/src/fftw3/rdft/scalar/r2cf/r2cfII_16.c
Line | Count | Source |
1 | | /* |
2 | | * Copyright (c) 2003, 2007-14 Matteo Frigo |
3 | | * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology |
4 | | * |
5 | | * This program is free software; you can redistribute it and/or modify |
6 | | * it under the terms of the GNU General Public License as published by |
7 | | * the Free Software Foundation; either version 2 of the License, or |
8 | | * (at your option) any later version. |
9 | | * |
10 | | * This program is distributed in the hope that it will be useful, |
11 | | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
12 | | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
13 | | * GNU General Public License for more details. |
14 | | * |
15 | | * You should have received a copy of the GNU General Public License |
16 | | * along with this program; if not, write to the Free Software |
17 | | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
18 | | * |
19 | | */ |
20 | | |
21 | | /* This file was automatically generated --- DO NOT EDIT */ |
22 | | /* Generated on Sat Mar 7 06:51:13 UTC 2026 */ |
23 | | |
24 | | #include "rdft/codelet-rdft.h" |
25 | | |
26 | | #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA) |
27 | | |
28 | | /* Generated by: ../../../genfft/gen_r2cf.native -fma -compact -variables 4 -pipeline-latency 4 -n 16 -name r2cfII_16 -dft-II -include rdft/scalar/r2cfII.h */ |
29 | | |
30 | | /* |
31 | | * This function contains 66 FP additions, 48 FP multiplications, |
32 | | * (or, 18 additions, 0 multiplications, 48 fused multiply/add), |
33 | | * 32 stack variables, 7 constants, and 32 memory accesses |
34 | | */ |
35 | | #include "rdft/scalar/r2cfII.h" |
36 | | |
37 | | static void r2cfII_16(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) |
38 | | { |
39 | | DK(KP980785280, +0.980785280403230449126182236134239036973933731); |
40 | | DK(KP198912367, +0.198912367379658006911597622644676228597850501); |
41 | | DK(KP831469612, +0.831469612302545237078788377617905756738560812); |
42 | | DK(KP668178637, +0.668178637919298919997757686523080761552472251); |
43 | | DK(KP923879532, +0.923879532511286756128183189396788286822416626); |
44 | | DK(KP414213562, +0.414213562373095048801688724209698078569671875); |
45 | | DK(KP707106781, +0.707106781186547524400844362104849039284835938); |
46 | | { |
47 | | INT i; |
48 | | for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(64, rs), MAKE_VOLATILE_STRIDE(64, csr), MAKE_VOLATILE_STRIDE(64, csi)) { |
49 | | E T5, TZ, TB, TT, Tr, TK, Tu, TJ, Ti, TH, Tl, TG, Tc, T10, TE; |
50 | | E TU; |
51 | | { |
52 | | E T1, TR, T4, TS, T2, T3; |
53 | | T1 = R0[0]; |
54 | | TR = R0[WS(rs, 4)]; |
55 | | T2 = R0[WS(rs, 2)]; |
56 | | T3 = R0[WS(rs, 6)]; |
57 | | T4 = T2 - T3; |
58 | | TS = T2 + T3; |
59 | | T5 = FNMS(KP707106781, T4, T1); |
60 | | TZ = FNMS(KP707106781, TS, TR); |
61 | | TB = FMA(KP707106781, T4, T1); |
62 | | TT = FMA(KP707106781, TS, TR); |
63 | | } |
64 | | { |
65 | | E Tn, Ts, Tq, Tt, To, Tp; |
66 | | Tn = R1[WS(rs, 7)]; |
67 | | Ts = R1[WS(rs, 3)]; |
68 | | To = R1[WS(rs, 1)]; |
69 | | Tp = R1[WS(rs, 5)]; |
70 | | Tq = To - Tp; |
71 | | Tt = To + Tp; |
72 | | Tr = FMA(KP707106781, Tq, Tn); |
73 | | TK = FMA(KP707106781, Tt, Ts); |
