Subversion Repositories DashDisplay

Rev

Details | Last modification | View Log | RSS feed

Rev Author Line No. Line
2 mjames 1
/* ----------------------------------------------------------------------    
2
* Copyright (C) 2010-2014 ARM Limited. All rights reserved.    
3
*    
4
* $Date:        19. March 2015
5
* $Revision:    V.1.4.5  
6
*    
7
* Project:          CMSIS DSP Library    
8
* Title:            arm_dct4_q15.c    
9
*    
10
* Description:  Processing function of DCT4 & IDCT4 Q15.    
11
*    
12
* Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
13
*  
14
* Redistribution and use in source and binary forms, with or without
15
* modification, are permitted provided that the following conditions
16
* are met:
17
*   - Redistributions of source code must retain the above copyright
18
*     notice, this list of conditions and the following disclaimer.
19
*   - Redistributions in binary form must reproduce the above copyright
20
*     notice, this list of conditions and the following disclaimer in
21
*     the documentation and/or other materials provided with the
22
*     distribution.
23
*   - Neither the name of ARM LIMITED nor the names of its contributors
24
*     may be used to endorse or promote products derived from this
25
*     software without specific prior written permission.
26
*
27
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
28
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
29
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
30
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
31
* COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
32
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
33
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
34
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
35
* CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
36
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
37
* ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
38
* POSSIBILITY OF SUCH DAMAGE.  
39
* -------------------------------------------------------------------- */
40
 
41
#include "arm_math.h"
42
 
43
/**    
44
 * @addtogroup DCT4_IDCT4    
45
 * @{    
46
 */
47
 
48
/**    
49
 * @brief Processing function for the Q15 DCT4/IDCT4.  
50
 * @param[in]       *S             points to an instance of the Q15 DCT4 structure.  
51
 * @param[in]       *pState        points to state buffer.  
52
 * @param[in,out]   *pInlineBuffer points to the in-place input and output buffer.  
53
 * @return none.  
54
 *    
55
 * \par Input an output formats:    
56
 * Internally inputs are downscaled in the RFFT process function to avoid overflows.    
57
 * Number of bits downscaled, depends on the size of the transform.    
58
 * The input and output formats for different DCT sizes and number of bits to upscale are mentioned in the table below:    
59
 *    
60
 * \image html dct4FormatsQ15Table.gif    
61
 */
62
 
63
void arm_dct4_q15(
64
  const arm_dct4_instance_q15 * S,
65
  q15_t * pState,
66
  q15_t * pInlineBuffer)
67
{
68
  uint32_t i;                                    /* Loop counter */
69
  q15_t *weights = S->pTwiddle;                  /* Pointer to the Weights table */
70
  q15_t *cosFact = S->pCosFactor;                /* Pointer to the cos factors table */
71
  q15_t *pS1, *pS2, *pbuff;                      /* Temporary pointers for input buffer and pState buffer */
72
  q15_t in;                                      /* Temporary variable */
73
 
74
 
75
  /* DCT4 computation involves DCT2 (which is calculated using RFFT)    
76
   * along with some pre-processing and post-processing.    
77
   * Computational procedure is explained as follows:    
78
   * (a) Pre-processing involves multiplying input with cos factor,    
79
   *     r(n) = 2 * u(n) * cos(pi*(2*n+1)/(4*n))    
80
   *              where,    
81
   *                 r(n) -- output of preprocessing    
82
   *                 u(n) -- input to preprocessing(actual Source buffer)    
83
   * (b) Calculation of DCT2 using FFT is divided into three steps:    
84
   *                  Step1: Re-ordering of even and odd elements of input.    
85
   *                  Step2: Calculating FFT of the re-ordered input.    
86
   *                  Step3: Taking the real part of the product of FFT output and weights.    
87
   * (c) Post-processing - DCT4 can be obtained from DCT2 output using the following equation:    
88
   *                   Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0)    
89
   *                        where,    
90
   *                           Y4 -- DCT4 output,   Y2 -- DCT2 output    
91
   * (d) Multiplying the output with the normalizing factor sqrt(2/N).    
92
   */
93
 
