Go to most recent revision | Details | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
56 | mjames | 1 | /* ---------------------------------------------------------------------- |
2 | * Project: CMSIS DSP Library |
||
3 | * Title: arm_shift_q15.c |
||
4 | * Description: Shifts the elements of a Q15 vector by a specified number of bits |
||
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 | * @brief Shifts the elements of a Q15 vector a specified number of bits. |
||
42 | * @param[in] *pSrc points to the input vector |
||
43 | * @param[in] shiftBits number of bits to shift. A positive value shifts left; a negative value shifts right. |
||
44 | * @param[out] *pDst points to the output vector |
||
45 | * @param[in] blockSize number of samples in the vector |
||
46 | * @return none. |
||
47 | * |
||
48 | * <b>Scaling and Overflow Behavior:</b> |
||
49 | * \par |
||
50 | * The function uses saturating arithmetic. |
||
51 | * Results outside of the allowable Q15 range [0x8000 0x7FFF] will be saturated. |
||
52 | */ |
||
53 | |||
54 | void arm_shift_q15( |
||
55 | q15_t * pSrc, |
||
56 | int8_t shiftBits, |
||
57 | q15_t * pDst, |
||
58 | uint32_t blockSize) |
||
59 | { |
||
60 | uint32_t blkCnt; /* loop counter */ |
||
61 | uint8_t sign; /* Sign of shiftBits */ |
||
62 | |||
63 | #if defined (ARM_MATH_DSP) |
||
64 | |||
65 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
||
66 | |||
67 | q15_t in1, in2; /* Temporary variables */ |
||
68 | |||
69 | |||
70 | /*loop Unrolling */ |
||
71 | blkCnt = blockSize >> 2U; |
||
72 | |||
73 | /* Getting the sign of shiftBits */ |
||
74 | sign = (shiftBits & 0x80); |
||
75 | |||
76 | /* If the shift value is positive then do right shift else left shift */ |
||
77 | if (sign == 0U) |
||
78 | { |
||
79 | /* First part of the processing with loop unrolling. Compute 4 outputs at a time. |
||
80 | ** a second loop below computes the remaining 1 to 3 samples. */ |
||
81 | while (blkCnt > 0U) |
||
82 | { |
||
83 | /* Read 2 inputs */ |
||
84 | in1 = *pSrc++; |
||
85 | in2 = *pSrc++; |
||
86 | /* C = A << shiftBits */ |
||
87 | /* Shift the inputs and then store the results in the destination buffer. */ |
||
88 | #ifndef ARM_MATH_BIG_ENDIAN |
||
89 | |||
90 | *__SIMD32(pDst)++ = __PKHBT(__SSAT((in1 << shiftBits), 16), |
||
91 | __SSAT((in2 << shiftBits), 16), 16); |
||
92 | |||
93 | #else |
||
94 | |||
95 | *__SIMD32(pDst)++ = __PKHBT(__SSAT((in2 << shiftBits), 16), |
||
96 | __SSAT((in1 << shiftBits), 16), 16); |
||
97 | |||
98 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
||
99 | |||
100 | in1 = *pSrc++; |
||
101 | in2 = *pSrc++; |
||
102 | |||
103 | #ifndef ARM_MATH_BIG_ENDIAN |
||
104 | |||
105 | *__SIMD32(pDst)++ = __PKHBT(__SSAT((in1 << shiftBits), 16), |
||
106 | __SSAT((in2 << shiftBits), 16), 16); |
||
107 | |||
108 | #else |
||
109 | |||
110 | *__SIMD32(pDst)++ = __PKHBT(__SSAT((in2 << shiftBits), 16), |
||
111 | __SSAT((in1 << shiftBits), 16), 16); |
||
112 | |||
113 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
||
114 | |||
115 | /* Decrement the loop counter */ |
||
116 | blkCnt--; |
||
117 | } |
||
118 | |||
119 | /* If the blockSize is not a multiple of 4, compute any remaining output samples here. |
||
120 | ** No loop unrolling is used. */ |
||
121 | blkCnt = blockSize % 0x4U; |
||
122 | |||
123 | while (blkCnt > 0U) |
||
124 | { |
||
125 | /* C = A << shiftBits */ |
||
