Rev 28 | Details | Compare with Previous | 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_cfft_radix4_f32.c |
||
9 | * |
||
10 | * Description: Radix-4 Decimation in Frequency CFFT & CIFFT Floating point processing function |
||
11 | * |
||
12 | * |
||
13 | * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0 |
||
14 | * |
||
15 | * Redistribution and use in source and binary forms, with or without |
||
16 | * modification, are permitted provided that the following conditions |
||
17 | * are met: |
||
18 | * - Redistributions of source code must retain the above copyright |
||
19 | * notice, this list of conditions and the following disclaimer. |
||
20 | * - Redistributions in binary form must reproduce the above copyright |
||
21 | * notice, this list of conditions and the following disclaimer in |
||
22 | * the documentation and/or other materials provided with the |
||
23 | * distribution. |
||
24 | * - Neither the name of ARM LIMITED nor the names of its contributors |
||
25 | * may be used to endorse or promote products derived from this |
||
26 | * software without specific prior written permission. |
||
27 | * |
||
28 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
||
29 | * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
||
30 | * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS |
||
31 | * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE |
||
32 | * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, |
||
33 | * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, |
||
34 | * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; |
||
35 | * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER |
||
36 | * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
||
37 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN |
||
38 | * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
||
39 | * POSSIBILITY OF SUCH DAMAGE. |
||
40 | * -------------------------------------------------------------------- */ |
||
41 | |||
42 | #include "arm_math.h" |
||
43 | |||
44 | extern void arm_bitreversal_f32( |
||
45 | float32_t * pSrc, |
||
46 | uint16_t fftSize, |
||
47 | uint16_t bitRevFactor, |
||
48 | uint16_t * pBitRevTab); |
||
49 | |||
50 | /** |
||
51 | * @ingroup groupTransforms |
||
52 | */ |
||
53 | |||
54 | /* ---------------------------------------------------------------------- |
||
55 | ** Internal helper function used by the FFTs |
||
56 | ** ------------------------------------------------------------------- */ |
||
57 | |||
58 | /* |
||
59 | * @brief Core function for the floating-point CFFT butterfly process. |
||
60 | * @param[in, out] *pSrc points to the in-place buffer of floating-point data type. |
||
61 | * @param[in] fftLen length of the FFT. |
||
62 | * @param[in] *pCoef points to the twiddle coefficient buffer. |
||
63 | * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table. |
||
64 | * @return none. |
||
65 | */ |
||
66 | |||
67 | void arm_radix4_butterfly_f32( |
||
68 | float32_t * pSrc, |
||
69 | uint16_t fftLen, |
||
70 | float32_t * pCoef, |
||
71 | uint16_t twidCoefModifier) |
||
72 | { |
||
73 | |||
74 | float32_t co1, co2, co3, si1, si2, si3; |
||
75 | uint32_t ia1, ia2, ia3; |
||
76 | uint32_t i0, i1, i2, i3; |
||
77 | uint32_t n1, n2, j, k; |
||
78 | |||
79 | #ifndef ARM_MATH_CM0_FAMILY_FAMILY |
||
80 | |||
81 | /* Run the below code for Cortex-M4 and Cortex-M3 */ |
||
82 | |||
83 | float32_t xaIn, yaIn, xbIn, ybIn, xcIn, ycIn, xdIn, ydIn; |
||
84 | float32_t Xaplusc, Xbplusd, Yaplusc, Ybplusd, Xaminusc, Xbminusd, Yaminusc, |
||
85 | Ybminusd; |
||
86 | float32_t Xb12C_out, Yb12C_out, Xc12C_out, Yc12C_out, Xd12C_out, Yd12C_out; |
||
87 | float32_t Xb12_out, Yb12_out, Xc12_out, Yc12_out, Xd12_out, Yd12_out; |
||
88 | float32_t *ptr1; |
||
89 | float32_t p0,p1,p2,p3,p4,p5; |
||
90 | float32_t a0,a1,a2,a3,a4,a5,a6,a7; |
||
91 | |||
92 | /* Initializations for the first stage */ |
||
93 | n2 = fftLen; |
||
94 | n1 = n2; |
||
95 | |||
96 | /* n2 = fftLen/4 */ |
||
97 | n2 >>= 2u; |
||
98 | i0 = 0u; |
||
99 | ia1 = 0u; |
||
100 | |||
101 | j = n2; |
||
102 | |||
103 | /* Calculation of first stage */ |
||
104 | do |
||
105 | { |
||
106 | /* index calculation for the input as, */ |
||
107 | /* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */ |
||
108 | i1 = i0 + n2; |
||
109 | i2 = i1 + n2; |
||
110 | i3 = i2 + n2; |
||
111 | |||
112 | xaIn = pSrc[(2u * i0)]; |
||
113 | yaIn = pSrc[(2u * i0) + 1u]; |
||
114 | |||
115 | xbIn = pSrc[(2u * i1)]; |
||
116 | ybIn = pSrc[(2u * i1) + 1u]; |
||
117 | |||
118 | xcIn = pSrc[(2u * i2)]; |
||
119 | ycIn = pSrc[(2u * i2) + 1u]; |
||
120 | |||
121 | xdIn = pSrc[(2u * i3)]; |
||
122 | ydIn = pSrc[(2u * i3) + 1u]; |
||
123 | |||
124 | /* xa + xc */ |
||
125 | Xaplusc = xaIn + xcIn; |
||
126 | /* xb + xd */ |
||
127 | Xbplusd = xbIn + xdIn; |
||
128 | /* ya + yc */ |
||
129 | Yaplusc = yaIn + ycIn; |
||
130 | /* yb + yd */ |
||
131 | Ybplusd = ybIn + ydIn; |
||
132 | |||
