arm_add_f32.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_add_f32.c 00009 * 00010 * Description: Floating-point vector addition. 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 00056 void arm_add_f32( 00057 float32_t * pSrcA, 00058 float32_t * pSrcB, 00059 float32_t * pDst, 00060 uint32_t blockSize) 00061 { 00062 uint32_t blkCnt; /* loop counter */ 00063 float32_t inA1, inA2, inA3, inA4; /* temporary input variabels */ 00064 float32_t inB1, inB2, inB3, inB4; /* temporary input variables */ 00065 00066 /*loop Unrolling */ 00067 blkCnt = blockSize >> 3u; 00068 00069 /* First part of the processing with loop unrolling. Compute 8 outputs at a time. 00070 ** a second loop below computes the remaining 1 to 7 samples. */ 00071 while(blkCnt > 0u) 00072 { 00073 /* C = A + B */ 00074 /* Add and then store the results in the destination buffer. */ 00075 00076 /* read four inputs from sourceA and four inputs from sourceB */ 00077 inA1 = *pSrcA; 00078 inB1 = *pSrcB; 00079 inA2 = *(pSrcA + 1); 00080 inB2 = *(pSrcB + 1); 00081 inA3 = *(pSrcA + 2); 00082 inB3 = *(pSrcB + 2); 00083 inA4 = *(pSrcA + 3); 00084 inB4 = *(pSrcB + 3); 00085 00086 /* C = A + B */ 00087 /* add and store result to destination */ 00088 *pDst = inA1 + inB1; 00089 *(pDst + 1) = inA2 + inB2; 00090 *(pDst + 2) = inA3 + inB3; 00091 *(pDst + 3) = inA4 + inB4; 00092 00093 /* read four inputs from sourceA and four inputs from sourceB */ 00094 inA1 = *(pSrcA + 4); 00095 inB1 = *(pSrcB + 4); 00096 inA2 = *(pSrcA + 5); 00097 inB2 = *(pSrcB + 5); 00098 inA3 = *(pSrcA + 6); 00099 inB3 = *(pSrcB + 6); 00100 inA4 = *(pSrcA + 7); 00101 inB4 = *(pSrcB + 7); 00102 00103 /* C = A + B */ 00104 /* add and store result to destination */ 00105 *(pDst + 4) = inA1 + inB1; 00106 *(pDst + 5) = inA2 + inB2; 00107 *(pDst + 6) = inA3 + inB3; 00108 *(pDst + 7) = inA4 + inB4; 00109 00110 /* Update pointers to process next sampels */ 00111 pSrcA += 8u; 00112 pSrcB += 8u; 00113 pDst += 8u; 00114 00115 /* Decrement the loop counter */ 00116 blkCnt--; 00117 } 00118 00119 /* If the blockSize is not a multiple of 8, compute any remaining output samples here. 00120 ** No loop unrolling is used. */ 00121 blkCnt = blockSize % 0x8u; 00122 00123 while(blkCnt > 0u) 00124 { 00125 /* C = A + B */ 00126 /* Add and then store the results in the destination buffer. */ 00127 inA1 = *pSrcA++; 00128 inB1 = *pSrcB++; 00129 00130 *pDst++ = inA1 + inB1; 00131 00132 /* Decrement the loop counter */ 00133 blkCnt--; 00134 } 00135 } 00136
