00001 /* ---------------------------------------------------------------------- 00002 * Copyright (C) 2010 ARM Limited. All rights reserved. 00003 * 00004 * $Date: 15. July 2011 00005 * $Revision: V1.0.10 00006 * 00007 * Project: CMSIS DSP Library 00008 * Title: arm_add_q31.c 00009 * 00010 * Description: Q31 vector addition. 00011 * 00012 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0 00013 * 00014 * Version 1.0.10 2011/7/15 00015 * Big Endian support added and Merged M0 and M3/M4 Source code. 00016 * 00017 * Version 1.0.3 2010/11/29 00018 * Re-organized the CMSIS folders and updated documentation. 00019 * 00020 * Version 1.0.2 2010/11/11 00021 * Documentation updated. 00022 * 00023 * Version 1.0.1 2010/10/05 00024 * Production release and review comments incorporated. 00025 * 00026 * Version 1.0.0 2010/09/20 00027 * Production release and review comments incorporated. 00028 * 00029 * Version 0.0.7 2010/06/10 00030 * Misra-C changes done 00031 * -------------------------------------------------------------------- */ 00032 00033 #include "arm_math.h" 00034 00059 void arm_add_q31( 00060 q31_t * pSrcA, 00061 q31_t * pSrcB, 00062 q31_t * pDst, 00063 uint32_t blockSize) 00064 { 00065 uint32_t blkCnt; /* loop counter */ 00066 00067 #ifndef ARM_MATH_CM0 00068 00069 /* Run the below code for Cortex-M4 and Cortex-M3 */ 00070 00071 00072 /*loop Unrolling */ 00073 blkCnt = blockSize >> 2u; 00074 00075 /* First part of the processing with loop unrolling. Compute 4 outputs at a time. 00076 ** a second loop below computes the remaining 1 to 3 samples. */ 00077 while(blkCnt > 0u) 00078 { 00079 /* C = A + B */ 00080 /* Add and then store the results in the destination buffer. */ 00081 *pDst++ = __QADD(*pSrcA++, *pSrcB++); 00082 *pDst++ = __QADD(*pSrcA++, *pSrcB++); 00083 *pDst++ = __QADD(*pSrcA++, *pSrcB++); 00084 *pDst++ = __QADD(*pSrcA++, *pSrcB++); 00085 00086 /* Decrement the loop counter */ 00087 blkCnt--; 00088 } 00089 00090 /* If the blockSize is not a multiple of 4, compute any remaining output samples here. 00091 ** No loop unrolling is used. */ 00092 blkCnt = blockSize % 0x4u; 00093 00094 while(blkCnt > 0u) 00095 { 00096 /* C = A + B */ 00097 /* Add and then store the results in the destination buffer. */ 00098 *pDst++ = __QADD(*pSrcA++, *pSrcB++); 00099 00100 /* Decrement the loop counter */ 00101 blkCnt--; 00102 } 00103 00104 #else 00105 00106 /* Run the below code for Cortex-M0 */ 00107 00108 00109 00110 /* Initialize blkCnt with number of samples */ 00111 blkCnt = blockSize; 00112 00113 while(blkCnt > 0u) 00114 { 00115 /* C = A + B */ 00116 /* Add and then store the results in the destination buffer. */ 00117 *pDst++ = (q31_t) clip_q63_to_q31((q63_t) * pSrcA++ + *pSrcB++); 00118 00119 /* Decrement the loop counter */ 00120 blkCnt--; 00121 } 00122 00123 #endif /* #ifndef ARM_MATH_CM0 */ 00124 00125 } 00126