133 | /* index calculation for the coefficients */ |
||
134 | ia2 = ia1 + ia1; |
||
135 | co2 = pCoef[ia2 * 2u]; |
||
136 | si2 = pCoef[(ia2 * 2u) + 1u]; |
||
137 | |||
138 | /* xa - xc */ |
||
139 | Xaminusc = xaIn - xcIn; |
||
140 | /* xb - xd */ |
||
141 | Xbminusd = xbIn - xdIn; |
||
142 | /* ya - yc */ |
||
143 | Yaminusc = yaIn - ycIn; |
||
144 | /* yb - yd */ |
||
145 | Ybminusd = ybIn - ydIn; |
||
146 | |||
147 | /* xa' = xa + xb + xc + xd */ |
||
148 | pSrc[(2u * i0)] = Xaplusc + Xbplusd; |
||
149 | /* ya' = ya + yb + yc + yd */ |
||
150 | pSrc[(2u * i0) + 1u] = Yaplusc + Ybplusd; |
||
151 | |||
152 | /* (xa - xc) + (yb - yd) */ |
||
153 | Xb12C_out = (Xaminusc + Ybminusd); |
||
154 | /* (ya - yc) + (xb - xd) */ |
||
155 | Yb12C_out = (Yaminusc - Xbminusd); |
||
156 | /* (xa + xc) - (xb + xd) */ |
||
157 | Xc12C_out = (Xaplusc - Xbplusd); |
||
158 | /* (ya + yc) - (yb + yd) */ |
||
159 | Yc12C_out = (Yaplusc - Ybplusd); |
||
160 | /* (xa - xc) - (yb - yd) */ |
||
161 | Xd12C_out = (Xaminusc - Ybminusd); |
||
162 | /* (ya - yc) + (xb - xd) */ |
||
163 | Yd12C_out = (Xbminusd + Yaminusc); |
||
164 | |||
165 | co1 = pCoef[ia1 * 2u]; |
||
166 | si1 = pCoef[(ia1 * 2u) + 1u]; |
||
167 | |||
168 | /* index calculation for the coefficients */ |
||
169 | ia3 = ia2 + ia1; |
||
170 | co3 = pCoef[ia3 * 2u]; |
||
171 | si3 = pCoef[(ia3 * 2u) + 1u]; |
||
172 | |||
173 | Xb12_out = Xb12C_out * co1; |
||
174 | Yb12_out = Yb12C_out * co1; |
||
175 | Xc12_out = Xc12C_out * co2; |
||
176 | Yc12_out = Yc12C_out * co2; |
||
177 | Xd12_out = Xd12C_out * co3; |
||
178 | Yd12_out = Yd12C_out * co3; |
||
179 | |||
180 | /* xb' = (xa+yb-xc-yd)co1 - (ya-xb-yc+xd)(si1) */ |
||
181 | //Xb12_out -= Yb12C_out * si1; |
||
182 | p0 = Yb12C_out * si1; |
||
183 | /* yb' = (ya-xb-yc+xd)co1 + (xa+yb-xc-yd)(si1) */ |
||
184 | //Yb12_out += Xb12C_out * si1; |
||
185 | p1 = Xb12C_out * si1; |
||
186 | /* xc' = (xa-xb+xc-xd)co2 - (ya-yb+yc-yd)(si2) */ |
||
187 | //Xc12_out -= Yc12C_out * si2; |
||
188 | p2 = Yc12C_out * si2; |
||
189 | /* yc' = (ya-yb+yc-yd)co2 + (xa-xb+xc-xd)(si2) */ |
||
190 | //Yc12_out += Xc12C_out * si2; |
||
191 | p3 = Xc12C_out * si2; |
||
192 | /* xd' = (xa-yb-xc+yd)co3 - (ya+xb-yc-xd)(si3) */ |
||
193 | //Xd12_out -= Yd12C_out * si3; |
||
194 | p4 = Yd12C_out * si3; |
||
195 | /* yd' = (ya+xb-yc-xd)co3 + (xa-yb-xc+yd)(si3) */ |
||
196 | //Yd12_out += Xd12C_out * si3; |
||
197 | p5 = Xd12C_out * si3; |
||
198 | |||
199 | Xb12_out += p0; |
||
200 | Yb12_out -= p1; |
||
201 | Xc12_out += p2; |
||
202 | Yc12_out -= p3; |
||
203 | Xd12_out += p4; |
||
204 | Yd12_out -= p5; |
||
205 | |||
206 | /* xc' = (xa-xb+xc-xd)co2 + (ya-yb+yc-yd)(si2) */ |
||
207 | pSrc[2u * i1] = Xc12_out; |
||
208 | |||
209 | /* yc' = (ya-yb+yc-yd)co2 - (xa-xb+xc-xd)(si2) */ |
||
210 | pSrc[(2u * i1) + 1u] = Yc12_out; |
||
211 | |||
212 | /* xb' = (xa+yb-xc-yd)co1 + (ya-xb-yc+xd)(si1) */ |
||
213 | pSrc[2u * i2] = Xb12_out; |
||
214 | |||
215 | /* yb' = (ya-xb-yc+xd)co1 - (xa+yb-xc-yd)(si1) */ |
||
216 | pSrc[(2u * i2) + 1u] = Yb12_out; |
||
217 | |||
218 | /* xd' = (xa-yb-xc+yd)co3 + (ya+xb-yc-xd)(si3) */ |
||
219 | pSrc[2u * i3] = Xd12_out; |
||
220 | |||
221 | /* yd' = (ya+xb-yc-xd)co3 - (xa-yb-xc+yd)(si3) */ |
||
222 | pSrc[(2u * i3) + 1u] = Yd12_out; |
||
223 | |||
224 | /* Twiddle coefficients index modifier */ |
||
225 | ia1 += twidCoefModifier; |
||
226 | |||
227 | /* Updating input index */ |
||
228 | i0++; |
||
229 | |||
230 | } |
||
231 | while(--j); |
||
232 | |||
233 | twidCoefModifier <<= 2u; |
||
234 | |||
235 | /* Calculation of second stage to excluding last stage */ |
||
236 | for (k = fftLen >> 2u; k > 4u; k >>= 2u) |
||
237 | { |
||
238 | /* Initializations for the first stage */ |
||
239 | n1 = n2; |
||
240 | n2 >>= 2u; |
||
241 | ia1 = 0u; |
||
242 | |||
243 | /* Calculation of first stage */ |
||
244 | j = 0; |
||
245 | do |
||
246 | { |
||
247 | /* index calculation for the coefficients */ |
||
248 | ia2 = ia1 + ia1; |
||
249 | ia3 = ia2 + ia1; |
||
250 | co1 = pCoef[ia1 * 2u]; |
||
251 | si1 = pCoef[(ia1 * 2u) + 1u]; |
||
252 | co2 = pCoef[ia2 * 2u]; |
||
253 | si2 = pCoef[(ia2 * 2u) + 1u]; |
||
254 | co3 = pCoef[ia3 * 2u]; |
||
255 | si3 = pCoef[(ia3 * 2u) + 1u]; |
||
256 | |||
257 | /* Twiddle coefficients index modifier */ |
||
258 | ia1 += twidCoefModifier; |
||
259 | |||
260 | i0 = j; |
||
261 | do |
||
262 | { |
||
263 | /* index calculation for the input as, */ |
||
264 | /* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */ |
||
265 | i1 = i0 + n2; |
||
266 | i2 = i1 + n2; |
||
267 | i3 = i2 + n2; |
||
268 | |||
269 | xaIn = pSrc[(2u * i0)]; |
||
270 | yaIn = pSrc[(2u * i0) + 1u]; |
||
271 | |||
272 | xbIn = pSrc[(2u * i1)]; |
||
273 | ybIn = pSrc[(2u * i1) + 1u]; |
||
274 | |||
275 | xcIn = pSrc[(2u * i2)]; |
||
276 | ycIn = pSrc[(2u * i2) + 1u]; |
||
277 | |||
278 | xdIn = pSrc[(2u * i3)]; |
||
279 | ydIn = pSrc[(2u * i3) + 1u]; |
||
280 | |||
281 | /* xa - xc */ |
||
282 | Xaminusc = xaIn - xcIn; |
||
