00136     real_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00137     /* outImag = imagA[1]* imagB[1] + imagA[3]* imagB[3] + imagA[5]* imagB[5] + .....+ imagA[numSamples-1]* imagB[numSamples-1] */
00138     imag_sum += (q63_t) * pSrcA++ * (*pSrcB++) >> 14;
00139 
00140     /* Decrement the loop counter */
00141     numSamples--;
00142   }
00143 
00144 #endif /* #ifndef ARM_MATH_CM0_FAMILY */
00145 
00146   /* Store the real and imaginary results in 16.48 format  */
00147   *realResult = real_sum;
00148   *imagResult = imag_sum;
00149 }
00150 
00151 /**    
00152  * @} end of cmplx_dot_prod group    
00153  */