94
        /*-------- Pre-processing ------------*/
95
  /* Multiplying input with cos factor i.e. r(n) = 2 * x(n) * cos(pi*(2*n+1)/(4*n)) */
96
  arm_mult_q15(pInlineBuffer, cosFact, pInlineBuffer, S->N);
97
  arm_shift_q15(pInlineBuffer, 1, pInlineBuffer, S->N);
98
 
99
  /* ----------------------------------------------------------------    
100
   * Step1: Re-ordering of even and odd elements as    
101
   *             pState[i] =  pInlineBuffer[2*i] and    
102
   *             pState[N-i-1] = pInlineBuffer[2*i+1] where i = 0 to N/2    
103
   ---------------------------------------------------------------------*/
104
 
105
  /* pS1 initialized to pState */
106
  pS1 = pState;
107
 
108
  /* pS2 initialized to pState+N-1, so that it points to the end of the state buffer */
109
  pS2 = pState + (S->N - 1u);
110
 
111
  /* pbuff initialized to input buffer */
112
  pbuff = pInlineBuffer;
113
 
114
 
115
#ifndef ARM_MATH_CM0_FAMILY
116
 
117
  /* Run the below code for Cortex-M4 and Cortex-M3 */
118
 
119
  /* Initializing the loop counter to N/2 >> 2 for loop unrolling by 4 */
120
  i = (uint32_t) S->Nby2 >> 2u;
121
 
122
  /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.    
123
   ** a second loop below computes the remaining 1 to 3 samples. */
124
  do
125
  {
126
    /* Re-ordering of even and odd elements */
127
    /* pState[i] =  pInlineBuffer[2*i] */
128
    *pS1++ = *pbuff++;
129
    /* pState[N-i-1] = pInlineBuffer[2*i+1] */
130
    *pS2-- = *pbuff++;
131
 
132
    *pS1++ = *pbuff++;
133
    *pS2-- = *pbuff++;
134
 
135
    *pS1++ = *pbuff++;
136
    *pS2-- = *pbuff++;
137
 
138
    *pS1++ = *pbuff++;
139
    *pS2-- = *pbuff++;
140
 
141
    /* Decrement the loop counter */
142
    i--;
143
  } while(i > 0u);
144
 
145
  /* pbuff initialized to input buffer */
146
  pbuff = pInlineBuffer;
147
 
148
  /* pS1 initialized to pState */
149
  pS1 = pState;
150
 
151
  /* Initializing the loop counter to N/4 instead of N for loop unrolling */
152
  i = (uint32_t) S->N >> 2u;
153
 
154
  /* Processing with loop unrolling 4 times as N is always multiple of 4.    
155
   * Compute 4 outputs at a time */
156
  do
157
  {
158
    /* Writing the re-ordered output back to inplace input buffer */
159
    *pbuff++ = *pS1++;
160
    *pbuff++ = *pS1++;
161
    *pbuff++ = *pS1++;
162
    *pbuff++ = *pS1++;
163
 
164
    /* Decrement the loop counter */
165
    i--;
166
  } while(i > 0u);
167
 
168
 
169
  /* ---------------------------------------------------------    
170
   *     Step2: Calculate RFFT for N-point input    
171
   * ---------------------------------------------------------- */
172
  /* pInlineBuffer is real input of length N , pState is the complex output of length 2N */
173
  arm_rfft_q15(S->pRfft, pInlineBuffer, pState);
174
 
175
 /*----------------------------------------------------------------------    
176
  *  Step3: Multiply the FFT output with the weights.    
177
  *----------------------------------------------------------------------*/
178
  arm_cmplx_mult_cmplx_q15(pState, weights, pState, S->N);
179
 
180
  /* The output of complex multiplication is in 3.13 format.    
181
   * Hence changing the format of N (i.e. 2*N elements) complex numbers to 1.15 format by shifting left by 2 bits. */
182
  arm_shift_q15(pState, 2, pState, S->N * 2);
183
 