126 | /* Shift and then store the results in the destination buffer. */ |
||
127 | *pDst++ = __SSAT((*pSrc++ << shiftBits), 16); |
||
128 | |||
129 | /* Decrement the loop counter */ |
||
130 | blkCnt--; |
||
131 | } |
||
132 | } |
||
133 | else |
||
134 | { |
||
135 | /* First part of the processing with loop unrolling. Compute 4 outputs at a time. |
||
136 | ** a second loop below computes the remaining 1 to 3 samples. */ |
||
137 | while (blkCnt > 0U) |
||
138 | { |
||
139 | /* Read 2 inputs */ |
||
140 | in1 = *pSrc++; |
||
141 | in2 = *pSrc++; |
||
142 | |||
143 | /* C = A >> shiftBits */ |
||
144 | /* Shift the inputs and then store the results in the destination buffer. */ |
||
145 | #ifndef ARM_MATH_BIG_ENDIAN |
||
146 | |||
147 | *__SIMD32(pDst)++ = __PKHBT((in1 >> -shiftBits), |
||
148 | (in2 >> -shiftBits), 16); |
||
149 | |||
150 | #else |
||
151 | |||
152 | *__SIMD32(pDst)++ = __PKHBT((in2 >> -shiftBits), |
||
153 | (in1 >> -shiftBits), 16); |
||
154 | |||
155 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
||
156 | |||
157 | in1 = *pSrc++; |
||
158 | in2 = *pSrc++; |
||
159 | |||
160 | #ifndef ARM_MATH_BIG_ENDIAN |
||
161 | |||
162 | *__SIMD32(pDst)++ = __PKHBT((in1 >> -shiftBits), |
||
163 | (in2 >> -shiftBits), 16); |
||
164 | |||
165 | #else |
||
166 | |||
167 | *__SIMD32(pDst)++ = __PKHBT((in2 >> -shiftBits), |
||
168 | (in1 >> -shiftBits), 16); |
||
169 | |||
170 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
||
171 | |||
172 | /* Decrement the loop counter */ |
||
173 | blkCnt--; |
||
174 | } |
||
175 | |||
176 | /* If the blockSize is not a multiple of 4, compute any remaining output samples here. |
||
177 | ** No loop unrolling is used. */ |
||
178 | blkCnt = blockSize % 0x4U; |
||
179 | |||
180 | while (blkCnt > 0U) |
||
181 | { |
||
182 | /* C = A >> shiftBits */ |
||
183 | /* Shift the inputs and then store the results in the destination buffer. */ |
||
184 | *pDst++ = (*pSrc++ >> -shiftBits); |
||
185 | |||
186 | /* Decrement the loop counter */ |
||
187 | blkCnt--; |
||
188 | } |
||
189 | } |
||
190 | |||
191 | #else |
||
192 | |||
193 | /* Run the below code for Cortex-M0 */ |
||
194 | |||
195 | /* Getting the sign of shiftBits */ |
||
196 | sign = (shiftBits & 0x80); |
||
197 | |||
198 | /* If the shift value is positive then do right shift else left shift */ |
||
199 | if (sign == 0U) |
||
200 | { |
||
201 | /* Initialize blkCnt with number of samples */ |
||
202 | blkCnt = blockSize; |
||
203 | |||
204 | while (blkCnt > 0U) |
||
205 | { |
||
206 | /* C = A << shiftBits */ |
||
207 | /* Shift and then store the results in the destination buffer. */ |
||
208 | *pDst++ = __SSAT(((q31_t) * pSrc++ << shiftBits), 16); |
||
209 | |||
210 | /* Decrement the loop counter */ |
||
211 | blkCnt--; |
||
212 | } |
||
213 | } |
||
214 | else |
||
215 | { |
||
216 | /* Initialize blkCnt with number of samples */ |
||
217 | blkCnt = blockSize; |
||
218 | |||
219 | while (blkCnt > 0U) |
||
220 | { |
||
221 | /* C = A >> shiftBits */ |
||
222 | /* Shift the inputs and then store the results in the destination buffer. */ |
||
223 | *pDst++ = (*pSrc++ >> -shiftBits); |
||
224 | |||
225 | /* Decrement the loop counter */ |
||
226 | blkCnt--; |
||
227 | } |
||
228 | } |
||
229 | |||
230 | #endif /* #if defined (ARM_MATH_DSP) */ |
||
231 | |||
232 | } |
||
233 | |||
234 | /** |
||
235 | * @} end of shift group |
||
236 | */ |