283 | /* (xb - xd) */ |
||
284 | Xbminusd = xbIn - xdIn; |
||
285 | /* ya - yc */ |
||
286 | Yaminusc = yaIn - ycIn; |
||
287 | /* (yb - yd) */ |
||
288 | Ybminusd = ybIn - ydIn; |
||
289 | |||
290 | /* xa + xc */ |
||
291 | Xaplusc = xaIn + xcIn; |
||
292 | /* xb + xd */ |
||
293 | Xbplusd = xbIn + xdIn; |
||
294 | /* ya + yc */ |
||
295 | Yaplusc = yaIn + ycIn; |
||
296 | /* yb + yd */ |
||
297 | Ybplusd = ybIn + ydIn; |
||
298 | |||
299 | /* (xa - xc) + (yb - yd) */ |
||
300 | Xb12C_out = (Xaminusc + Ybminusd); |
||
301 | /* (ya - yc) - (xb - xd) */ |
||
302 | Yb12C_out = (Yaminusc - Xbminusd); |
||
303 | /* xa + xc -(xb + xd) */ |
||
304 | Xc12C_out = (Xaplusc - Xbplusd); |
||
305 | /* (ya + yc) - (yb + yd) */ |
||
306 | Yc12C_out = (Yaplusc - Ybplusd); |
||
307 | /* (xa - xc) - (yb - yd) */ |
||
308 | Xd12C_out = (Xaminusc - Ybminusd); |
||
309 | /* (ya - yc) + (xb - xd) */ |
||
310 | Yd12C_out = (Xbminusd + Yaminusc); |
||
311 | |||
312 | pSrc[(2u * i0)] = Xaplusc + Xbplusd; |
||
313 | pSrc[(2u * i0) + 1u] = Yaplusc + Ybplusd; |
||
314 | |||
315 | Xb12_out = Xb12C_out * co1; |
||
316 | Yb12_out = Yb12C_out * co1; |
||
317 | Xc12_out = Xc12C_out * co2; |
||
318 | Yc12_out = Yc12C_out * co2; |
||
319 | Xd12_out = Xd12C_out * co3; |
||
320 | Yd12_out = Yd12C_out * co3; |
||
321 | |||
322 | /* xb' = (xa+yb-xc-yd)co1 - (ya-xb-yc+xd)(si1) */ |
||
323 | //Xb12_out -= Yb12C_out * si1; |
||
324 | p0 = Yb12C_out * si1; |
||
325 | /* yb' = (ya-xb-yc+xd)co1 + (xa+yb-xc-yd)(si1) */ |
||
326 | //Yb12_out += Xb12C_out * si1; |
||
327 | p1 = Xb12C_out * si1; |
||
328 | /* xc' = (xa-xb+xc-xd)co2 - (ya-yb+yc-yd)(si2) */ |
||
329 | //Xc12_out -= Yc12C_out * si2; |
||
330 | p2 = Yc12C_out * si2; |
||
331 | /* yc' = (ya-yb+yc-yd)co2 + (xa-xb+xc-xd)(si2) */ |
||
332 | //Yc12_out += Xc12C_out * si2; |
||
333 | p3 = Xc12C_out * si2; |
||
334 | /* xd' = (xa-yb-xc+yd)co3 - (ya+xb-yc-xd)(si3) */ |
||
335 | //Xd12_out -= Yd12C_out * si3; |
||
336 | p4 = Yd12C_out * si3; |
||
337 | /* yd' = (ya+xb-yc-xd)co3 + (xa-yb-xc+yd)(si3) */ |
||
338 | //Yd12_out += Xd12C_out * si3; |
||
339 | p5 = Xd12C_out * si3; |
||
340 | |||
341 | Xb12_out += p0; |
||
342 | Yb12_out -= p1; |
||
343 | Xc12_out += p2; |
||
344 | Yc12_out -= p3; |
||
345 | Xd12_out += p4; |
||
346 | Yd12_out -= p5; |
||
347 | |||
348 | /* xc' = (xa-xb+xc-xd)co2 + (ya-yb+yc-yd)(si2) */ |
||
349 | pSrc[2u * i1] = Xc12_out; |
||
350 | |||
351 | /* yc' = (ya-yb+yc-yd)co2 - (xa-xb+xc-xd)(si2) */ |
||
352 | pSrc[(2u * i1) + 1u] = Yc12_out; |
||
353 | |||
354 | /* xb' = (xa+yb-xc-yd)co1 + (ya-xb-yc+xd)(si1) */ |
||
355 | pSrc[2u * i2] = Xb12_out; |
||
356 | |||
357 | /* yb' = (ya-xb-yc+xd)co1 - (xa+yb-xc-yd)(si1) */ |
||
358 | pSrc[(2u * i2) + 1u] = Yb12_out; |
||
359 | |||
360 | /* xd' = (xa-yb-xc+yd)co3 + (ya+xb-yc-xd)(si3) */ |
||
361 | pSrc[2u * i3] = Xd12_out; |
||
362 | |||
363 | /* yd' = (ya+xb-yc-xd)co3 - (xa-yb-xc+yd)(si3) */ |
||
364 | pSrc[(2u * i3) + 1u] = Yd12_out; |
||
365 | |||
366 | i0 += n1; |
||
367 | } while(i0 < fftLen); |
||
368 | j++; |
||
369 | } while(j <= (n2 - 1u)); |
||
370 | twidCoefModifier <<= 2u; |
||
371 | } |
||
372 | |||
373 | j = fftLen >> 2; |
||
374 | ptr1 = &pSrc[0]; |
||
375 | |||
376 | /* Calculations of last stage */ |
||
377 | do |
||
378 | { |
||
379 | xaIn = ptr1[0]; |
||
380 | yaIn = ptr1[1]; |
||
381 | xbIn = ptr1[2]; |
||
382 | ybIn = ptr1[3]; |
||
383 | xcIn = ptr1[4]; |
||
384 | ycIn = ptr1[5]; |
||
385 | xdIn = ptr1[6]; |
||
386 | ydIn = ptr1[7]; |
||
387 | |||
388 | /* xa + xc */ |
||
389 | Xaplusc = xaIn + xcIn; |
||
390 | |||
391 | /* xa - xc */ |
||
392 | Xaminusc = xaIn - xcIn; |
||
393 | |||
394 | /* ya + yc */ |
||
395 | Yaplusc = yaIn + ycIn; |
||
396 | |||
397 | /* ya - yc */ |
||
398 | Yaminusc = yaIn - ycIn; |
||
399 | |||
400 | /* xb + xd */ |
||
401 | Xbplusd = xbIn + xdIn; |
||
402 | |||
403 | /* yb + yd */ |
||
404 | Ybplusd = ybIn + ydIn; |
||
405 | |||
406 | /* (xb-xd) */ |
||
407 | Xbminusd = xbIn - xdIn; |
||
408 | |||
409 | /* (yb-yd) */ |
||
410 | Ybminusd = ybIn - ydIn; |
||
411 | |||
412 | /* xa' = xa + xb + xc + xd */ |
||
413 | a0 = (Xaplusc + Xbplusd); |
||
414 | /* ya' = ya + yb + yc + yd */ |
||
415 | a1 = (Yaplusc + Ybplusd); |
||
416 | /* xc' = (xa-xb+xc-xd) */ |
||
417 | a2 = (Xaplusc - Xbplusd); |
||
418 | /* yc' = (ya-yb+yc-yd) */ |
||
419 | a3 = (Yaplusc - Ybplusd); |
||
420 | /* xb' = (xa+yb-xc-yd) */ |
||
421 | a4 = (Xaminusc + Ybminusd); |
||
422 | /* yb' = (ya-xb-yc+xd) */ |
||
423 | a5 = (Yaminusc - Xbminusd); |
||
424 | /* xd' = (xa-yb-xc+yd)) */ |
||
425 | a6 = (Xaminusc - Ybminusd); |
||
426 | /* yd' = (ya+xb-yc-xd) */ |
||
427 | a7 = (Xbminusd + Yaminusc); |
||
428 | |||
429 | ptr1[0] = a0; |
||
430 | ptr1[1] = a1; |
||
431 | ptr1[2] = a2; |
||
432 | ptr1[3] = a3; |
||
433 | ptr1[4] = a4; |
||
434 | ptr1[5] = a5; |
||
435 | ptr1[6] = a6; |
||
436 | ptr1[7] = a7; |
||
437 | |||
438 | /* increment pointer by 8 */ |
||
439 | ptr1 += 8u; |
||
440 | } while(--j); |
||
441 | |||
442 | #else |