184
  /* ----------- Post-processing ---------- */
185
  /* DCT-IV can be obtained from DCT-II by the equation,    
186
   *       Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0)    
187
   *       Hence, Y4(0) = Y2(0)/2  */
188
  /* Getting only real part from the output and Converting to DCT-IV */
189
 
190
  /* Initializing the loop counter to N >> 2 for loop unrolling by 4 */
191
  i = ((uint32_t) S->N - 1u) >> 2u;
192
 
193
  /* pbuff initialized to input buffer. */
194
  pbuff = pInlineBuffer;
195
 
196
  /* pS1 initialized to pState */
197
  pS1 = pState;
198
 
199
  /* Calculating Y4(0) from Y2(0) using Y4(0) = Y2(0)/2 */
200
  in = *pS1++ >> 1u;
201
  /* input buffer acts as inplace, so output values are stored in the input itself. */
202
  *pbuff++ = in;
203
 
204
  /* pState pointer is incremented twice as the real values are located alternatively in the array */
205
  pS1++;
206
 
207
  /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.    
208
   ** a second loop below computes the remaining 1 to 3 samples. */
209
  do
210
  {
211
    /* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */
212
    /* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */
213
    in = *pS1++ - in;
214
    *pbuff++ = in;
215
    /* points to the next real value */
216
    pS1++;
217
 
218
    in = *pS1++ - in;
219
    *pbuff++ = in;
220
    pS1++;
221
 
222
    in = *pS1++ - in;
223
    *pbuff++ = in;
224
    pS1++;
225
 
226
    in = *pS1++ - in;
227
    *pbuff++ = in;
228
    pS1++;
229
 
230
    /* Decrement the loop counter */
231
    i--;
232
  } while(i > 0u);
233
 
234
  /* If the blockSize is not a multiple of 4, compute any remaining output samples here.    
235
   ** No loop unrolling is used. */
236
  i = ((uint32_t) S->N - 1u) % 0x4u;
237
 
238
  while(i > 0u)
239
  {
240
    /* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */
241
    /* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */
242
    in = *pS1++ - in;
243
    *pbuff++ = in;
244
    /* points to the next real value */
245
    pS1++;
246
 
247
    /* Decrement the loop counter */
248
    i--;
249
  }
250
 
251
 
252
   /*------------ Normalizing the output by multiplying with the normalizing factor ----------*/
253
 
254
  /* Initializing the loop counter to N/4 instead of N for loop unrolling */
255
  i = (uint32_t) S->N >> 2u;
256
 
257
  /* pbuff initialized to the pInlineBuffer(now contains the output values) */
258
  pbuff = pInlineBuffer;
259
 
260
  /* Processing with loop unrolling 4 times as N is always multiple of 4.  Compute 4 outputs at a time */
261
  do
262
  {
263
    /* Multiplying pInlineBuffer with the normalizing factor sqrt(2/N) */
264
    in = *pbuff;
265
    *pbuff++ = ((q15_t) (((q31_t) in * S->normalize) >> 15));
266
 
267
    in = *pbuff;
268
    *pbuff++ = ((q15_t) (((q31_t) in * S->normalize) >> 15));
269
 
270
    in = *pbuff;
271
    *pbuff++ = ((q15_t) (((q31_t) in * S->normalize) >> 15));
272
 
273
    in = *pbuff;
274
    *pbuff++ = ((q15_t) (((q31_t) in * S->normalize) >> 15));
275
 
276
    /* Decrement the loop counter */
277
    i--;
278
  } while(i > 0u);
279
 
280
 
281
#else
282
 
283
  /* Run the below code for Cortex-M0 */
284
 
285
  /* Initializing the loop counter to N/2 */
286
  i = (uint32_t) S->Nby2;
287
 
288
  do
289
  {
290
    /* Re-ordering of even and odd elements */
291
    /* pState[i] =  pInlineBuffer[2*i] */
292
    *pS1++ = *pbuff++;
293
    /* pState[N-i-1] = pInlineBuffer[2*i+1] */
294
    *pS2-- = *pbuff++;
295
 
