arm_cmplx_mult_cmplx_q15.c

00001 /* ----------------------------------------------------------------------
00002  * Project:      CMSIS DSP Library
00003  * Title:        arm_cmplx_mult_cmplx_q15.c
00004  * Description:  Q15 complex-by-complex multiplication
00005  *
00006  * $Date:        27. January 2017
00007  * $Revision:    V.1.5.1
00008  *
00009  * Target Processor: Cortex-M cores
00010  * -------------------------------------------------------------------- */
00011 /*
00012  * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved.
00013  *
00014  * SPDX-License-Identifier: Apache-2.0
00015  *
00016  * Licensed under the Apache License, Version 2.0 (the License); you may
00017  * not use this file except in compliance with the License.
00018  * You may obtain a copy of the License at
00019  *
00020  * www.apache.org/licenses/LICENSE-2.0
00021  *
00022  * Unless required by applicable law or agreed to in writing, software
00023  * distributed under the License is distributed on an AS IS BASIS, WITHOUT
00024  * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
00025  * See the License for the specific language governing permissions and
00026  * limitations under the License.
00027  */
00028 
00029 #include "arm_math.h"
00030 
00031 /**
00032  * @ingroup groupCmplxMath
00033  */
00034 
00035 /**
00036  * @addtogroup CmplxByCmplxMult
00037  * @{
00038  */
00039 
00040 /**
00041  * @brief  Q15 complex-by-complex multiplication
00042  * @param[in]  *pSrcA points to the first input vector
00043  * @param[in]  *pSrcB points to the second input vector
00044  * @param[out]  *pDst  points to the output vector
00045  * @param[in]  numSamples number of complex samples in each vector
00046  * @return none.
00047  *
00048  * <b>Scaling and Overflow Behavior:</b>
00049  * \par
00050  * The function implements 1.15 by 1.15 multiplications and finally output is converted into 3.13 format.
00051  */
00052 
00053 void arm_cmplx_mult_cmplx_q15(
00054   q15_t * pSrcA,
00055   q15_t * pSrcB,
00056   q15_t * pDst,
00057   uint32_t numSamples)
00058 {
00059   q15_t a, b, c, d;                              /* Temporary variables to store real and imaginary values */
00060 
00061 #if defined (ARM_MATH_DSP)
00062 
00063   /* Run the below code for Cortex-M4 and Cortex-M3 */
00064   uint32_t blkCnt;                               /* loop counters */
00065 
00066   /* loop Unrolling */
00067   blkCnt = numSamples >> 2U;
00068 
00069   /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.
00070    ** a second loop below computes the remaining 1 to 3 samples. */
00071   while (blkCnt > 0U)
00072   {
00073     /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1].  */
00074     /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i].  */
00075     a = *pSrcA++;
00076     b = *pSrcA++;
00077     c = *pSrcB++;
00078     d = *pSrcB++;
00079 
00080     /* store the result in 3.13 format in the destination buffer. */
00081     *pDst++ =
00082       (q15_t) (q31_t) (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17);
00083     /* store the result in 3.13 format in the destination buffer. */
00084     *pDst++ =
00085       (q15_t) (q31_t) (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17);
00086 
00087     a = *pSrcA++;
00088     b = *pSrcA++;
00089     c = *pSrcB++;
00090     d = *pSrcB++;
00091 
00092     /* store the result in 3.13 format in the destination buffer. */
00093     *pDst++ =
00094       (q15_t) (q31_t) (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17);
00095     /* store the result in 3.13 format in the destination buffer. */
00096     *pDst++ =
00097       (q15_t) (q31_t) (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17);
00098 
00099     a = *pSrcA++;
00100     b = *pSrcA++;
00101     c = *pSrcB++;
00102     d = *pSrcB++;
00103 
00104     /* store the result in 3.13 format in the destination buffer. */
00105     *pDst++ =
00106       (q15_t) (q31_t) (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17);
00107     /* store the result in 3.13 format in the destination buffer. */
00108     *pDst++ =
00109       (q15_t) (q31_t) (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17);
00110 
00111     a = *pSrcA++;
00112     b = *pSrcA++;
00113     c = *pSrcB++;
00114     d = *pSrcB++;
00115 
00116     /* store the result in 3.13 format in the destination buffer. */
00117     *pDst++ =
00118       (q15_t) (q31_t) (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17);
00119     /* store the result in 3.13 format in the destination buffer. */
00120     *pDst++ =
00121       (q15_t) (q31_t) (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17);
00122 
00123     /* Decrement the blockSize loop counter */
00124     blkCnt--;
00125   }
00126 
00127   /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
00128    ** No loop unrolling is used. */
00129   blkCnt = numSamples % 0x4U;
00130 
00131   while (blkCnt > 0U)
00132   {
00133     /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1].  */
00134     /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i].  */
00135     a = *pSrcA++;
00136     b = *pSrcA++;
00137     c = *pSrcB++;
00138     d = *pSrcB++;
00139 
00140     /* store the result in 3.13 format in the destination buffer. */
00141     *pDst++ =
00142       (q15_t) (q31_t) (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17);
00143     /* store the result in 3.13 format in the destination buffer. */
00144     *pDst++ =
00145       (q15_t) (q31_t) (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17);
00146 
00147     /* Decrement the blockSize loop counter */
00148     blkCnt--;
00149   }
00150 
00151 #else
00152 
00153   /* Run the below code for Cortex-M0 */
00154 
00155   while (numSamples > 0U)
00156   {
00157     /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1].  */
00158     /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i].  */
00159     a = *pSrcA++;
00160     b = *pSrcA++;
00161     c = *pSrcB++;
00162     d = *pSrcB++;
00163 
00164     /* store the result in 3.13 format in the destination buffer. */
00165     *pDst++ =
00166       (q15_t) (q31_t) (((q31_t) a * c) >> 17) - (((q31_t) b * d) >> 17);
00167     /* store the result in 3.13 format in the destination buffer. */
00168     *pDst++ =
00169       (q15_t) (q31_t) (((q31_t) a * d) >> 17) + (((q31_t) b * c) >> 17);
00170 
00171     /* Decrement the blockSize loop counter */
00172     numSamples--;
00173   }
00174 
00175 #endif /* #if defined (ARM_MATH_DSP) */
00176 
00177 }
00178 
00179 /**
00180  * @} end of CmplxByCmplxMult group
00181  */
00182