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_mat_sub_q15.c 00009 * 00010 * Description: Q15 Matrix 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 0.0.5 2010/04/26 00030 * incorporated review comments and updated with latest CMSIS layer 00031 * 00032 * Version 0.0.3 2010/03/10 00033 * Initial version 00034 * -------------------------------------------------------------------- */ 00035 00036 #include "arm_math.h" 00037 00061 arm_status arm_mat_sub_q15( 00062 const arm_matrix_instance_q15 * pSrcA, 00063 const arm_matrix_instance_q15 * pSrcB, 00064 arm_matrix_instance_q15 * pDst) 00065 { 00066 q15_t *pInA = pSrcA->pData; /* input data matrix pointer A */ 00067 q15_t *pInB = pSrcB->pData; /* input data matrix pointer B */ 00068 q15_t *pOut = pDst->pData; /* output data matrix pointer */ 00069 uint32_t numSamples; /* total number of elements in the matrix */ 00070 uint32_t blkCnt; /* loop counters */ 00071 arm_status status; /* status of matrix subtraction */ 00072 00073 00074 #ifdef ARM_MATH_MATRIX_CHECK 00075 00076 00077 /* Check for matrix mismatch condition */ 00078 if((pSrcA->numRows != pSrcB->numRows) || 00079 (pSrcA->numCols != pSrcB->numCols) || 00080 (pSrcA->numRows != pDst->numRows) || (pSrcA->numCols != pDst->numCols)) 00081 { 00082 /* Set status as ARM_MATH_SIZE_MISMATCH */ 00083 status = ARM_MATH_SIZE_MISMATCH; 00084 } 00085 else 00086 #endif /* #ifdef ARM_MATH_MATRIX_CHECK */ 00087 00088 { 00089 /* Total number of samples in the input matrix */ 00090 numSamples = (uint32_t) pSrcA->numRows * pSrcA->numCols; 00091 00092 #ifndef ARM_MATH_CM0 00093 00094 /* Run the below code for Cortex-M4 and Cortex-M3 */ 00095 00096 /* Apply loop unrolling */ 00097 blkCnt = numSamples >> 2u; 00098 00099 /* First part of the processing with loop unrolling. Compute 4 outputs at a time. 00100 ** a second loop below computes the remaining 1 to 3 samples. */ 00101 while(blkCnt > 0u) 00102 { 00103 /* C(m,n) = A(m,n) - B(m,n) */ 00104 /* Subtract, Saturate and then store the results in the destination buffer. */ 00105 *__SIMD32(pOut)++ = __QSUB16(*__SIMD32(pInA)++, *__SIMD32(pInB)++); 00106 *__SIMD32(pOut)++ = __QSUB16(*__SIMD32(pInA)++, *__SIMD32(pInB)++); 00107 00108 /* Decrement the loop counter */ 00109 blkCnt--; 00110 } 00111 00112 /* If the blockSize is not a multiple of 4, compute any remaining output samples here. 00113 ** No loop unrolling is used. */ 00114 blkCnt = numSamples % 0x4u; 00115 00116 while(blkCnt > 0u) 00117 { 00118 /* C(m,n) = A(m,n) - B(m,n) */ 00119 /* Subtract and then store the results in the destination buffer. */ 00120 *pOut++ = (q15_t) __QSUB16(*pInA++, *pInB++); 00121 00122 /* Decrement the loop counter */ 00123 blkCnt--; 00124 } 00125 00126 #else 00127 00128 /* Run the below code for Cortex-M0 */ 00129 00130 /* Initialize blkCnt with number of samples */ 00131 blkCnt = numSamples; 00132 00133 while(blkCnt > 0u) 00134 { 00135 /* C(m,n) = A(m,n) - B(m,n) */ 00136 /* Subtract and then store the results in the destination buffer. */ 00137 *pOut++ = (q15_t) __SSAT(((q31_t) * pInA++ - *pInB++), 16); 00138 00139 /* Decrement the loop counter */ 00140 blkCnt--; 00141 } 00142 00143 #endif /* #ifndef ARM_MATH_CM0 */ 00144 00145 /* Set status as ARM_MATH_SUCCESS */ 00146 status = ARM_MATH_SUCCESS; 00147 } 00148 00149 /* Return to application */ 00150 return (status); 00151 } 00152