||
443 | |||
444 | float32_t t1, t2, r1, r2, s1, s2; |
||
445 | |||
446 | /* Run the below code for Cortex-M0 */ |
||
447 | |||
448 | /* Initializations for the fft calculation */ |
||
449 | n2 = fftLen; |
||
450 | n1 = n2; |
||
451 | for (k = fftLen; k > 1u; k >>= 2u) |
||
452 | { |
||
453 | /* Initializations for the fft calculation */ |
||
454 | n1 = n2; |
||
455 | n2 >>= 2u; |
||
456 | ia1 = 0u; |
||
457 | |||
458 | /* FFT Calculation */ |
||
459 | j = 0; |
||
460 | do |
||
461 | { |
||
462 | /* index calculation for the coefficients */ |
||
463 | ia2 = ia1 + ia1; |
||
464 | ia3 = ia2 + ia1; |
||
465 | co1 = pCoef[ia1 * 2u]; |
||
466 | si1 = pCoef[(ia1 * 2u) + 1u]; |
||
467 | co2 = pCoef[ia2 * 2u]; |
||
468 | si2 = pCoef[(ia2 * 2u) + 1u]; |
||
469 | co3 = pCoef[ia3 * 2u]; |
||
470 | si3 = pCoef[(ia3 * 2u) + 1u]; |
||
471 | |||
472 | /* Twiddle coefficients index modifier */ |
||
473 | ia1 = ia1 + twidCoefModifier; |
||
474 | |||
475 | i0 = j; |
||
476 | do |
||
477 | { |
||
478 | /* index calculation for the input as, */ |
||
479 | /* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */ |
||
480 | i1 = i0 + n2; |
||
481 | i2 = i1 + n2; |
||
482 | i3 = i2 + n2; |
||
483 | |||
484 | /* xa + xc */ |
||
485 | r1 = pSrc[(2u * i0)] + pSrc[(2u * i2)]; |
||
486 | |||
487 | /* xa - xc */ |
||
488 | r2 = pSrc[(2u * i0)] - pSrc[(2u * i2)]; |
||
489 | |||
490 | /* ya + yc */ |
||
491 | s1 = pSrc[(2u * i0) + 1u] + pSrc[(2u * i2) + 1u]; |
||
492 | |||
493 | /* ya - yc */ |
||
494 | s2 = pSrc[(2u * i0) + 1u] - pSrc[(2u * i2) + 1u]; |
||
495 | |||
496 | /* xb + xd */ |
||
497 | t1 = pSrc[2u * i1] + pSrc[2u * i3]; |
||
498 | |||
499 | /* xa' = xa + xb + xc + xd */ |
||
500 | pSrc[2u * i0] = r1 + t1; |
||
501 | |||
502 | /* xa + xc -(xb + xd) */ |
||
503 | r1 = r1 - t1; |
||
504 | |||
505 | /* yb + yd */ |
||
506 | t2 = pSrc[(2u * i1) + 1u] + pSrc[(2u * i3) + 1u]; |
||
507 | |||
508 | /* ya' = ya + yb + yc + yd */ |
||
509 | pSrc[(2u * i0) + 1u] = s1 + t2; |
||
510 | |||
511 | /* (ya + yc) - (yb + yd) */ |
||
512 | s1 = s1 - t2; |
||
513 | |||
514 | /* (yb - yd) */ |
||
515 | t1 = pSrc[(2u * i1) + 1u] - pSrc[(2u * i3) + 1u]; |
||
516 | |||
517 | /* (xb - xd) */ |
||
518 | t2 = pSrc[2u * i1] - pSrc[2u * i3]; |
||
519 | |||
520 | /* xc' = (xa-xb+xc-xd)co2 + (ya-yb+yc-yd)(si2) */ |
||
521 | pSrc[2u * i1] = (r1 * co2) + (s1 * si2); |
||
522 | |||
523 | /* yc' = (ya-yb+yc-yd)co2 - (xa-xb+xc-xd)(si2) */ |
||
524 | pSrc[(2u * i1) + 1u] = (s1 * co2) - (r1 * si2); |
||
525 | |||
526 | /* (xa - xc) + (yb - yd) */ |
||
527 | r1 = r2 + t1; |
||
528 | |||
529 | /* (xa - xc) - (yb - yd) */ |
||
530 | r2 = r2 - t1; |
||
531 | |||
532 | /* (ya - yc) - (xb - xd) */ |
||
533 | s1 = s2 - t2; |
||
534 | |||
535 | /* (ya - yc) + (xb - xd) */ |
||
536 | s2 = s2 + t2; |
||
537 | |||
538 | /* xb' = (xa+yb-xc-yd)co1 + (ya-xb-yc+xd)(si1) */ |
||
539 | pSrc[2u * i2] = (r1 * co1) + (s1 * si1); |
||
540 | |||
541 | /* yb' = (ya-xb-yc+xd)co1 - (xa+yb-xc-yd)(si1) */ |
||
542 | pSrc[(2u * i2) + 1u] = (s1 * co1) - (r1 * si1); |
||
543 | |||
544 | /* xd' = (xa-yb-xc+yd)co3 + (ya+xb-yc-xd)(si3) */ |
||
545 | pSrc[2u * i3] = (r2 * co3) + (s2 * si3); |
||
546 | |||
547 | /* yd' = (ya+xb-yc-xd)co3 - (xa-yb-xc+yd)(si3) */ |
||
548 | pSrc[(2u * i3) + 1u] = (s2 * co3) - (r2 * si3); |
||
549 | |||
550 | i0 += n1; |
||
551 | } while( i0 < fftLen); |
||
552 | j++; |
||
553 | } while(j <= (n2 - 1u)); |
||
554 | twidCoefModifier <<= 2u; |
||
555 | } |
||
556 | |||
557 | #endif /* #ifndef ARM_MATH_CM0_FAMILY_FAMILY */ |
||
558 | |||
559 | } |
||
560 | |||
561 | /* |
||
562 | * @brief Core function for the floating-point CIFFT butterfly process. |
||
563 | * @param[in, out] *pSrc points to the in-place buffer of floating-point data type. |
||
564 | * @param[in] fftLen length of the FFT. |
||
565 | * @param[in] *pCoef points to twiddle coefficient buffer. |
||
566 | * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table. |
||
567 | * @param[in] onebyfftLen value of 1/fftLen. |
||
568 | * @return none. |
||
569 | */ |
||
570 | |||
571 | void arm_radix4_butterfly_inverse_f32( |
||
572 | float32_t * pSrc, |
||
573 | uint16_t fftLen, |
||
574 | float32_t * pCoef, |
||
575 | uint16_t twidCoefModifier, |
||
576 | float32_t onebyfftLen) |
||
577 | { |
||
578 | float32_t co1, co2, co3, si1, si2, si3; |
||
579 | uint32_t ia1, ia2, ia3; |
||
580 | uint32_t i0, i1, i2, i3; |
||
581 | uint32_t n1, n2, j, k; |
||
582 | |||
583 | #ifndef ARM_MATH_CM0_FAMILY_FAMILY |
||
584 | |||
585 | float32_t xaIn, yaIn, xbIn, ybIn, xcIn, ycIn, xdIn, ydIn; |
||
586 | float32_t Xaplusc, Xbplusd, Yaplusc, Ybplusd, Xaminusc, Xbminusd, Yaminusc, |
||
587 | Ybminusd; |
||
588 | float32_t Xb12C_out, Yb12C_out, Xc12C_out, Yc12C_out, Xd12C_out, Yd12C_out; |
||
589 | float32_t Xb12_out, Yb12_out, Xc12_out, Yc12_out, Xd12_out, Yd12_out; |
||
590 | float32_t *ptr1; |
||
591 | float32_t p0,p1,p2,p3,p4,p5,p6,p7; |
||
592 | float32_t a0,a1,a2,a3,a4,a5,a6,a7; |
||
593 | |||
594 | |||
