Rev 56 | Details | Compare with Previous | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
56 | mjames | 1 | /* ---------------------------------------------------------------------- |
2 | * Project: CMSIS DSP Library |
||
3 | * Title: arm_mat_trans_q15.c |
||
4 | * Description: Q15 matrix transpose |
||
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 groupMatrix |
||
33 | */ |
||
34 | |||
35 | /** |
||
36 | * @addtogroup MatrixTrans |
||
37 | * @{ |
||
38 | */ |
||
39 | |||
40 | /* |
||
41 | * @brief Q15 matrix transpose. |
||
42 | * @param[in] *pSrc points to the input matrix |
||
43 | * @param[out] *pDst points to the output matrix |
||
44 | * @return The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> |
||
45 | * or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking. |
||
46 | */ |
||
47 | |||
48 | arm_status arm_mat_trans_q15( |
||
49 | const arm_matrix_instance_q15 * pSrc, |
||
50 | arm_matrix_instance_q15 * pDst) |
||
51 | { |
||
52 | q15_t *pSrcA = pSrc->pData; /* input data matrix pointer */ |
||
53 | q15_t *pOut = pDst->pData; /* output data matrix pointer */ |
||
54 | uint16_t nRows = pSrc->numRows; /* number of nRows */ |
||
55 | uint16_t nColumns = pSrc->numCols; /* number of nColumns */ |
||
56 | uint16_t col, row = nRows, i = 0U; /* row and column loop counters */ |
||
57 | arm_status status; /* status of matrix transpose */ |
||
58 | |||
59 | #if defined (ARM_MATH_DSP) |
||
60 | |||
61 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
||
62 | #ifndef UNALIGNED_SUPPORT_DISABLE |
||
63 | |||
64 | q31_t in; /* variable to hold temporary output */ |
||
65 | |||
66 | #else |
||
67 | |||
68 | q15_t in; |
||
69 | |||
70 | #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */ |
||
71 | |||
72 | #ifdef ARM_MATH_MATRIX_CHECK |
||
73 | |||
74 | |||
75 | /* Check for matrix mismatch condition */ |
||
76 | if ((pSrc->numRows != pDst->numCols) || (pSrc->numCols != pDst->numRows)) |
||
77 | { |
||
78 | /* Set status as ARM_MATH_SIZE_MISMATCH */ |
||
79 | status = ARM_MATH_SIZE_MISMATCH; |
||
80 | } |
||
81 | else |
||
82 | #endif /* #ifdef ARM_MATH_MATRIX_CHECK */ |
||
83 | |||
84 | { |
||
85 | /* Matrix transpose by exchanging the rows with columns */ |
||
86 | /* row loop */ |
||
87 | do |
||
88 | { |
||
89 | |||
90 | /* Apply loop unrolling and exchange the columns with row elements */ |
||
91 | col = nColumns >> 2U; |
||
92 | |||
93 | /* The pointer pOut is set to starting address of the column being processed */ |
||
94 | pOut = pDst->pData + i; |
||
95 | |||
96 | /* First part of the processing with loop unrolling. Compute 4 outputs at a time. |
||
97 | ** a second loop below computes the remaining 1 to 3 samples. */ |
||
98 | while (col > 0U) |
||
99 | { |
||
100 | #ifndef UNALIGNED_SUPPORT_DISABLE |
||
101 | |||
102 | /* Read two elements from the row */ |
||
103 | in = *__SIMD32(pSrcA)++; |
||
104 | |||
105 | /* Unpack and store one element in the destination */ |
||
106 | #ifndef ARM_MATH_BIG_ENDIAN |
||
107 | |||
108 | *pOut = (q15_t) in; |
||
109 | |||
110 | #else |
||
111 | |||
112 | *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16); |
||
113 | |||
114 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
||
115 | |||
116 | /* Update the pointer pOut to point to the next row of the transposed matrix */ |
||
117 | pOut += nRows; |
||
118 | |||
119 | /* Unpack and store the second element in the destination */ |
||
120 | |||
121 | #ifndef ARM_MATH_BIG_ENDIAN |
||
122 | |||
123 | *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16); |
||
124 | |||
125 | #else |
||
126 | |||
127 | *pOut = (q15_t) in; |
||
128 | |||
129 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
||
130 | |||
131 | /* Update the pointer pOut to point to the next row of the transposed matrix */ |
||
132 | pOut += nRows; |
||
133 | |||
134 | /* Read two elements from the row */ |
||
135 | #ifndef ARM_MATH_BIG_ENDIAN |
||
136 | |||
137 | in = *__SIMD32(pSrcA)++; |
||
138 | |||
139 | #else |
||
140 | |||
