arm_cmplx_mult_cmplx_q15.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_mult_cmplx_q15.c 00009 * 00010 * Description: Q15 complex-by-complex multiplication 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 00052 void arm_cmplx_mult_cmplx_q15( 00053 q15_t * pSrcA, 00054 q15_t * pSrcB, 00055 q15_t * pDst, 00056 uint32_t numSamples) 00057 { 00058 q15_t a1, b1, c1, d1; /* Temporary variables to store real and imaginary values */ 00059 q15_t a2, b2, c2, d2; /* Temporary variables to store real and imaginary values */ 00060 q15_t out1, out2, out3, out4; 00061 00062 uint32_t blkCnt; /* loop counters */ 00063 00064 /* loop Unrolling */ 00065 blkCnt = numSamples >> 2u; 00066 00067 /* First part of the processing with loop unrolling. Compute 4 outputs at a time. 00068 ** a second loop below computes the remaining 1 to 3 samples. */ 00069 while(blkCnt > 0u) 00070 { 00071 /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1]. */ 00072 /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i]. */ 00073 a1 = *pSrcA; 00074 c1 = *pSrcB; 00075 b1 = *(pSrcA + 1); 00076 d1 = *(pSrcB + 1); 00077 00078 /* store the result in 3.13 format in the destination buffer. */ 00079 out1 = 00080 (q15_t) (q31_t) (((q31_t) a1 * c1) >> 17) - (((q31_t) b1 * d1) >> 17); 00081 /* store the result in 3.13 format in the destination buffer. */ 00082 out2 = 00083 (q15_t) (q31_t) (((q31_t) a1 * d1) >> 17) + (((q31_t) b1 * c1) >> 17); 00084 00085 a2 = *(pSrcA + 2); 00086 c2 = *(pSrcB + 2); 00087 b2 = *(pSrcA + 3); 00088 d2 = *(pSrcB + 3); 00089 00090 /* store the result in 3.13 format in the destination buffer. */ 00091 out3 = 00092 (q15_t) (q31_t) (((q31_t) a2 * c2) >> 17) - (((q31_t) b2 * d2) >> 17); 00093 /* store the result in 3.13 format in the destination buffer. */ 00094 out4 = 00095 (q15_t) (q31_t) (((q31_t) a2 * d2) >> 17) + (((q31_t) b2 * c2) >> 17); 00096 00097 a1 = *(pSrcA + 4); 00098 c1 = *(pSrcB + 4); 00099 b1 = *(pSrcA + 5); 00100 d1 = *(pSrcB + 5); 00101 00102 *pDst = out1; 00103 *(pDst + 1) = out2; 00104 *(pDst + 2) = out3; 00105 *(pDst + 3) = out4; 00106 00107 /* store the result in 3.13 format in the destination buffer. */ 00108 out1 = 00109 (q15_t) (q31_t) (((q31_t) a1 * c1) >> 17) - (((q31_t) b1 * d1) >> 17); 00110 /* store the result in 3.13 format in the destination buffer. */ 00111 out2 = 00112 (q15_t) (q31_t) (((q31_t) a1 * d1) >> 17) + (((q31_t) b1 * c1) >> 17); 00113 00114 a2 = *(pSrcA + 6); 00115 c2 = *(pSrcB + 6); 00116 b2 = *(pSrcA + 7); 00117 d2 = *(pSrcB + 7); 00118 00119 /* store the result in 3.13 format in the destination buffer. */ 00120 out3 = 00121 (q15_t) (q31_t) (((q31_t) a2 * c2) >> 17) - (((q31_t) b2 * d2) >> 17); 00122 /* store the result in 3.13 format in the destination buffer. */ 00123 out4 = 00124 (q15_t) (q31_t) (((q31_t) a2 * d2) >> 17) + (((q31_t) b2 * c2) >> 17); 00125 00126 00127 *(pDst + 4) = out1; 00128 *(pDst + 5) = out2; 00129 pSrcA += 8u; 00130 *(pDst + 6) = out3; 00131 *(pDst + 7) = out4; 00132 pSrcB += 8u; 00133 pDst += 8u; 00134 00135 /* Decrement the blockSize loop counter */ 00136 blkCnt--; 00137 } 00138 00139 /* If the blockSize is not a multiple of 4, compute any remaining output samples here. 00140 ** No loop unrolling is used. */ 00141 blkCnt = numSamples % 0x4u; 00142 00143 while(blkCnt > 0u) 00144 { 00145 /* C[2 * i] = A[2 * i] * B[2 * i] - A[2 * i + 1] * B[2 * i + 1]. */ 00146 /* C[2 * i + 1] = A[2 * i] * B[2 * i + 1] + A[2 * i + 1] * B[2 * i]. */ 00147 a1 = *pSrcA++; 00148 b1 = *pSrcA++; 00149 c1 = *pSrcB++; 00150 d1 = *pSrcB++; 00151 00152 /* store the result in 3.13 format in the destination buffer. */ 00153 *pDst++ = 00154 (q15_t) (q31_t) (((q31_t) a1 * c1) >> 17) - (((q31_t) b1 * d1) >> 17); 00155 /* store the result in 3.13 format in the destination buffer. */ 00156 *pDst++ = 00157 (q15_t) (q31_t) (((q31_t) a1 * d1) >> 17) + (((q31_t) b1 * c1) >> 17); 00158 00159 /* Decrement the blockSize loop counter */ 00160 blkCnt--; 00161 } 00162 } 00163
