1 /* ----------------------------------------------------------------------
2 * Project: CMSIS DSP Library
3 * Title: arm_cfft_radix8_f32.c
4 * Description: Radix-8 Decimation in Frequency CFFT & CIFFT 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
31
32 /* ----------------------------------------------------------------------
33 * Internal helper function used by the FFTs
34 * -------------------------------------------------------------------- */
35
36 /**
37 brief Core function for the floating-point CFFT butterfly process.
38 param[in,out] pSrc points to the in-place buffer of floating-point data type.
39 param[in] fftLen length of the FFT.
40 param[in] pCoef points to the twiddle coefficient buffer.
41 param[in] twidCoefModifier twiddle coefficient modifier that supports different size FFTs with the same twiddle factor table.
42 return none
43 */
44
arm_radix8_butterfly_f32(float32_t * pSrc,uint16_t fftLen,const float32_t * pCoef,uint16_t twidCoefModifier)45 void arm_radix8_butterfly_f32(
46 float32_t * pSrc,
47 uint16_t fftLen,
48 const float32_t * pCoef,
49 uint16_t twidCoefModifier)
50 {
51 uint32_t ia1, ia2, ia3, ia4, ia5, ia6, ia7;
52 uint32_t i1, i2, i3, i4, i5, i6, i7, i8;
53 uint32_t id;
54 uint32_t n1, n2, j;
55
56 float32_t r1, r2, r3, r4, r5, r6, r7, r8;
57 float32_t t1, t2;
58 float32_t s1, s2, s3, s4, s5, s6, s7, s8;
59 float32_t p1, p2, p3, p4;
60 float32_t co2, co3, co4, co5, co6, co7, co8;
61 float32_t si2, si3, si4, si5, si6, si7, si8;
62 const float32_t C81 = 0.70710678118f;
63
64 n2 = fftLen;
65
66 do
67 {
68 n1 = n2;
69 n2 = n2 >> 3;
70 i1 = 0;
71
72 do
73 {
74 i2 = i1 + n2;
75 i3 = i2 + n2;
76 i4 = i3 + n2;
77 i5 = i4 + n2;
78 i6 = i5 + n2;
79 i7 = i6 + n2;
80 i8 = i7 + n2;
81 r1 = pSrc[2 * i1] + pSrc[2 * i5];
82 r5 = pSrc[2 * i1] - pSrc[2 * i5];
83 r2 = pSrc[2 * i2] + pSrc[2 * i6];
84 r6 = pSrc[2 * i2] - pSrc[2 * i6];
85 r3 = pSrc[2 * i3] + pSrc[2 * i7];
86 r7 = pSrc[2 * i3] - pSrc[2 * i7];
87 r4 = pSrc[2 * i4] + pSrc[2 * i8];
88 r8 = pSrc[2 * i4] - pSrc[2 * i8];
89 t1 = r1 - r3;
90 r1 = r1 + r3;
91 r3 = r2 - r4;
92 r2 = r2 + r4;
93 pSrc[2 * i1] = r1 + r2;
94 pSrc[2 * i5] = r1 - r2;
95 r1 = pSrc[2 * i1 + 1] + pSrc[2 * i5 + 1];
96 s5 = pSrc[2 * i1 + 1] - pSrc[2 * i5 + 1];
