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_sub_q15.c 00009 * 00010 * Description: Q15 vector subtraction. 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 00058 void arm_sub_q15( 00059 q15_t * pSrcA, 00060 q15_t * pSrcB, 00061 q15_t * pDst, 00062 uint32_t blockSize) 00063 { 00064 uint32_t blkCnt; /* loop counter */ 00065 00066 00067 #ifndef ARM_MATH_CM0 00068 00069 /* Run the below code for Cortex-M4 and Cortex-M3 */ 00070 /*loop Unrolling */ 00071 blkCnt = blockSize >> 2u; 00072 00073 /* First part of the processing with loop unrolling. Compute 4 outputs at a time. 00074 ** a second loop below computes the remaining 1 to 3 samples. */ 00075 while(blkCnt > 0u) 00076 { 00077 /* C = A - B */ 00078 /* Subtract and then store the results in the destination buffer two samples at a time. */ 00079 *__SIMD32(pDst)++ = __QSUB16(*__SIMD32(pSrcA)++, *__SIMD32(pSrcB)++); 00080 *__SIMD32(pDst)++ = __QSUB16(*__SIMD32(pSrcA)++, *__SIMD32(pSrcB)++); 00081 00082 /* Decrement the loop counter */ 00083 blkCnt--; 00084 } 00085 00086 /* If the blockSize is not a multiple of 4, compute any remaining output samples here. 00087 ** No loop unrolling is used. */ 00088 blkCnt = blockSize % 0x4u; 00089 00090 while(blkCnt > 0u) 00091 { 00092 /* C = A - B */ 00093 /* Subtract and then store the result in the destination buffer. */ 00094 *pDst++ = (q15_t) __QSUB16(*pSrcA++, *pSrcB++); 00095 00096 /* Decrement the loop counter */ 00097 blkCnt--; 00098 } 00099 00100 #else 00101 00102 /* Run the below code for Cortex-M0 */ 00103 00104 /* Initialize blkCnt with number of samples */ 00105 blkCnt = blockSize; 00106 00107 while(blkCnt > 0u) 00108 { 00109 /* C = A - B */ 00110 /* Subtract and then store the result in the destination buffer. */ 00111 *pDst++ = (q15_t) __SSAT(((q31_t) * pSrcA++ - *pSrcB++), 16); 00112 00113 /* Decrement the loop counter */ 00114 blkCnt--; 00115 } 00116 00117 #endif /* #ifndef ARM_MATH_CM0 */ 00118 00119 00120 } 00121