141 | in = *__SIMD32(pSrcA)++; |
||
142 | |||
143 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
||
144 | |||
145 | /* Unpack and store one element in the destination */ |
||
146 | #ifndef ARM_MATH_BIG_ENDIAN |
||
147 | |||
148 | *pOut = (q15_t) in; |
||
149 | |||
150 | #else |
||
151 | |||
152 | *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16); |
||
153 | |||
154 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
||
155 | |||
156 | /* Update the pointer pOut to point to the next row of the transposed matrix */ |
||
157 | pOut += nRows; |
||
158 | |||
159 | /* Unpack and store the second element in the destination */ |
||
160 | #ifndef ARM_MATH_BIG_ENDIAN |
||
161 | |||
162 | *pOut = (q15_t) ((in & (q31_t) 0xffff0000) >> 16); |
||
163 | |||
164 | #else |
||
165 | |||
166 | *pOut = (q15_t) in; |
||
167 | |||
168 | #endif /* #ifndef ARM_MATH_BIG_ENDIAN */ |
||
169 | |||
170 | #else |
||
171 | /* Read one element from the row */ |
||
172 | in = *pSrcA++; |
||
173 | |||
174 | /* Store one element in the destination */ |
||
175 | *pOut = in; |
||
176 | |||
177 | /* Update the pointer px to point to the next row of the transposed matrix */ |
||
178 | pOut += nRows; |
||
179 | |||
180 | /* Read one element from the row */ |
||
181 | in = *pSrcA++; |
||
182 | |||
183 | /* Store one element in the destination */ |
||
184 | *pOut = in; |
||
185 | |||
186 | /* Update the pointer px to point to the next row of the transposed matrix */ |
||
187 | pOut += nRows; |
||
188 | |||
189 | /* Read one element from the row */ |
||
190 | in = *pSrcA++; |
||
191 | |||
192 | /* Store one element in the destination */ |
||
193 | *pOut = in; |
||
194 | |||
195 | /* Update the pointer px to point to the next row of the transposed matrix */ |
||
196 | pOut += nRows; |
||
197 | |||
198 | /* Read one element from the row */ |
||
199 | in = *pSrcA++; |
||
200 | |||
201 | /* Store one element in the destination */ |
||
202 | *pOut = in; |
||
203 | |||
204 | #endif /* #ifndef UNALIGNED_SUPPORT_DISABLE */ |
||
205 | |||
206 | /* Update the pointer pOut to point to the next row of the transposed matrix */ |
||
207 | pOut += nRows; |
||
208 | |||
209 | /* Decrement the column loop counter */ |
||
210 | col--; |
||
211 | } |
||
212 | |||
213 | /* Perform matrix transpose for last 3 samples here. */ |
||
214 | col = nColumns % 0x4U; |
||
215 | |||
216 | #else |
||
217 | |||
218 | /* Run the below code for Cortex-M0 */ |
||
219 | |||
220 | #ifdef ARM_MATH_MATRIX_CHECK |
||
221 | |||
222 | /* Check for matrix mismatch condition */ |
||
223 | if ((pSrc->numRows != pDst->numCols) || (pSrc->numCols != pDst->numRows)) |
||
224 | { |
||
225 | /* Set status as ARM_MATH_SIZE_MISMATCH */ |
||
226 | status = ARM_MATH_SIZE_MISMATCH; |
||
227 | } |
||
228 | else |
||
229 | #endif /* #ifdef ARM_MATH_MATRIX_CHECK */ |
||
230 | |||
231 | { |
||
232 | /* Matrix transpose by exchanging the rows with columns */ |
||
233 | /* row loop */ |
||
234 | do |
||
235 | { |
||
236 | /* The pointer pOut is set to starting address of the column being processed */ |
||
237 | pOut = pDst->pData + i; |
||
238 | |||
239 | /* Initialize column loop counter */ |
||
240 | col = nColumns; |
||
241 | |||
242 | #endif /* #if defined (ARM_MATH_DSP) */ |
||
243 | |||
244 | while (col > 0U) |
||
245 | { |
||
246 | /* Read and store the input element in the destination */ |
||
247 | *pOut = *pSrcA++; |
||
248 | |||
249 | /* Update the pointer pOut to point to the next row of the transposed matrix */ |
||
250 | pOut += nRows; |
||
251 | |||
252 | /* Decrement the column loop counter */ |
||
253 | col--; |
||
254 | } |
||
255 | |||
256 | i++; |
||
257 | |||
258 | /* Decrement the row loop counter */ |
||
259 | row--; |
||
260 | |||
261 | } while (row > 0U); |
||
262 | |||
263 | /* set status as ARM_MATH_SUCCESS */ |
||
264 | status = ARM_MATH_SUCCESS; |
||
265 | } |
||
266 | /* Return to application */ |
||
267 | return (status); |
||
268 | } |
||
269 | |||
270 | /** |
||
271 | * @} end of MatrixTrans group |
||
272 | */ |