595 | /* Initializations for the first stage */ |
||
596 | n2 = fftLen; |
||
597 | n1 = n2; |
||
598 | |||
599 | /* n2 = fftLen/4 */ |
||
600 | n2 >>= 2u; |
||
601 | i0 = 0u; |
||
602 | ia1 = 0u; |
||
603 | |||
604 | j = n2; |
||
605 | |||
606 | /* Calculation of first stage */ |
||
607 | do |
||
608 | { |
||
609 | /* index calculation for the input as, */ |
||
610 | /* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */ |
||
611 | i1 = i0 + n2; |
||
612 | i2 = i1 + n2; |
||
613 | i3 = i2 + n2; |
||
614 | |||
615 | /* Butterfly implementation */ |
||
616 | xaIn = pSrc[(2u * i0)]; |
||
617 | yaIn = pSrc[(2u * i0) + 1u]; |
||
618 | |||
619 | xcIn = pSrc[(2u * i2)]; |
||
620 | ycIn = pSrc[(2u * i2) + 1u]; |
||
621 | |||
622 | xbIn = pSrc[(2u * i1)]; |
||
623 | ybIn = pSrc[(2u * i1) + 1u]; |
||
624 | |||
625 | xdIn = pSrc[(2u * i3)]; |
||
626 | ydIn = pSrc[(2u * i3) + 1u]; |
||
627 | |||
628 | /* xa + xc */ |
||
629 | Xaplusc = xaIn + xcIn; |
||
630 | /* xb + xd */ |
||
631 | Xbplusd = xbIn + xdIn; |
||
632 | /* ya + yc */ |
||
633 | Yaplusc = yaIn + ycIn; |
||
634 | /* yb + yd */ |
||
635 | Ybplusd = ybIn + ydIn; |
||
636 | |||
637 | /* index calculation for the coefficients */ |
||
638 | ia2 = ia1 + ia1; |
||
639 | co2 = pCoef[ia2 * 2u]; |
||
640 | si2 = pCoef[(ia2 * 2u) + 1u]; |
||
641 | |||
642 | /* xa - xc */ |
||
643 | Xaminusc = xaIn - xcIn; |
||
644 | /* xb - xd */ |
||
645 | Xbminusd = xbIn - xdIn; |
||
646 | /* ya - yc */ |
||
647 | Yaminusc = yaIn - ycIn; |
||
648 | /* yb - yd */ |
||
649 | Ybminusd = ybIn - ydIn; |
||
650 | |||
651 | /* xa' = xa + xb + xc + xd */ |
||
652 | pSrc[(2u * i0)] = Xaplusc + Xbplusd; |
||
653 | |||
654 | /* ya' = ya + yb + yc + yd */ |
||
655 | pSrc[(2u * i0) + 1u] = Yaplusc + Ybplusd; |
||
656 | |||
657 | /* (xa - xc) - (yb - yd) */ |
||
658 | Xb12C_out = (Xaminusc - Ybminusd); |
||
659 | /* (ya - yc) + (xb - xd) */ |
||
660 | Yb12C_out = (Yaminusc + Xbminusd); |
||
661 | /* (xa + xc) - (xb + xd) */ |
||
662 | Xc12C_out = (Xaplusc - Xbplusd); |
||
663 | /* (ya + yc) - (yb + yd) */ |
||
664 | Yc12C_out = (Yaplusc - Ybplusd); |
||
665 | /* (xa - xc) + (yb - yd) */ |
||
666 | Xd12C_out = (Xaminusc + Ybminusd); |
||
667 | /* (ya - yc) - (xb - xd) */ |
||
668 | Yd12C_out = (Yaminusc - Xbminusd); |
||
669 | |||
670 | co1 = pCoef[ia1 * 2u]; |
||
671 | si1 = pCoef[(ia1 * 2u) + 1u]; |
||
672 | |||
673 | /* index calculation for the coefficients */ |
||
674 | ia3 = ia2 + ia1; |
||
675 | co3 = pCoef[ia3 * 2u]; |
||
676 | si3 = pCoef[(ia3 * 2u) + 1u]; |
||
677 | |||
678 | Xb12_out = Xb12C_out * co1; |
||
679 | Yb12_out = Yb12C_out * co1; |
||
680 | Xc12_out = Xc12C_out * co2; |
||
681 | Yc12_out = Yc12C_out * co2; |
||
682 | Xd12_out = Xd12C_out * co3; |
||
683 | Yd12_out = Yd12C_out * co3; |
||
684 | |||
685 | /* xb' = (xa+yb-xc-yd)co1 - (ya-xb-yc+xd)(si1) */ |
||
686 | //Xb12_out -= Yb12C_out * si1; |
||
687 | p0 = Yb12C_out * si1; |
||
688 | /* yb' = (ya-xb-yc+xd)co1 + (xa+yb-xc-yd)(si1) */ |
||
689 | //Yb12_out += Xb12C_out * si1; |
||
690 | p1 = Xb12C_out * si1; |
||
691 | /* xc' = (xa-xb+xc-xd)co2 - (ya-yb+yc-yd)(si2) */ |
||
692 | //Xc12_out -= Yc12C_out * si2; |
||
693 | p2 = Yc12C_out * si2; |
||
694 | /* yc' = (ya-yb+yc-yd)co2 + (xa-xb+xc-xd)(si2) */ |
||
695 | //Yc12_out += Xc12C_out * si2; |
||
696 | p3 = Xc12C_out * si2; |
||
697 | /* xd' = (xa-yb-xc+yd)co3 - (ya+xb-yc-xd)(si3) */ |
||
698 | //Xd12_out -= Yd12C_out * si3; |
||
699 | p4 = Yd12C_out * si3; |
||
700 | /* yd' = (ya+xb-yc-xd)co3 + (xa-yb-xc+yd)(si3) */ |
||
701 | //Yd12_out += Xd12C_out * si3; |
||
702 | p5 = Xd12C_out * si3; |
||
703 | |||
704 | Xb12_out -= p0; |
||
705 | Yb12_out += p1; |
||
706 | Xc12_out -= p2; |
||
707 | Yc12_out += p3; |
||
708 | Xd12_out -= p4; |
||
709 | Yd12_out += p5; |
||
710 | |||
711 | /* xc' = (xa-xb+xc-xd)co2 - (ya-yb+yc-yd)(si2) */ |
||
712 | pSrc[2u * i1] = Xc12_out; |
||
713 | |||
714 | /* yc' = (ya-yb+yc-yd)co2 + (xa-xb+xc-xd)(si2) */ |
||
715 | pSrc[(2u * i1) + 1u] = Yc12_out; |
||
716 | |||
717 | /* xb' = (xa+yb-xc-yd)co1 - (ya-xb-yc+xd)(si1) */ |
||
718 | pSrc[2u * i2] = Xb12_out; |
||
719 | |||
720 | /* yb' = (ya-xb-yc+xd)co1 + (xa+yb-xc-yd)(si1) */ |
||
721 | pSrc[(2u * i2) + 1u] = Yb12_out; |
||
722 | |||
723 | /* xd' = (xa-yb-xc+yd)co3 - (ya+xb-yc-xd)(si3) */ |
||
724 | pSrc[2u * i3] = Xd12_out; |
||
725 | |||
726 | /* yd' = (ya+xb-yc-xd)co3 + (xa-yb-xc+yd)(si3) */ |
||
727 | pSrc[(2u * i3) + 1u] = Yd12_out; |
||
728 | |||
729 | /* Twiddle coefficients index modifier */ |
||
730 | ia1 = ia1 + twidCoefModifier; |
||
731 | |||
732 | /* Updating input index */ |
||
733 | i0 = i0 + 1u; |
||
734 | |||
735 | } while(--j); |
||
736 | |||
737 | twidCoefModifier <<= 2u; |
||
738 | |||
739 | /* Calculation of second stage to excluding last stage */ |
||
740 | for (k = fftLen >> 2u; k > 4u; k >>= 2u) |
||
741 | { |
||
742 | /* Initializations for the first stage */ |
||
743 | n1 = n2; |
||
744 | n2 >>= 2u; |
||
745 | ia1 = 0u; |
||
746 | |||