296
    /* Decrement the loop counter */
297
    i--;
298
  } while(i > 0u);
299
 
300
  /* pbuff initialized to input buffer */
301
  pbuff = pInlineBuffer;
302
 
303
  /* pS1 initialized to pState */
304
  pS1 = pState;
305
 
306
  /* Initializing the loop counter */
307
  i = (uint32_t) S->N;
308
 
309
  do
310
  {
311
    /* Writing the re-ordered output back to inplace input buffer */
312
    *pbuff++ = *pS1++;
313
 
314
    /* Decrement the loop counter */
315
    i--;
316
  } while(i > 0u);
317
 
318
 
319
  /* ---------------------------------------------------------    
320
   *     Step2: Calculate RFFT for N-point input    
321
   * ---------------------------------------------------------- */
322
  /* pInlineBuffer is real input of length N , pState is the complex output of length 2N */
323
  arm_rfft_q15(S->pRfft, pInlineBuffer, pState);
324
 
325
 /*----------------------------------------------------------------------    
326
  *  Step3: Multiply the FFT output with the weights.    
327
  *----------------------------------------------------------------------*/
328
  arm_cmplx_mult_cmplx_q15(pState, weights, pState, S->N);
329
 
330
  /* The output of complex multiplication is in 3.13 format.    
331
   * Hence changing the format of N (i.e. 2*N elements) complex numbers to 1.15 format by shifting left by 2 bits. */
332
  arm_shift_q15(pState, 2, pState, S->N * 2);
333
 
334
  /* ----------- Post-processing ---------- */
335
  /* DCT-IV can be obtained from DCT-II by the equation,    
336
   *       Y4(k) = Y2(k) - Y4(k-1) and Y4(-1) = Y4(0)    
337
   *       Hence, Y4(0) = Y2(0)/2  */
338
  /* Getting only real part from the output and Converting to DCT-IV */
339
 
340
  /* Initializing the loop counter */
341
  i = ((uint32_t) S->N - 1u);
342
 
343
  /* pbuff initialized to input buffer. */
344
  pbuff = pInlineBuffer;
345
 
346
  /* pS1 initialized to pState */
347
  pS1 = pState;
348
 
349
  /* Calculating Y4(0) from Y2(0) using Y4(0) = Y2(0)/2 */
350
  in = *pS1++ >> 1u;
351
  /* input buffer acts as inplace, so output values are stored in the input itself. */
352
  *pbuff++ = in;
353
 
354
  /* pState pointer is incremented twice as the real values are located alternatively in the array */
355
  pS1++;
356
 
357
  do
358
  {
359
    /* Calculating Y4(1) to Y4(N-1) from Y2 using equation Y4(k) = Y2(k) - Y4(k-1) */
360
    /* pState pointer (pS1) is incremented twice as the real values are located alternatively in the array */
361
    in = *pS1++ - in;
362
    *pbuff++ = in;
363
    /* points to the next real value */
364
    pS1++;
365
 
366
    /* Decrement the loop counter */
367
    i--;
368
  } while(i > 0u);
369
 
370
   /*------------ Normalizing the output by multiplying with the normalizing factor ----------*/
371
 
372
  /* Initializing the loop counter */
373
  i = (uint32_t) S->N;
374
 
375
  /* pbuff initialized to the pInlineBuffer(now contains the output values) */
376
  pbuff = pInlineBuffer;
377
 
378
  do
379
  {
380
    /* Multiplying pInlineBuffer with the normalizing factor sqrt(2/N) */
381
    in = *pbuff;
382
    *pbuff++ = ((q15_t) (((q31_t) in * S->normalize) >> 15));
383
 
384
    /* Decrement the loop counter */
385
    i--;
386
  } while(i > 0u);
387
 
388
#endif /* #ifndef ARM_MATH_CM0_FAMILY */
389
 
390
}
391
 
392
/**    
393
   * @} end of DCT4_IDCT4 group    
394
   */