arm_cmplx_mag_squared_q15.c

00001 /* ----------------------------------------------------------------------
00002  * Project:      CMSIS DSP Library
00003  * Title:        arm_cmplx_mag_squared_q15.c
00004  * Description:  Q15 complex magnitude squared
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 cmplx_mag_squared
00037  * @{
00038  */
00039 
00040 /**
00041  * @brief  Q15 complex magnitude squared
00042  * @param  *pSrc points to the complex input vector
00043  * @param  *pDst points to the real output vector
00044  * @param  numSamples number of complex samples in the input vector
00045  * @return none.
00046  *
00047  * <b>Scaling and Overflow Behavior:</b>
00048  * \par
00049  * The function implements 1.15 by 1.15 multiplications and finally output is converted into 3.13 format.
00050  */
00051 
00052 void arm_cmplx_mag_squared_q15(
00053   q15_t * pSrc,
00054   q15_t * pDst,
00055   uint32_t numSamples)
00056 {
00057   q31_t acc0, acc1;                              /* Accumulators */
00058 
00059 #if defined (ARM_MATH_DSP)
00060 
00061   /* Run the below code for Cortex-M4 and Cortex-M3 */
00062   uint32_t blkCnt;                               /* loop counter */
00063   q31_t in1, in2, in3, in4;
00064   q31_t acc2, acc3;
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[0] = (A[0] * A[0] + A[1] * A[1]) */
00074     in1 = *__SIMD32(pSrc)++;
00075     in2 = *__SIMD32(pSrc)++;
00076     in3 = *__SIMD32(pSrc)++;
00077     in4 = *__SIMD32(pSrc)++;
00078 
00079     acc0 = __SMUAD(in1, in1);
00080     acc1 = __SMUAD(in2, in2);
00081     acc2 = __SMUAD(in3, in3);
00082     acc3 = __SMUAD(in4, in4);
00083 
00084     /* store the result in 3.13 format in the destination buffer. */
00085     *pDst++ = (q15_t) (acc0 >> 17);
00086     *pDst++ = (q15_t) (acc1 >> 17);
00087     *pDst++ = (q15_t) (acc2 >> 17);
00088     *pDst++ = (q15_t) (acc3 >> 17);
00089 
00090     /* Decrement the loop counter */
00091     blkCnt--;
00092   }
00093 
00094   /* If the numSamples is not a multiple of 4, compute any remaining output samples here.
00095    ** No loop unrolling is used. */
00096   blkCnt = numSamples % 0x4U;
00097 
00098   while (blkCnt > 0U)
00099   {
00100     /* C[0] = (A[0] * A[0] + A[1] * A[1]) */
00101     in1 = *__SIMD32(pSrc)++;
00102     acc0 = __SMUAD(in1, in1);
00103 
00104     /* store the result in 3.13 format in the destination buffer. */
00105     *pDst++ = (q15_t) (acc0 >> 17);
00106 
00107     /* Decrement the loop counter */
00108     blkCnt--;
00109   }
00110 
00111 #else
00112 
00113   /* Run the below code for Cortex-M0 */
00114   q15_t real, imag;                              /* Temporary variables to store real and imaginary values */
00115 
00116   while (numSamples > 0U)
00117   {
00118     /* out = ((real * real) + (imag * imag)) */
00119     real = *pSrc++;
00120     imag = *pSrc++;
00121     acc0 = (real * real);
00122     acc1 = (imag * imag);
00123     /* store the result in 3.13 format in the destination buffer. */
00124     *pDst++ = (q15_t) (((q63_t) acc0 + acc1) >> 17);
00125 
00126     /* Decrement the loop counter */
00127     numSamples--;
00128   }
00129 
00130 #endif /* #if defined (ARM_MATH_DSP) */
00131 
00132 }
00133 
00134 /**
00135  * @} end of cmplx_mag_squared group
00136  */
00137