747 | /* Calculation of first stage */ |
||
748 | j = 0; |
||
749 | do |
||
750 | { |
||
751 | /* index calculation for the coefficients */ |
||
752 | ia2 = ia1 + ia1; |
||
753 | ia3 = ia2 + ia1; |
||
754 | co1 = pCoef[ia1 * 2u]; |
||
755 | si1 = pCoef[(ia1 * 2u) + 1u]; |
||
756 | co2 = pCoef[ia2 * 2u]; |
||
757 | si2 = pCoef[(ia2 * 2u) + 1u]; |
||
758 | co3 = pCoef[ia3 * 2u]; |
||
759 | si3 = pCoef[(ia3 * 2u) + 1u]; |
||
760 | |||
761 | /* Twiddle coefficients index modifier */ |
||
762 | ia1 = ia1 + twidCoefModifier; |
||
763 | |||
764 | i0 = j; |
||
765 | do |
||
766 | { |
||
767 | /* index calculation for the input as, */ |
||
768 | /* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */ |
||
769 | i1 = i0 + n2; |
||
770 | i2 = i1 + n2; |
||
771 | i3 = i2 + n2; |
||
772 | |||
773 | xaIn = pSrc[(2u * i0)]; |
||
774 | yaIn = pSrc[(2u * i0) + 1u]; |
||
775 | |||
776 | xbIn = pSrc[(2u * i1)]; |
||
777 | ybIn = pSrc[(2u * i1) + 1u]; |
||
778 | |||
779 | xcIn = pSrc[(2u * i2)]; |
||
780 | ycIn = pSrc[(2u * i2) + 1u]; |
||
781 | |||
782 | xdIn = pSrc[(2u * i3)]; |
||
783 | ydIn = pSrc[(2u * i3) + 1u]; |
||
784 | |||
785 | /* xa - xc */ |
||
786 | Xaminusc = xaIn - xcIn; |
||
787 | /* (xb - xd) */ |
||
788 | Xbminusd = xbIn - xdIn; |
||
789 | /* ya - yc */ |
||
790 | Yaminusc = yaIn - ycIn; |
||
791 | /* (yb - yd) */ |
||
792 | Ybminusd = ybIn - ydIn; |
||
793 | |||
794 | /* xa + xc */ |
||
795 | Xaplusc = xaIn + xcIn; |
||
796 | /* xb + xd */ |
||
797 | Xbplusd = xbIn + xdIn; |
||
798 | /* ya + yc */ |
||
799 | Yaplusc = yaIn + ycIn; |
||
800 | /* yb + yd */ |
||
801 | Ybplusd = ybIn + ydIn; |
||
802 | |||
803 | /* (xa - xc) - (yb - yd) */ |
||
804 | Xb12C_out = (Xaminusc - Ybminusd); |
||
805 | /* (ya - yc) + (xb - xd) */ |
||
806 | Yb12C_out = (Yaminusc + Xbminusd); |
||
807 | /* xa + xc -(xb + xd) */ |
||
808 | Xc12C_out = (Xaplusc - Xbplusd); |
||
809 | /* (ya + yc) - (yb + yd) */ |
||
810 | Yc12C_out = (Yaplusc - Ybplusd); |
||
811 | /* (xa - xc) + (yb - yd) */ |
||
812 | Xd12C_out = (Xaminusc + Ybminusd); |
||
813 | /* (ya - yc) - (xb - xd) */ |
||
814 | Yd12C_out = (Yaminusc - Xbminusd); |
||
815 | |||
816 | pSrc[(2u * i0)] = Xaplusc + Xbplusd; |
||
817 | pSrc[(2u * i0) + 1u] = Yaplusc + Ybplusd; |
||
818 | |||
819 | Xb12_out = Xb12C_out * co1; |
||
820 | Yb12_out = Yb12C_out * co1; |
||
821 | Xc12_out = Xc12C_out * co2; |
||
822 | Yc12_out = Yc12C_out * co2; |
||
823 | Xd12_out = Xd12C_out * co3; |
||
824 | Yd12_out = Yd12C_out * co3; |
||
825 | |||
826 | /* xb' = (xa+yb-xc-yd)co1 - (ya-xb-yc+xd)(si1) */ |
||
827 | //Xb12_out -= Yb12C_out * si1; |
||
828 | p0 = Yb12C_out * si1; |
||
829 | /* yb' = (ya-xb-yc+xd)co1 + (xa+yb-xc-yd)(si1) */ |
||
830 | //Yb12_out += Xb12C_out * si1; |
||
831 | p1 = Xb12C_out * si1; |
||
832 | /* xc' = (xa-xb+xc-xd)co2 - (ya-yb+yc-yd)(si2) */ |
||
833 | //Xc12_out -= Yc12C_out * si2; |
||
834 | p2 = Yc12C_out * si2; |
||
835 | /* yc' = (ya-yb+yc-yd)co2 + (xa-xb+xc-xd)(si2) */ |
||
836 | //Yc12_out += Xc12C_out * si2; |
||
837 | p3 = Xc12C_out * si2; |
||
838 | /* xd' = (xa-yb-xc+yd)co3 - (ya+xb-yc-xd)(si3) */ |
||
839 | //Xd12_out -= Yd12C_out * si3; |
||
840 | p4 = Yd12C_out * si3; |
||
841 | /* yd' = (ya+xb-yc-xd)co3 + (xa-yb-xc+yd)(si3) */ |
||
842 | //Yd12_out += Xd12C_out * si3; |
||
843 | p5 = Xd12C_out * si3; |
||
844 | |||
845 | Xb12_out -= p0; |
||
846 | Yb12_out += p1; |
||
847 | Xc12_out -= p2; |
||
848 | Yc12_out += p3; |
||
849 | Xd12_out -= p4; |
||
850 | Yd12_out += p5; |
||
851 | |||
852 | /* xc' = (xa-xb+xc-xd)co2 - (ya-yb+yc-yd)(si2) */ |
||
853 | pSrc[2u * i1] = Xc12_out; |
||
854 | |||
855 | /* yc' = (ya-yb+yc-yd)co2 + (xa-xb+xc-xd)(si2) */ |
||
856 | pSrc[(2u * i1) + 1u] = Yc12_out; |
||
857 | |||
858 | /* xb' = (xa+yb-xc-yd)co1 - (ya-xb-yc+xd)(si1) */ |
||
859 | pSrc[2u * i2] = Xb12_out; |
||
860 | |||
861 | /* yb' = (ya-xb-yc+xd)co1 + (xa+yb-xc-yd)(si1) */ |
||
862 | pSrc[(2u * i2) + 1u] = Yb12_out; |
||
863 | |||
864 | /* xd' = (xa-yb-xc+yd)co3 - (ya+xb-yc-xd)(si3) */ |
||
865 | pSrc[2u * i3] = Xd12_out; |
||
866 | |||
867 | /* yd' = (ya+xb-yc-xd)co3 + (xa-yb-xc+yd)(si3) */ |
||
868 | pSrc[(2u * i3) + 1u] = Yd12_out; |
||
869 | |||
870 | i0 += n1; |
||
871 | } while(i0 < fftLen); |
||
872 | j++; |
||
873 | } while(j <= (n2 - 1u)); |
||
874 | twidCoefModifier <<= 2u; |
||
875 | } |
||
876 | /* Initializations of last stage */ |
||
877 | |||
878 | j = fftLen >> 2; |
||
879 | ptr1 = &pSrc[0]; |
||
880 | |||
881 | /* Calculations of last stage */ |
||
882 | do |
||
883 | { |
||
884 | xaIn = ptr1[0]; |
||
885 | yaIn = ptr1[1]; |
||
886 | xbIn = ptr1[2]; |
||
887 | ybIn = ptr1[3]; |
||
888 | xcIn = ptr1[4]; |
||
889 | ycIn = ptr1[5]; |
||
890 | xdIn = ptr1[6]; |
||
891 | ydIn = ptr1[7]; |
||
892 | |||
893 | /* Butterfly implementation */ |
||
894 | /* xa + xc */ |
||
895 | Xaplusc = xaIn + xcIn; |
||
896 | |||
897 | /* xa - xc */ |
||
898 | Xaminusc = xaIn - xcIn; |
||
899 | |||
900 | /* ya + yc */ |
||
901 | Yaplusc = yaIn + ycIn; |
||
902 | |||
903 | /* ya - yc */ |
