arm_cmplx_mag_squared_q31.c
Go to the documentation of this file.
00001 /* ---------------------------------------------------------------------- 00002 * Copyright (C) 2011 ARM Limited. All rights reserved. 00003 * 00004 * $Date: 15. December 2011 00005 * $Revision: V2.0.0 00006 * 00007 * Project: Cortex-R DSP Library 00008 * Title: arm_cmplx_mag_squared_q31.c 00009 * 00010 * Description: Q31 complex magnitude squared. 00011 * 00012 * Target Processor: Cortex-R4/R5 00013 * 00014 * Version 1.0.0 2011/03/08 00015 * Alpha release. 00016 * 00017 * Version 1.0.1 2011/09/30 00018 * Beta release. 00019 * 00020 * Version 2.0.0 2011/12/15 00021 * Final release. 00022 * 00023 * ---------------------------------------------------------------------------- */ 00024 #include "arm_math.h" 00025 00049 void arm_cmplx_mag_squared_q31( 00050 q31_t * pSrc, 00051 q31_t * pDst, 00052 uint32_t numSamples) 00053 { 00054 uint32_t blkCnt; /* loop counter */ 00055 q31_t real1, real2, imag1, imag2; /* temporary variables to hold input data */ 00056 q31_t real3, real4, imag3, imag4; /* Temporary variables to hold input data */ 00057 q31_t out1, out2; /* temporary variabels to hold output data */ 00058 00059 /* loop Unrolling */ 00060 blkCnt = numSamples >> 2u; 00061 00062 /* First part of the processing with loop unrolling. Compute 4 outputs at a time. 00063 ** a second loop below computes the remaining 1 to 3 samples. */ 00064 while(blkCnt > 0u) 00065 { 00066 /* read complex data from soruce buffer */ 00067 real1 = pSrc[0]; 00068 imag1 = pSrc[1]; 00069 real2 = pSrc[2]; 00070 imag2 = pSrc[3]; 00071 00072 /* calculate power in 2.30 format */ 00073 real1 = (q31_t)(((q63_t) real1 * real1) >> 32); 00074 imag1 = (q31_t)(((q63_t) imag1 * imag1) >> 32); 00075 real2 = (q31_t)(((q63_t) real2 * real2) >> 32); 00076 imag2 = (q31_t)(((q63_t) imag2 * imag2) >> 32); 00077 00078 /* read complex data from soruce buffer */ 00079 real3 = pSrc[4]; 00080 imag3 = pSrc[5]; 00081 real4 = pSrc[6]; 00082 imag4 = pSrc[7]; 00083 00084 /* calculate power in 2.30 format */ 00085 real3 = (q31_t)(((q63_t) real3 * real3) >> 32); 00086 imag3 = (q31_t)(((q63_t) imag3 * imag3) >> 32); 00087 real4 = (q31_t)(((q63_t) real4 * real4) >> 32); 00088 imag4 = (q31_t)(((q63_t) imag4 * imag4) >> 32); 00089 00090 /* get the result in 3.29 format */ 00091 real1 = (real1 >> 1); 00092 imag1 = (imag1 >> 1); 00093 real2 = (real2 >> 1); 00094 imag2 = (imag2 >> 1); 00095 00096 /* accumulate real and imaginary powers */ 00097 out1 = real1 + imag1; 00098 out2 = real2 + imag2; 00099 00100 00101 /* get the result in 3.29 format */ 00102 real3 = (real3 >> 1); 00103 imag3 = (imag3 >> 1); 00104 real4 = (real4 >> 1); 00105 00106 /* store the result to destination buffer */ 00107 pDst[0] = out1; 00108 00109 /* get the result in 3.29 format */ 00110 imag4 = (imag4 >> 1); 00111 00112 /* store the result to destination buffer */ 00113 pDst[1] = out2; 00114 00115 /* accumulate real and imaginary powers */ 00116 out1 = real3 + imag3; 00117 out2 = real4 + imag4; 00118 00119 /* store the result to destination buffer */ 00120 pDst[2] = out1; 00121 00122 /* increment destination buffer by 8 to process next samples */ 00123 pSrc += 8u; 00124 00125 /* store the result to destination buffer */ 00126 pDst[3] = out2; 00127 00128 /* increment destination buffer by 4 to process next samples */ 00129 pDst += 4u; 00130 00131 /* Decrement the loop counter */ 00132 blkCnt--; 00133 } 00134 00135 /* If the numSamples is not a multiple of 4, compute any remaining output samples here. 00136 ** No loop unrolling is used. */ 00137 blkCnt = numSamples % 0x4u; 00138 00139 while(blkCnt > 0u) 00140 { 00141 /* C[0] = (A[0] * A[0] + A[1] * A[1]) */ 00142 real1 = *pSrc++; 00143 imag1 = *pSrc++; 00144 real1 = (q31_t) (((q63_t) real1 * real1) >> 33); 00145 imag1 = (q31_t) (((q63_t) imag1 * imag1) >> 33); 00146 /* store the result in 3.29 format in the destination buffer. */ 00147 *pDst++ = real1 + imag1; 00148 00149 /* Decrement the loop counter */ 00150 blkCnt--; 00151 } 00152 } 00153
