1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_cfft_f64.c
4 * Description: Combined Radix Decimation in Frequency CFFT Double Precision Floating point processing function
5 *
6 * $Date: 23 April 2021
7 * $Revision: V1.9.0
8 *
9 * Target Processor: Cortex-M and Cortex-A cores
10 * -------------------------------------------------------------------- */
11 /*
12 * Copyright (C) 2010-2021 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 "dsp/transform_functions.h"
30 #include "arm_common_tables.h"
31
32
33 extern void arm_radix4_butterfly_f64(
34 float64_t * pSrc,
35 uint16_t fftLen,
36 const float64_t * pCoef,
37 uint16_t twidCoefModifier);
38
39 extern void arm_bitreversal_64(
40 uint64_t * pSrc,
41 const uint16_t bitRevLen,
42 const uint16_t * pBitRevTable);
43
44 /**
45 * @} end of ComplexFFT group
46 */
47
48 /* ----------------------------------------------------------------------
49 * Internal helper function used by the FFTs
50 * ---------------------------------------------------------------------- */
51
52 /*
53 * @brief Core function for the Double Precision floating-point CFFT butterfly process.
54 * @param[in, out] *pSrc points to the in-place buffer of F64 data type.
55 * @param[in] fftLen length of the FFT.
56 * @param[in] *pCoef points to the twiddle coefficient buffer.
57 * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
58 * @return none.
59 */
60
arm_radix4_butterfly_f64(float64_t * pSrc,uint16_t fftLen,const float64_t * pCoef,uint16_t twidCoefModifier)61 void arm_radix4_butterfly_f64(
62 float64_t * pSrc,
63 uint16_t fftLen,
64 const float64_t * pCoef,
65 uint16_t twidCoefModifier)
66 {
67
68 float64_t co1, co2, co3, si1, si2, si3;
69 uint32_t ia1, ia2, ia3;
70 uint32_t i0, i1, i2, i3;
71 uint32_t n1, n2, j, k;
72
73 float64_t t1, t2, r1, r2, s1, s2;
74
75
76 /* Initializations for the fft calculation */
77 n2 = fftLen;
78 n1 = n2;
79 for (k = fftLen; k > 1U; k >>= 2U)
80 {
81 /* Initializations for the fft calculation */
82 n1 = n2;
83 n2 >>= 2U;
84 ia1 = 0U;
85
86 /* FFT Calculation */
87 j = 0;
88 do
89 {
90 /* index calculation for the coefficients */
91 ia2 = ia1 + ia1;
92 ia3 = ia2 + ia1;
93 co1 = pCoef[ia1 * 2U];
94 si1 = pCoef[(ia1 * 2U) + 1U];
95 co2 = pCoef[ia2 * 2U];
96 si2 = pCoef[(ia2 * 2U) + 1U];
97 co3 = pCoef[ia3 * 2U];
98 si3 = pCoef[(ia3 * 2U) + 1U];
99
100 /* Twiddle coefficients index modifier */
101 ia1 = ia1 + twidCoefModifier;
102
103 i0 = j;
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 /* xa + xc */
113 r1 = pSrc[(2U * i0)] + pSrc[(2U * i2)];
114
115 /* xa - xc */
116 r2 = pSrc[(2U * i0)] - pSrc[(2U * i2)];
117
118 /* ya + yc */
119 s1 = pSrc[(2U * i0) + 1U] + pSrc[(2U * i2) + 1U];
120
121 /* ya - yc */
122 s2 = pSrc[(2U * i0) + 1U] - pSrc[(2U * i2) + 1U];
123
124 /* xb + xd */
125 t1 = pSrc[2U * i1] + pSrc[2U * i3];
126
127 /* xa' = xa + xb + xc + xd */
128 pSrc[2U * i0] = r1 + t1;
129
130 /* xa + xc -(xb + xd) */
131 r1 = r1 - t1;
132
133 /* yb + yd */
134 t2 = pSrc[(2U * i1) + 1U] + pSrc[(2U * i3) + 1U];
135
136 /* ya' = ya + yb + yc + yd */
137 pSrc[(2U * i0) + 1U] = s1 + t2;
138
139 /* (ya + yc) - (yb + yd) */
140 s1 = s1 - t2;
141
142 /* (yb - yd) */
143 t1 = pSrc[(2U * i1) + 1U] - pSrc[(2U * i3) + 1U];
144
145 /* (xb - xd) */
146 t2 = pSrc[2U * i1] - pSrc[2U * i3];
147
148 /* xc' = (xa-xb+xc-xd)co2 + (ya-yb+yc-yd)(si2) */
149 pSrc[2U * i1] = (r1 * co2) + (s1 * si2);
150
151 /* yc' = (ya-yb+yc-yd)co2 - (xa-xb+xc-xd)(si2) */
152 pSrc[(2U * i1) + 1U] = (s1 * co2) - (r1 * si2);
153
154 /* (xa - xc) + (yb - yd) */
155 r1 = r2 + t1;
156
157 /* (xa - xc) - (yb - yd) */
158 r2 = r2 - t1;
159
160 /* (ya - yc) - (xb - xd) */
161 s1 = s2 - t2;
162
163 /* (ya - yc) + (xb - xd) */
164 s2 = s2 + t2;
165
166 /* xb' = (xa+yb-xc-yd)co1 + (ya-xb-yc+xd)(si1) */
167 pSrc[2U * i2] = (r1 * co1) + (s1 * si1);
168
169 /* yb' = (ya-xb-yc+xd)co1 - (xa+yb-xc-yd)(si1) */
170 pSrc[(2U * i2) + 1U] = (s1 * co1) - (r1 * si1);
171
172 /* xd' = (xa-yb-xc+yd)co3 + (ya+xb-yc-xd)(si3) */
173 pSrc[2U * i3] = (r2 * co3) + (s2 * si3);
174
175 /* yd' = (ya+xb-yc-xd)co3 - (xa-yb-xc+yd)(si3) */
176 pSrc[(2U * i3) + 1U] = (s2 * co3) - (r2 * si3);
177
178 i0 += n1;
179 } while ( i0 < fftLen);
180 j++;
181 } while (j <= (n2 - 1U));
182 twidCoefModifier <<= 2U;
183 }
184 }
185
186 /*
187 * @brief Core function for the Double Precision floating-point CFFT butterfly process.