||
904 | Yaminusc = yaIn - ycIn; |
||
905 | |||
906 | /* xb + xd */ |
||
907 | Xbplusd = xbIn + xdIn; |
||
908 | |||
909 | /* yb + yd */ |
||
910 | Ybplusd = ybIn + ydIn; |
||
911 | |||
912 | /* (xb-xd) */ |
||
913 | Xbminusd = xbIn - xdIn; |
||
914 | |||
915 | /* (yb-yd) */ |
||
916 | Ybminusd = ybIn - ydIn; |
||
917 | |||
918 | /* xa' = (xa+xb+xc+xd) * onebyfftLen */ |
||
919 | a0 = (Xaplusc + Xbplusd); |
||
920 | /* ya' = (ya+yb+yc+yd) * onebyfftLen */ |
||
921 | a1 = (Yaplusc + Ybplusd); |
||
922 | /* xc' = (xa-xb+xc-xd) * onebyfftLen */ |
||
923 | a2 = (Xaplusc - Xbplusd); |
||
924 | /* yc' = (ya-yb+yc-yd) * onebyfftLen */ |
||
925 | a3 = (Yaplusc - Ybplusd); |
||
926 | /* xb' = (xa-yb-xc+yd) * onebyfftLen */ |
||
927 | a4 = (Xaminusc - Ybminusd); |
||
928 | /* yb' = (ya+xb-yc-xd) * onebyfftLen */ |
||
929 | a5 = (Yaminusc + Xbminusd); |
||
930 | /* xd' = (xa-yb-xc+yd) * onebyfftLen */ |
||
931 | a6 = (Xaminusc + Ybminusd); |
||
932 | /* yd' = (ya-xb-yc+xd) * onebyfftLen */ |
||
933 | a7 = (Yaminusc - Xbminusd); |
||
934 | |||
935 | p0 = a0 * onebyfftLen; |
||
936 | p1 = a1 * onebyfftLen; |
||
937 | p2 = a2 * onebyfftLen; |
||
938 | p3 = a3 * onebyfftLen; |
||
939 | p4 = a4 * onebyfftLen; |
||
940 | p5 = a5 * onebyfftLen; |
||
941 | p6 = a6 * onebyfftLen; |
||
942 | p7 = a7 * onebyfftLen; |
||
943 | |||
944 | /* xa' = (xa+xb+xc+xd) * onebyfftLen */ |
||
945 | ptr1[0] = p0; |
||
946 | /* ya' = (ya+yb+yc+yd) * onebyfftLen */ |
||
947 | ptr1[1] = p1; |
||
948 | /* xc' = (xa-xb+xc-xd) * onebyfftLen */ |
||
949 | ptr1[2] = p2; |
||
950 | /* yc' = (ya-yb+yc-yd) * onebyfftLen */ |
||
951 | ptr1[3] = p3; |
||
952 | /* xb' = (xa-yb-xc+yd) * onebyfftLen */ |
||
953 | ptr1[4] = p4; |
||
954 | /* yb' = (ya+xb-yc-xd) * onebyfftLen */ |
||
955 | ptr1[5] = p5; |
||
956 | /* xd' = (xa-yb-xc+yd) * onebyfftLen */ |
||
957 | ptr1[6] = p6; |
||
958 | /* yd' = (ya-xb-yc+xd) * onebyfftLen */ |
||
959 | ptr1[7] = p7; |
||
960 | |||
961 | /* increment source pointer by 8 for next calculations */ |
||
962 | ptr1 = ptr1 + 8u; |
||
963 | |||
964 | } while(--j); |
||
965 | |||
966 | #else |
||
967 | |||
968 | float32_t t1, t2, r1, r2, s1, s2; |
||
969 | |||
970 | /* Run the below code for Cortex-M0 */ |
||
971 | |||
972 | /* Initializations for the first stage */ |
||
973 | n2 = fftLen; |
||
974 | n1 = n2; |
||
975 | |||
976 | /* Calculation of first stage */ |
||
977 | for (k = fftLen; k > 4u; k >>= 2u) |
||
978 | { |
||
979 | /* Initializations for the first stage */ |
||
980 | n1 = n2; |
||
981 | n2 >>= 2u; |
||
982 | ia1 = 0u; |
||
983 | |||
984 | /* Calculation of first stage */ |
||
985 | j = 0; |
||
986 | do |
||
987 | { |
||
988 | /* index calculation for the coefficients */ |
||
989 | ia2 = ia1 + ia1; |
||
990 | ia3 = ia2 + ia1; |
||
991 | co1 = pCoef[ia1 * 2u]; |
||
992 | si1 = pCoef[(ia1 * 2u) + 1u]; |
||
993 | co2 = pCoef[ia2 * 2u]; |
||
994 | si2 = pCoef[(ia2 * 2u) + 1u]; |
||
995 | co3 = pCoef[ia3 * 2u]; |
||
996 | si3 = pCoef[(ia3 * 2u) + 1u]; |
||
997 | |||
998 | /* Twiddle coefficients index modifier */ |
||
999 | ia1 = ia1 + twidCoefModifier; |
||
1000 | |||
1001 | i0 = j; |
||
1002 | do |
||
1003 | { |
||
1004 | /* index calculation for the input as, */ |
||
1005 | /* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */ |
||
1006 | i1 = i0 + n2; |
||
1007 | i2 = i1 + n2; |
||
1008 | i3 = i2 + n2; |
||
1009 | |||
1010 | /* xa + xc */ |
||
1011 | r1 = pSrc[(2u * i0)] + pSrc[(2u * i2)]; |
||
1012 | |||
1013 | /* xa - xc */ |
||
1014 | r2 = pSrc[(2u * i0)] - pSrc[(2u * i2)]; |
||
1015 | |||
1016 | /* ya + yc */ |
||
1017 | s1 = pSrc[(2u * i0) + 1u] + pSrc[(2u * i2) + 1u]; |
||
1018 | |||
1019 | /* ya - yc */ |
||
1020 | s2 = pSrc[(2u * i0) + 1u] - pSrc[(2u * i2) + 1u]; |
||
1021 | |||
1022 | /* xb + xd */ |
||
1023 | t1 = pSrc[2u * i1] + pSrc[2u * i3]; |
||
1024 | |||
1025 | /* xa' = xa + xb + xc + xd */ |
||
1026 | pSrc[2u * i0] = r1 + t1; |
||
1027 | |||
1028 | /* xa + xc -(xb + xd) */ |
||
1029 | r1 = r1 - t1; |
||
1030 | |||
1031 | /* yb + yd */ |
||
1032 | t2 = pSrc[(2u * i1) + 1u] + pSrc[(2u * i3) + 1u]; |
||
1033 | |||
1034 | /* ya' = ya + yb + yc + yd */ |
||
1035 | pSrc[(2u * i0) + 1u] = s1 + t2; |
||
1036 | |||
1037 | /* (ya + yc) - (yb + yd) */ |
||
1038 | s1 = s1 - t2; |
||
1039 | |||
1040 | /* (yb - yd) */ |
||
1041 | t1 = pSrc[(2u * i1) + 1u] - pSrc[(2u * i3) + 1u]; |
||
1042 | |||
1043 | /* (xb - xd) */ |
||
1044 | t2 = pSrc[2u * i1] - pSrc[2u * i3]; |
||
1045 | |||
1046 | /* xc' = (xa-xb+xc-xd)co2 - (ya-yb+yc-yd)(si2) */ |
||
1047 | pSrc[2u * i1] = (r1 * co2) - (s1 * si2); |
||
1048 | |||
1049 | /* yc' = (ya-yb+yc-yd)co2 + (xa-xb+xc-xd)(si2) */ |
||
1050 | pSrc[(2u * i1) + 1u] = (s1 * co2) + (r1 * si2); |
||
1051 | |||
1052 | /* (xa - xc) - (yb - yd) */ |
||
1053 | r1 = r2 - t1; |
||
1054 | |||
1055 | /* (xa - xc) + (yb - yd) */ |
||
1056 | r2 = r2 + t1; |
||
1057 | |||
1058 | /* (ya - yc) + (xb - xd) */ |
||
1059 | s1 = s2 + t2; |
||
1060 | |||
1061 | /* (ya - yc) - (xb - xd) */ |
