Rev 2 | Details | Compare with Previous | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 2 | mjames | 1 | /* ---------------------------------------------------------------------- |
| 2 | * Project: CMSIS DSP Library |
||
| 3 | * Title: arm_shift_q7.c |
||
| 4 | * Description: Processing function for the Q7 Shifting |
||
| 5 | * |
||
| 6 | * $Date: 27. January 2017 |
||
| 7 | * $Revision: V.1.5.1 |
||
| 8 | * |
||
| 9 | * Target Processor: Cortex-M cores |
||
| 10 | * -------------------------------------------------------------------- */ |
||
| 11 | /* |
||
| 12 | * Copyright (C) 2010-2017 ARM Limited or its affiliates. All rights reserved. |
||
| 13 | * |
||
| 14 | * SPDX-License-Identifier: Apache-2.0 |
||
| 15 | * |
||
| 16 | * Licensed under the Apache License, Version 2.0 (the License); you may |
||
| 17 | * not use this file except in compliance with the License. |
||
| 18 | * You may obtain a copy of the License at |
||
| 19 | * |
||
| 20 | * www.apache.org/licenses/LICENSE-2.0 |
||
| 21 | * |
||
| 22 | * Unless required by applicable law or agreed to in writing, software |
||
| 23 | * distributed under the License is distributed on an AS IS BASIS, WITHOUT |
||
| 24 | * WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
||
| 25 | * See the License for the specific language governing permissions and |
||
| 26 | * limitations under the License. |
||
| 27 | */ |
||
| 28 | |||
| 29 | #include "arm_math.h" |
||
| 30 | |||
| 31 | /** |
||
| 32 | * @ingroup groupMath |
||
| 33 | */ |
||
| 34 | |||
| 35 | /** |
||
| 36 | * @addtogroup shift |
||
| 37 | * @{ |
||
| 38 | */ |
||
| 39 | |||
| 40 | |||
| 41 | /** |
||
| 42 | * @brief Shifts the elements of a Q7 vector a specified number of bits. |
||
| 43 | * @param[in] *pSrc points to the input vector |
||
| 44 | * @param[in] shiftBits number of bits to shift. A positive value shifts left; a negative value shifts right. |
||
| 45 | * @param[out] *pDst points to the output vector |
||
| 46 | * @param[in] blockSize number of samples in the vector |
||
| 47 | * @return none. |
||
| 48 | * |
||
| 49 | * \par Conditions for optimum performance |
||
| 50 | * Input and output buffers should be aligned by 32-bit |
||
| 51 | * |
||
| 52 | * |
||
| 53 | * <b>Scaling and Overflow Behavior:</b> |
||
| 54 | * \par |
||
| 55 | * The function uses saturating arithmetic. |
||
| 56 | * Results outside of the allowable Q7 range [0x8 0x7F] will be saturated. |
||
| 57 | */ |
||
| 58 | |||
| 59 | void arm_shift_q7( |
||
| 60 | q7_t * pSrc, |
||
| 61 | int8_t shiftBits, |
||
| 62 | q7_t * pDst, |
||
| 63 | uint32_t blockSize) |
||
| 64 | { |
||
| 65 | uint32_t blkCnt; /* loop counter */ |
||
| 66 | uint8_t sign; /* Sign of shiftBits */ |
||
| 67 | |||
| 68 | #if defined (ARM_MATH_DSP) |
||
| 69 | |||
| 70 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
||
| 71 | q7_t in1; /* Input value1 */ |
||
| 72 | q7_t in2; /* Input value2 */ |
||
| 73 | q7_t in3; /* Input value3 */ |
||
| 74 | q7_t in4; /* Input value4 */ |
||
| 75 | |||
| 76 | |||
| 77 | /*loop Unrolling */ |
||
| 78 | blkCnt = blockSize >> 2U; |
||
| 79 | |||
| 80 | /* Getting the sign of shiftBits */ |
||
| 81 | sign = (shiftBits & 0x80); |
||
| 82 | |||
| 83 | /* If the shift value is positive then do right shift else left shift */ |
||
| 84 | if (sign == 0U) |
||
| 85 | { |
||
| 86 | /* First part of the processing with loop unrolling. Compute 4 outputs at a time. |
||
| 87 | ** a second loop below computes the remaining 1 to 3 samples. */ |
||
| 88 | while (blkCnt > 0U) |
||
| 89 | { |
||
| 90 | /* C = A << shiftBits */ |
||
| 91 | /* Read 4 inputs */ |
||
| 92 | in1 = *pSrc; |
||
| 93 | in2 = *(pSrc + 1); |
||
| 94 | in3 = *(pSrc + 2); |
||
| 95 | in4 = *(pSrc + 3); |
||
| 96 | |||
| 97 | /* Store the Shifted result in the destination buffer in single cycle by packing the outputs */ |