97 r2 = pSrc[2 * i2 + 1] + pSrc[2 * i6 + 1];
98 s6 = pSrc[2 * i2 + 1] - pSrc[2 * i6 + 1];
99 s3 = pSrc[2 * i3 + 1] + pSrc[2 * i7 + 1];
100 s7 = pSrc[2 * i3 + 1] - pSrc[2 * i7 + 1];
101 r4 = pSrc[2 * i4 + 1] + pSrc[2 * i8 + 1];
102 s8 = pSrc[2 * i4 + 1] - pSrc[2 * i8 + 1];
103 t2 = r1 - s3;
104 r1 = r1 + s3;
105 s3 = r2 - r4;
106 r2 = r2 + r4;
107 pSrc[2 * i1 + 1] = r1 + r2;
108 pSrc[2 * i5 + 1] = r1 - r2;
109 pSrc[2 * i3] = t1 + s3;
110 pSrc[2 * i7] = t1 - s3;
111 pSrc[2 * i3 + 1] = t2 - r3;
112 pSrc[2 * i7 + 1] = t2 + r3;
113 r1 = (r6 - r8) * C81;
114 r6 = (r6 + r8) * C81;
115 r2 = (s6 - s8) * C81;
116 s6 = (s6 + s8) * C81;
117 t1 = r5 - r1;
118 r5 = r5 + r1;
119 r8 = r7 - r6;
120 r7 = r7 + r6;
121 t2 = s5 - r2;
122 s5 = s5 + r2;
123 s8 = s7 - s6;
124 s7 = s7 + s6;
125 pSrc[2 * i2] = r5 + s7;
126 pSrc[2 * i8] = r5 - s7;
127 pSrc[2 * i6] = t1 + s8;
128 pSrc[2 * i4] = t1 - s8;
129 pSrc[2 * i2 + 1] = s5 - r7;
130 pSrc[2 * i8 + 1] = s5 + r7;
131 pSrc[2 * i6 + 1] = t2 - r8;
132 pSrc[2 * i4 + 1] = t2 + r8;
133
134 i1 += n1;
135 } while (i1 < fftLen);
136
137 if (n2 < 8)
138 break;
139
140 ia1 = 0;
141 j = 1;
142
143 do
144 {
145 /* index calculation for the coefficients */
146 id = ia1 + twidCoefModifier;
147 ia1 = id;
148 ia2 = ia1 + id;
149 ia3 = ia2 + id;
150 ia4 = ia3 + id;
151 ia5 = ia4 + id;
152 ia6 = ia5 + id;
153 ia7 = ia6 + id;
154
155 co2 = pCoef[2 * ia1];
156 co3 = pCoef[2 * ia2];
157 co4 = pCoef[2 * ia3];
158 co5 = pCoef[2 * ia4];
159 co6 = pCoef[2 * ia5];
160 co7 = pCoef[2 * ia6];
161 co8 = pCoef[2 * ia7];
162 si2 = pCoef[2 * ia1 + 1];
163 si3 = pCoef[2 * ia2 + 1];
164 si4 = pCoef[2 * ia3 + 1];
165 si5 = pCoef[2 * ia4 + 1];
166 si6 = pCoef[2 * ia5 + 1];
167 si7 = pCoef[2 * ia6 + 1];
168 si8 = pCoef[2 * ia7 + 1];
169
170 i1 = j;
171
172 do
173 {
174 /* index calculation for the input */
175 i2 = i1 + n2;
176 i3 = i2 + n2;
177 i4 = i3 + n2;
178 i5 = i4 + n2;
179 i6 = i5 + n2;
180 i7 = i6 + n2;
181 i8 = i7 + n2;
182 r1 = pSrc[2 * i1] + pSrc[2 * i5];
183 r5 = pSrc[2 * i1] - pSrc[2 * i5];
184 r2 = pSrc[2 * i2] + pSrc[2 * i6];
185 r6 = pSrc[2 * i2] - pSrc[2 * i6];
186 r3 = pSrc[2 * i3] + pSrc[2 * i7];
187 r7 = pSrc[2 * i3] - pSrc[2 * i7];
188 r4 = pSrc[2 * i4] + pSrc[2 * i8];
