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_shift_q7.c 00009 * 00010 * Description: Processing function for the Q7 Shifting 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_shift_q7( 00060 q7_t * pSrc, 00061 int8_t shiftBits, 00062 q7_t * pDst, 00063 uint32_t blockSize) 00064 { 00065 uint32_t blkCnt; /* loop counter */ 00066 uint8_t sign; /* Sign of shiftBits */ 00067 00068 #ifndef ARM_MATH_CM0 00069 00070 /* Run the below code for Cortex-M4 and Cortex-M3 */ 00071 q7_t in1; /* Input value1 */ 00072 q7_t in2; /* Input value2 */ 00073 q7_t in3; /* Input value3 */ 00074 q7_t in4; /* Input value4 */ 00075 00076 00077 /*loop Unrolling */ 00078 blkCnt = blockSize >> 2u; 00079 00080 /* Getting the sign of shiftBits */ 00081 sign = (shiftBits & 0x80); 00082 00083 /* If the shift value is positive then do right shift else left shift */ 00084 if(sign == 0u) 00085 { 00086 /* First part of the processing with loop unrolling. Compute 4 outputs at a time. 00087 ** a second loop below computes the remaining 1 to 3 samples. */ 00088 while(blkCnt > 0u) 00089 { 00090 /* C = A << shiftBits */ 00091 /* Read 4 inputs */ 00092 in1 = *pSrc++; 00093 in2 = *pSrc++; 00094 in3 = *pSrc++; 00095 in4 = *pSrc++; 00096 00097 /* Store the Shifted result in the destination buffer in single cycle by packing the outputs */ 00098 *__SIMD32(pDst)++ = __PACKq7(__SSAT((in1 << shiftBits), 8), 00099 __SSAT((in2 << shiftBits), 8), 00100 __SSAT((in3 << shiftBits), 8), 00101 __SSAT((in4 << shiftBits), 8)); 00102 00103 /* Decrement the loop counter */ 00104 blkCnt--; 00105 } 00106 00107 /* If the blockSize is not a multiple of 4, compute any remaining output samples here. 00108 ** No loop unrolling is used. */ 00109 blkCnt = blockSize % 0x4u; 00110 00111 while(blkCnt > 0u) 00112 { 00113 /* C = A << shiftBits */ 00114 /* Shift the input and then store the result in the destination buffer. */ 00115 *pDst++ = (q7_t) __SSAT((*pSrc++ << shiftBits), 8); 00116 00117 /* Decrement the loop counter */ 00118 blkCnt--; 00119 } 00120 } 00121 else 00122 { 00123 /* First part of the processing with loop unrolling. Compute 4 outputs at a time. 00124 ** a second loop below computes the remaining 1 to 3 samples. */ 00125 while(blkCnt > 0u) 00126 { 00127 /* C = A >> shiftBits */ 00128 /* Read 4 inputs */ 00129 in1 = *pSrc++; 00130 in2 = *pSrc++; 00131 in3 = *pSrc++; 00132 in4 = *pSrc++; 00133 00134 /* Store the Shifted result in the destination buffer in single cycle by packing the outputs */ 00135 *__SIMD32(pDst)++ = __PACKq7((in1 >> -shiftBits), (in2 >> -shiftBits), 00136 (in3 >> -shiftBits), (in4 >> -shiftBits)); 00137 00138 /* Decrement the loop counter */ 00139 blkCnt--; 00140 } 00141 00142 /* If the blockSize is not a multiple of 4, compute any remaining output samples here. 00143 ** No loop unrolling is used. */ 00144 blkCnt = blockSize % 0x4u; 00145 00146 while(blkCnt > 0u) 00147 { 00148 /* C = A >> shiftBits */ 00149 /* Shift the input and then store the result in the destination buffer. */ 00150 *pDst++ = (*pSrc++ >> -shiftBits); 00151 00152 /* Decrement the loop counter */ 00153 blkCnt--; 00154 } 00155 } 00156 00157 #else 00158 00159 /* Run the below code for Cortex-M0 */ 00160 00161 /* Getting the sign of shiftBits */ 00162 sign = (shiftBits & 0x80); 00163 00164 /* If the shift value is positive then do right shift else left shift */ 00165 if(sign == 0u) 00166 { 00167 /* Initialize blkCnt with number of samples */ 00168 blkCnt = blockSize; 00169 00170 while(blkCnt > 0u) 00171 { 00172 /* C = A << shiftBits */ 00173 /* Shift the input and then store the result in the destination buffer. */ 00174 *pDst++ = (q7_t) __SSAT(((q15_t) * pSrc++ << shiftBits), 8); 00175 00176 /* Decrement the loop counter */ 00177 blkCnt--; 00178 } 00179 } 00180 else 00181 { 00182 /* Initialize blkCnt with number of samples */ 00183 blkCnt = blockSize; 00184 00185 while(blkCnt > 0u) 00186 { 00187 /* C = A >> shiftBits */ 00188 /* Shift the input and then store the result in the destination buffer. */ 00189 *pDst++ = (*pSrc++ >> -shiftBits); 00190 00191 /* Decrement the loop counter */ 00192 blkCnt--; 00193 } 00194 } 00195 00196 #endif /* #ifndef ARM_MATH_CM0 */ 00197 00198 } 00199