||
1062 | s2 = s2 - t2; |
||
1063 | |||
1064 | /* xb' = (xa+yb-xc-yd)co1 - (ya-xb-yc+xd)(si1) */ |
||
1065 | pSrc[2u * i2] = (r1 * co1) - (s1 * si1); |
||
1066 | |||
1067 | /* yb' = (ya-xb-yc+xd)co1 + (xa+yb-xc-yd)(si1) */ |
||
1068 | pSrc[(2u * i2) + 1u] = (s1 * co1) + (r1 * si1); |
||
1069 | |||
1070 | /* xd' = (xa-yb-xc+yd)co3 - (ya+xb-yc-xd)(si3) */ |
||
1071 | pSrc[2u * i3] = (r2 * co3) - (s2 * si3); |
||
1072 | |||
1073 | /* yd' = (ya+xb-yc-xd)co3 + (xa-yb-xc+yd)(si3) */ |
||
1074 | pSrc[(2u * i3) + 1u] = (s2 * co3) + (r2 * si3); |
||
1075 | |||
1076 | i0 += n1; |
||
1077 | } while( i0 < fftLen); |
||
1078 | j++; |
||
1079 | } while(j <= (n2 - 1u)); |
||
1080 | twidCoefModifier <<= 2u; |
||
1081 | } |
||
1082 | /* Initializations of last stage */ |
||
1083 | n1 = n2; |
||
1084 | n2 >>= 2u; |
||
1085 | |||
1086 | /* Calculations of last stage */ |
||
1087 | for (i0 = 0u; i0 <= (fftLen - n1); i0 += n1) |
||
1088 | { |
||
1089 | /* index calculation for the input as, */ |
||
1090 | /* pSrc[i0 + 0], pSrc[i0 + fftLen/4], pSrc[i0 + fftLen/2], pSrc[i0 + 3fftLen/4] */ |
||
1091 | i1 = i0 + n2; |
||
1092 | i2 = i1 + n2; |
||
1093 | i3 = i2 + n2; |
||
1094 | |||
1095 | /* Butterfly implementation */ |
||
1096 | /* xa + xc */ |
||
1097 | r1 = pSrc[2u * i0] + pSrc[2u * i2]; |
||
1098 | |||
1099 | /* xa - xc */ |
||
1100 | r2 = pSrc[2u * i0] - pSrc[2u * i2]; |
||
1101 | |||
1102 | /* ya + yc */ |
||
1103 | s1 = pSrc[(2u * i0) + 1u] + pSrc[(2u * i2) + 1u]; |
||
1104 | |||
1105 | /* ya - yc */ |
||
1106 | s2 = pSrc[(2u * i0) + 1u] - pSrc[(2u * i2) + 1u]; |
||
1107 | |||
1108 | /* xc + xd */ |
||
1109 | t1 = pSrc[2u * i1] + pSrc[2u * i3]; |
||
1110 | |||
1111 | /* xa' = xa + xb + xc + xd */ |
||
1112 | pSrc[2u * i0] = (r1 + t1) * onebyfftLen; |
||
1113 | |||
1114 | /* (xa + xb) - (xc + xd) */ |
||
1115 | r1 = r1 - t1; |
||
1116 | |||
1117 | /* yb + yd */ |
||
1118 | t2 = pSrc[(2u * i1) + 1u] + pSrc[(2u * i3) + 1u]; |
||
1119 | |||
1120 | /* ya' = ya + yb + yc + yd */ |
||
1121 | pSrc[(2u * i0) + 1u] = (s1 + t2) * onebyfftLen; |
||
1122 | |||
1123 | /* (ya + yc) - (yb + yd) */ |
||
1124 | s1 = s1 - t2; |
||
1125 | |||
1126 | /* (yb-yd) */ |
||
1127 | t1 = pSrc[(2u * i1) + 1u] - pSrc[(2u * i3) + 1u]; |
||
1128 | |||
1129 | /* (xb-xd) */ |
||
1130 | t2 = pSrc[2u * i1] - pSrc[2u * i3]; |
||
1131 | |||
1132 | /* xc' = (xa-xb+xc-xd)co2 - (ya-yb+yc-yd)(si2) */ |
||
1133 | pSrc[2u * i1] = r1 * onebyfftLen; |
||
1134 | |||
1135 | /* yc' = (ya-yb+yc-yd)co2 + (xa-xb+xc-xd)(si2) */ |
||
1136 | pSrc[(2u * i1) + 1u] = s1 * onebyfftLen; |
||
1137 | |||
1138 | /* (xa - xc) - (yb-yd) */ |
||
1139 | r1 = r2 - t1; |
||
1140 | |||
1141 | /* (xa - xc) + (yb-yd) */ |
||
1142 | r2 = r2 + t1; |
||
1143 | |||
1144 | /* (ya - yc) + (xb-xd) */ |
||
1145 | s1 = s2 + t2; |
||
1146 | |||
1147 | /* (ya - yc) - (xb-xd) */ |
||
1148 | s2 = s2 - t2; |
||
1149 | |||
1150 | /* xb' = (xa+yb-xc-yd)co1 - (ya-xb-yc+xd)(si1) */ |
||
1151 | pSrc[2u * i2] = r1 * onebyfftLen; |
||
1152 | |||
1153 | /* yb' = (ya-xb-yc+xd)co1 + (xa+yb-xc-yd)(si1) */ |
||
1154 | pSrc[(2u * i2) + 1u] = s1 * onebyfftLen; |
||
1155 | |||
1156 | /* xd' = (xa-yb-xc+yd)co3 - (ya+xb-yc-xd)(si3) */ |
||
1157 | pSrc[2u * i3] = r2 * onebyfftLen; |
||
1158 | |||
1159 | /* yd' = (ya+xb-yc-xd)co3 + (xa-yb-xc+yd)(si3) */ |
||
1160 | pSrc[(2u * i3) + 1u] = s2 * onebyfftLen; |
||
1161 | } |
||
1162 | |||
1163 | #endif /* #ifndef ARM_MATH_CM0_FAMILY_FAMILY */ |
||
1164 | } |
||
1165 | |||
1166 | /** |
||
1167 | * @addtogroup ComplexFFT |
||
1168 | * @{ |
||
1169 | */ |
||
1170 | |||
1171 | /** |
||
1172 | * @details |
||
1173 | * @brief Processing function for the floating-point Radix-4 CFFT/CIFFT. |
||
1174 | * @deprecated Do not use this function. It has been superseded by \ref arm_cfft_f32 and will be removed |
||
1175 | * in the future. |
||
1176 | * @param[in] *S points to an instance of the floating-point Radix-4 CFFT/CIFFT structure. |
||
1177 | * @param[in, out] *pSrc points to the complex data buffer of size <code>2*fftLen</code>. Processing occurs in-place. |
||
1178 | * @return none. |
||
1179 | */ |
||
1180 | |||
1181 | void arm_cfft_radix4_f32( |
||
1182 | const arm_cfft_radix4_instance_f32 * S, |
||
1183 | float32_t * pSrc) |
||
1184 | { |
||
1185 | |||
1186 | if(S->ifftFlag == 1u) |
||
1187 | { |
||
1188 | /* Complex IFFT radix-4 */ |
||
1189 | arm_radix4_butterfly_inverse_f32(pSrc, S->fftLen, S->pTwiddle, |
||
1190 | S->twidCoefModifier, S->onebyfftLen); |
||
1191 | } |
||
1192 | else |
||
1193 | { |
||
1194 | /* Complex FFT radix-4 */ |
||
1195 | arm_radix4_butterfly_f32(pSrc, S->fftLen, S->pTwiddle, |
||
1196 | S->twidCoefModifier); |
||
1197 | } |
||
1198 | |||
1199 | if(S->bitReverseFlag == 1u) |
||
1200 | { |
||
1201 | /* Bit Reversal */ |
||
1202 | arm_bitreversal_f32(pSrc, S->fftLen, S->bitRevFactor, S->pBitRevTable); |
||
1203 | } |
||
1204 | |||
1205 | } |
||
1206 | |||
1207 | /** |
||
1208 | * @} end of ComplexFFT group |
||
1209 | */ |
||
1210 |