189 r8 = pSrc[2 * i4] - pSrc[2 * i8];
190 t1 = r1 - r3;
191 r1 = r1 + r3;
192 r3 = r2 - r4;
193 r2 = r2 + r4;
194 pSrc[2 * i1] = r1 + r2;
195 r2 = r1 - r2;
196 s1 = pSrc[2 * i1 + 1] + pSrc[2 * i5 + 1];
197 s5 = pSrc[2 * i1 + 1] - pSrc[2 * i5 + 1];
198 s2 = pSrc[2 * i2 + 1] + pSrc[2 * i6 + 1];
199 s6 = pSrc[2 * i2 + 1] - pSrc[2 * i6 + 1];
200 s3 = pSrc[2 * i3 + 1] + pSrc[2 * i7 + 1];
201 s7 = pSrc[2 * i3 + 1] - pSrc[2 * i7 + 1];
202 s4 = pSrc[2 * i4 + 1] + pSrc[2 * i8 + 1];
203 s8 = pSrc[2 * i4 + 1] - pSrc[2 * i8 + 1];
204 t2 = s1 - s3;
205 s1 = s1 + s3;
206 s3 = s2 - s4;
207 s2 = s2 + s4;
208 r1 = t1 + s3;
209 t1 = t1 - s3;
210 pSrc[2 * i1 + 1] = s1 + s2;
211 s2 = s1 - s2;
212 s1 = t2 - r3;
213 t2 = t2 + r3;
214 p1 = co5 * r2;
215 p2 = si5 * s2;
216 p3 = co5 * s2;
217 p4 = si5 * r2;
218 pSrc[2 * i5] = p1 + p2;
219 pSrc[2 * i5 + 1] = p3 - p4;
220 p1 = co3 * r1;
221 p2 = si3 * s1;
222 p3 = co3 * s1;
223 p4 = si3 * r1;
224 pSrc[2 * i3] = p1 + p2;
225 pSrc[2 * i3 + 1] = p3 - p4;
226 p1 = co7 * t1;
227 p2 = si7 * t2;
228 p3 = co7 * t2;
229 p4 = si7 * t1;
230 pSrc[2 * i7] = p1 + p2;
231 pSrc[2 * i7 + 1] = p3 - p4;
232 r1 = (r6 - r8) * C81;
233 r6 = (r6 + r8) * C81;
234 s1 = (s6 - s8) * C81;
235 s6 = (s6 + s8) * C81;
236 t1 = r5 - r1;
237 r5 = r5 + r1;
238 r8 = r7 - r6;
239 r7 = r7 + r6;
240 t2 = s5 - s1;
241 s5 = s5 + s1;
242 s8 = s7 - s6;
243 s7 = s7 + s6;
244 r1 = r5 + s7;
245 r5 = r5 - s7;
246 r6 = t1 + s8;
247 t1 = t1 - s8;
248 s1 = s5 - r7;
249 s5 = s5 + r7;
250 s6 = t2 - r8;
251 t2 = t2 + r8;
252 p1 = co2 * r1;
253 p2 = si2 * s1;
254 p3 = co2 * s1;
255 p4 = si2 * r1;
256 pSrc[2 * i2] = p1 + p2;
257 pSrc[2 * i2 + 1] = p3 - p4;
258 p1 = co8 * r5;
259 p2 = si8 * s5;
260 p3 = co8 * s5;
261 p4 = si8 * r5;
262 pSrc[2 * i8] = p1 + p2;
263 pSrc[2 * i8 + 1] = p3 - p4;
264 p1 = co6 * r6;
265 p2 = si6 * s6;
266 p3 = co6 * s6;
267 p4 = si6 * r6;
268 pSrc[2 * i6] = p1 + p2;
269 pSrc[2 * i6 + 1] = p3 - p4;
270 p1 = co4 * t1;
271 p2 = si4 * t2;
272 p3 = co4 * t2;
273 p4 = si4 * t1;
274 pSrc[2 * i4] = p1 + p2;
275 pSrc[2 * i4 + 1] = p3 - p4;
276
277 i1 += n1;
278 } while (i1 < fftLen);
279
280 j++;
281 } while (j < n2);
282
283 twidCoefModifier <<= 3;
284 } while (n2 > 7);
285 }
286