74 | | Tu = FNMS(KP707106781, Tt, Ts); |
75 | | TJ = FMS(KP707106781, Tq, Tn); |
76 | | } |
77 | | { |
78 | | E Te, Tj, Th, Tk, Tf, Tg; |
79 | | Te = R1[0]; |
80 | | Tj = R1[WS(rs, 4)]; |
81 | | Tf = R1[WS(rs, 2)]; |
82 | | Tg = R1[WS(rs, 6)]; |
83 | | Th = Tf - Tg; |
84 | | Tk = Tf + Tg; |
85 | | Ti = FNMS(KP707106781, Th, Te); |
86 | | TH = FMA(KP707106781, Tk, Tj); |
87 | | Tl = FNMS(KP707106781, Tk, Tj); |
88 | | TG = FMA(KP707106781, Th, Te); |
89 | | } |
90 | | { |
91 | | E T8, TC, Tb, TD; |
92 | | { |
93 | | E T6, T7, T9, Ta; |
94 | | T6 = R0[WS(rs, 5)]; |
95 | | T7 = R0[WS(rs, 1)]; |
96 | | T8 = FMA(KP414213562, T7, T6); |
97 | | TC = FNMS(KP414213562, T6, T7); |
98 | | T9 = R0[WS(rs, 3)]; |
99 | | Ta = R0[WS(rs, 7)]; |
100 | | Tb = FMA(KP414213562, Ta, T9); |
101 | | TD = FMS(KP414213562, T9, Ta); |
102 | | } |
103 | | Tc = T8 - Tb; |
104 | | T10 = TD - TC; |
105 | | TE = TC + TD; |
106 | | TU = T8 + Tb; |
107 | | } |
108 | | { |
109 | | E Td, T13, Tw, T14, Tm, Tv; |
110 | | Td = FMA(KP923879532, Tc, T5); |
111 | | T13 = FNMS(KP923879532, T10, TZ); |
112 | | Tm = FMA(KP668178637, Tl, Ti); |
113 | | Tv = FMA(KP668178637, Tu, Tr); |
114 | | Tw = Tm - Tv; |
115 | | T14 = Tm + Tv; |
116 | | Cr[WS(csr, 6)] = FNMS(KP831469612, Tw, Td); |
117 | | Ci[WS(csi, 5)] = FNMS(KP831469612, T14, T13); |
118 | | Cr[WS(csr, 1)] = FMA(KP831469612, Tw, Td); |
119 | | Ci[WS(csi, 2)] = -(FMA(KP831469612, T14, T13)); |
120 | | } |
121 | | { |
122 | | E Tx, T11, TA, T12, Ty, Tz; |
123 | | Tx = FNMS(KP923879532, Tc, T5); |
124 | | T11 = FMA(KP923879532, T10, TZ); |
125 | | Ty = FNMS(KP668178637, Tr, Tu); |
126 | | Tz = FNMS(KP668178637, Ti, Tl); |
127 | | TA = Ty - Tz; |
128 | | T12 = Tz + Ty; |
129 | | Cr[WS(csr, 5)] = FNMS(KP831469612, TA, Tx); |
130 | | Ci[WS(csi, 1)] = FMA(KP831469612, T12, T11); |
131 | | Cr[WS(csr, 2)] = FMA(KP831469612, TA, Tx); |
132 | | Ci[WS(csi, 6)] = FMS(KP831469612, T12, T11); |
133 | | } |
134 | | { |
135 | | E TF, TX, TM, TY, TI, TL; |
136 | | TF = FMA(KP923879532, TE, TB); |
137 | | TX = FNMS(KP923879532, TU, TT); |
138 | | TI = FNMS(KP198912367, TH, TG); |
139 | | TL = FMA(KP198912367, TK, TJ); |
140 | | TM = TI + TL; |
141 | | TY = TL - TI; |
142 | | Cr[WS(csr, 7)] = FNMS(KP980785280, TM, TF); |
143 | | Ci[WS(csi, 3)] = FMA(KP980785280, TY, TX); |
144 | | Cr[0] = FMA(KP980785280, TM, TF); |
145 | | Ci[WS(csi, 4)] = FMS(KP980785280, TY, TX); |
146 | | } |
147 | | { |
148 | | E TN, TV, TQ, TW, TO, TP; |
149 | | TN = FNMS(KP923879532, TE, TB); |
150 | | TV = FMA(KP923879532, TU, TT); |
151 | | TO = FMA(KP198912367, TG, TH); |
152 | | TP = FNMS(KP198912367, TJ, TK); |
153 | | TQ = TO - TP; |
154 | | TW = TO + TP; |
155 | | Cr[WS(csr, 4)] = FNMS(KP980785280, TQ, TN); |
156 | | Ci[WS(csi, 7)] = FNMS(KP980785280, TW, TV); |
157 | | Cr[WS(csr, 3)] = FMA(KP980785280, TQ, TN); |
158 | | Ci[0] = -(FMA(KP980785280, TW, TV)); |
159 | | } |
160 | | } |
161 | | } |
162 | | } |
163 | | |
164 | | static const kr2c_desc desc = { 16, "r2cfII_16", { 18, 0, 48, 0 }, &GENUS }; |
165 | | |
166 | | void X(codelet_r2cfII_16) (planner *p) { X(kr2c_register) (p, r2cfII_16, &desc); |
167 | | } |
168 | | |
169 | | #else |
170 | | |
171 | | /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 16 -name r2cfII_16 -dft-II -include rdft/scalar/r2cfII.h */ |