188 * @param[in, out] *pSrc points to the in-place buffer of F64 data type.
189 * @param[in] fftLen length of the FFT.
190 * @param[in] *pCoef points to the twiddle coefficient buffer.
191 * @param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
192 * @return none.
193 */
194
arm_cfft_radix4by2_f64(float64_t * pSrc,uint32_t fftLen,const float64_t * pCoef)195 void arm_cfft_radix4by2_f64(
196 float64_t * pSrc,
197 uint32_t fftLen,
198 const float64_t * pCoef)
199 {
200 uint32_t i, l;
201 uint32_t n2, ia;
202 float64_t xt, yt, cosVal, sinVal;
203 float64_t p0, p1,p2,p3,a0,a1;
204
205 n2 = fftLen >> 1;
206 ia = 0;
207 for (i = 0; i < n2; i++)
208 {
209 cosVal = pCoef[2*ia];
210 sinVal = pCoef[2*ia + 1];
211 ia++;
212
213 l = i + n2;
214
215 /* Butterfly implementation */
216 a0 = pSrc[2 * i] + pSrc[2 * l];
217 xt = pSrc[2 * i] - pSrc[2 * l];
218
219 yt = pSrc[2 * i + 1] - pSrc[2 * l + 1];
220 a1 = pSrc[2 * l + 1] + pSrc[2 * i + 1];
221
222 p0 = xt * cosVal;
223 p1 = yt * sinVal;
224 p2 = yt * cosVal;
225 p3 = xt * sinVal;
226
227 pSrc[2 * i] = a0;
228 pSrc[2 * i + 1] = a1;
229
230 pSrc[2 * l] = p0 + p1;
231 pSrc[2 * l + 1] = p2 - p3;
232
233 }
234
235 // first col
236 arm_radix4_butterfly_f64( pSrc, n2, (float64_t*)pCoef, 2U);
237 // second col
238 arm_radix4_butterfly_f64( pSrc + fftLen, n2, (float64_t*)pCoef, 2U);
239
240 }
241
242 /**
243 @addtogroup ComplexFFT
244 @{
245 */
246
247 /**
248 @brief Processing function for the Double Precision floating-point complex FFT.
249 @param[in] S points to an instance of the Double Precision floating-point CFFT structure
250 @param[in,out] p1 points to the complex data buffer of size <code>2*fftLen</code>. Processing occurs in-place
251 @param[in] ifftFlag flag that selects transform direction
252 - value = 0: forward transform
253 - value = 1: inverse transform
254 @param[in] bitReverseFlag flag that enables / disables bit reversal of output
255 - value = 0: disables bit reversal of output
256 - value = 1: enables bit reversal of output
257 @return none
258 */
259
arm_cfft_f64(const arm_cfft_instance_f64 * S,float64_t * p1,uint8_t ifftFlag,uint8_t bitReverseFlag)260 void arm_cfft_f64(
261 const arm_cfft_instance_f64 * S,
262 float64_t * p1,
263 uint8_t ifftFlag,
264 uint8_t bitReverseFlag)
265 {
266 uint32_t L = S->fftLen, l;
267 float64_t invL, * pSrc;
268
269 if (ifftFlag == 1U)
270 {
271 /* Conjugate input data */
272 pSrc = p1 + 1;
273 for(l=0; l<L; l++)
274 {
275 *pSrc = -*pSrc;
276 pSrc += 2;
277 }
278 }
279
280 switch (L)
281 {
282 case 16:
283 case 64:
284 case 256:
285 case 1024:
286 case 4096:
287 arm_radix4_butterfly_f64 (p1, L, (float64_t*)S->pTwiddle, 1U);
288 break;
289
290 case 32:
291 case 128:
292 case 512:
293 case 2048:
294 arm_cfft_radix4by2_f64 ( p1, L, (float64_t*)S->pTwiddle);
295 break;
296
297 }
298
299 if ( bitReverseFlag )
300 arm_bitreversal_64((uint64_t*)p1, S->bitRevLength,S->pBitRevTable);
301
302 if (ifftFlag == 1U)
303 {
304 invL = 1.0 / (float64_t)L;
305 /* Conjugate and scale output data */
306 pSrc = p1;
307 for(l=0; l<L; l++)
308 {
309 *pSrc++ *= invL ;
310 *pSrc = -(*pSrc) * invL;
311 pSrc++;
312 }
313 }
314 }
315
316 /**
317 @} end of ComplexFFT group
318 */
319