||
| 98 | *__SIMD32(pDst)++ = __PACKq7(__SSAT((in1 << shiftBits), 8), |
||
| 99 | __SSAT((in2 << shiftBits), 8), |
||
| 100 | __SSAT((in3 << shiftBits), 8), |
||
| 101 | __SSAT((in4 << shiftBits), 8)); |
||
| 102 | /* Update source pointer to process next sampels */ |
||
| 103 | pSrc += 4U; |
||
| 104 | |||
| 105 | /* Decrement the loop counter */ |
||
| 106 | blkCnt--; |
||
| 107 | } |
||
| 108 | |||
| 109 | /* If the blockSize is not a multiple of 4, compute any remaining output samples here. |
||
| 110 | ** No loop unrolling is used. */ |
||
| 111 | blkCnt = blockSize % 0x4U; |
||
| 112 | |||
| 113 | while (blkCnt > 0U) |
||
| 114 | { |
||
| 115 | /* C = A << shiftBits */ |
||
| 116 | /* Shift the input and then store the result in the destination buffer. */ |
||
| 117 | *pDst++ = (q7_t) __SSAT((*pSrc++ << shiftBits), 8); |
||
| 118 | |||
| 119 | /* Decrement the loop counter */ |
||
| 120 | blkCnt--; |
||
| 121 | } |
||
| 122 | } |
||
| 123 | else |
||
| 124 | { |
||
| 125 | shiftBits = -shiftBits; |
||
| 126 | /* First part of the processing with loop unrolling. Compute 4 outputs at a time. |
||
| 127 | ** a second loop below computes the remaining 1 to 3 samples. */ |
||
| 128 | while (blkCnt > 0U) |
||
| 129 | { |
||
| 130 | /* C = A >> shiftBits */ |
||
| 131 | /* Read 4 inputs */ |
||
| 132 | in1 = *pSrc; |
||
| 133 | in2 = *(pSrc + 1); |
||
| 134 | in3 = *(pSrc + 2); |
||
| 135 | in4 = *(pSrc + 3); |
||
| 136 | |||
| 137 | /* Store the Shifted result in the destination buffer in single cycle by packing the outputs */ |
||
| 138 | *__SIMD32(pDst)++ = __PACKq7((in1 >> shiftBits), (in2 >> shiftBits), |
||
| 139 | (in3 >> shiftBits), (in4 >> shiftBits)); |
||
| 140 | |||
| 141 | |||
| 142 | pSrc += 4U; |
||
| 143 | |||
| 144 | /* Decrement the loop counter */ |
||
| 145 | blkCnt--; |
||
| 146 | } |
||
| 147 | |||
| 148 | /* If the blockSize is not a multiple of 4, compute any remaining output samples here. |
||
| 149 | ** No loop unrolling is used. */ |
||
| 150 | blkCnt = blockSize % 0x4U; |
||
| 151 | |||
| 152 | while (blkCnt > 0U) |
||
| 153 | { |
||
| 154 | /* C = A >> shiftBits */ |
||
| 155 | /* Shift the input and then store the result in the destination buffer. */ |
||
| 156 | in1 = *pSrc++; |
||
| 157 | *pDst++ = (in1 >> shiftBits); |
||
| 158 | |||
| 159 | /* Decrement the loop counter */ |
||
| 160 | blkCnt--; |
||
| 161 | } |
||
| 162 | } |
||
| 163 | |||
| 164 | #else |
||
| 165 | |||
| 166 | /* Run the below code for Cortex-M0 */ |
||
| 167 | |||
| 168 | /* Getting the sign of shiftBits */ |
||
| 169 | sign = (shiftBits & 0x80); |
||
| 170 | |||
| 171 | /* If the shift value is positive then do right shift else left shift */ |
||
| 172 | if (sign == 0U) |
||
| 173 | { |
||
| 174 | /* Initialize blkCnt with number of samples */ |
||
| 175 | blkCnt = blockSize; |
||
| 176 | |||
| 177 | while (blkCnt > 0U) |
||
| 178 | { |
||
| 179 | /* C = A << shiftBits */ |
||
| 180 | /* Shift the input and then store the result in the destination buffer. */ |
||
| 181 | *pDst++ = (q7_t) __SSAT(((q15_t) * pSrc++ << shiftBits), 8); |
||
| 182 | |||
| 183 | /* Decrement the loop counter */ |
||
| 184 | blkCnt--; |
||
| 185 | } |
||
| 186 | } |
||
| 187 | else |
||
| 188 | { |
||
| 189 | /* Initialize blkCnt with number of samples */ |
||
| 190 | blkCnt = blockSize; |
||
| 191 | |||
| 192 | while (blkCnt > 0U) |
||
| 193 | { |
||
| 194 | /* C = A >> shiftBits */ |
||
| 195 | /* Shift the input and then store the result in the destination buffer. */ |
||
| 196 | *pDst++ = (*pSrc++ >> -shiftBits); |
||
| 197 | |||
| 198 | /* Decrement the loop counter */ |
||
| 199 | blkCnt--; |
||
| 200 | } |
||
| 201 | } |
||
| 202 | |||
| 203 | #endif /* #if defined (ARM_MATH_DSP) */ |
||
| 204 | } |
||
| 205 | |||
| 206 | /** |
||
| 207 | * @} end of shift group |
||
| 208 | */ |