172 | | |
173 | | /* |
174 | | * This function contains 66 FP additions, 30 FP multiplications, |
175 | | * (or, 54 additions, 18 multiplications, 12 fused multiply/add), |
176 | | * 32 stack variables, 7 constants, and 32 memory accesses |
177 | | */ |
178 | | #include "rdft/scalar/r2cfII.h" |
179 | | |
180 | | static void r2cfII_16(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) |
181 | 0 | { |
182 | 0 | DK(KP555570233, +0.555570233019602224742830813948532874374937191); |
183 | 0 | DK(KP831469612, +0.831469612302545237078788377617905756738560812); |
184 | 0 | DK(KP980785280, +0.980785280403230449126182236134239036973933731); |
185 | 0 | DK(KP195090322, +0.195090322016128267848284868477022240927691618); |
186 | 0 | DK(KP382683432, +0.382683432365089771728459984030398866761344562); |
187 | 0 | DK(KP923879532, +0.923879532511286756128183189396788286822416626); |
188 | 0 | DK(KP707106781, +0.707106781186547524400844362104849039284835938); |
189 | 0 | { |
190 | 0 | INT i; |
191 | 0 | for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(64, rs), MAKE_VOLATILE_STRIDE(64, csr), MAKE_VOLATILE_STRIDE(64, csi)) { |
192 | 0 | E T5, T11, TB, TV, Tr, TK, Tu, TJ, Ti, TH, Tl, TG, Tc, T10, TE; |
193 | 0 | E TS; |
194 | 0 | { |
195 | 0 | E T1, TU, T4, TT, T2, T3; |
196 | 0 | T1 = R0[0]; |
197 | 0 | TU = R0[WS(rs, 4)]; |
198 | 0 | T2 = R0[WS(rs, 2)]; |
199 | 0 | T3 = R0[WS(rs, 6)]; |
200 | 0 | T4 = KP707106781 * (T2 - T3); |
201 | 0 | TT = KP707106781 * (T2 + T3); |
202 | 0 | T5 = T1 + T4; |
203 | 0 | T11 = TU - TT; |
204 | 0 | TB = T1 - T4; |
205 | 0 | TV = TT + TU; |
206 | 0 | } |
207 | 0 | { |
208 | 0 | E Tq, Tt, Tp, Ts, Tn, To; |
209 | 0 | Tq = R1[WS(rs, 7)]; |
210 | 0 | Tt = R1[WS(rs, 3)]; |
211 | 0 | Tn = R1[WS(rs, 1)]; |
212 | 0 | To = R1[WS(rs, 5)]; |
213 | 0 | Tp = KP707106781 * (Tn - To); |
214 | 0 | Ts = KP707106781 * (Tn + To); |
215 | 0 | Tr = Tp - Tq; |
216 | 0 | TK = Tt - Ts; |
217 | 0 | Tu = Ts + Tt; |
218 | 0 | TJ = Tp + Tq; |
219 | 0 | } |
220 | 0 | { |
221 | 0 | E Te, Tk, Th, Tj, Tf, Tg; |
222 | 0 | Te = R1[0]; |
223 | 0 | Tk = R1[WS(rs, 4)]; |
224 | 0 | Tf = R1[WS(rs, 2)]; |
225 | 0 | Tg = R1[WS(rs, 6)]; |
226 | 0 | Th = KP707106781 * (Tf - Tg); |
227 | 0 | Tj = KP707106781 * (Tf + Tg); |
228 | 0 | Ti = Te + Th; |
229 | 0 | TH = Tk - Tj; |
230 | 0 | Tl = Tj + Tk; |
231 | 0 | TG = Te - Th; |
232 | 0 | } |
233 | 0 | { |
234 | 0 | E T8, TC, Tb, TD; |
235 | 0 | { |
236 | 0 | E T6, T7, T9, Ta; |
237 | 0 | T6 = R0[WS(rs, 1)]; |
238 | 0 | T7 = R0[WS(rs, 5)]; |
239 | 0 | T8 = FNMS(KP382683432, T7, KP923879532 * T6); |
240 | 0 | TC = FMA(KP382683432, T6, KP923879532 * T7); |
241 | 0 | T9 = R0[WS(rs, 3)]; |
242 | 0 | Ta = R0[WS(rs, 7)]; |
243 | 0 | Tb = FNMS(KP923879532, Ta, KP382683432 * T9); |
244 | 0 | TD = FMA(KP923879532, T9, KP382683432 * Ta); |
245 | 0 | } |
246 | 0 | Tc = T8 + Tb; |
247 | 0 | T10 = Tb - T8; |
248 | 0 | TE = TC - TD; |
249 | 0 | TS = TC + TD; |
250 | 0 | } |
251 | 0 | { |
252 | 0 | E Td, TW, Tw, TR, Tm, Tv; |
253 | 0 | Td = T5 - Tc; |
254 | 0 | TW = TS + TV; |
255 | 0 | Tm = FMA(KP195090322, Ti, KP980785280 * Tl); |
256 | 0 | Tv = FNMS(KP980785280, Tu, KP195090322 * Tr); |
257 | 0 | Tw = Tm + Tv; |
258 | 0 | TR = Tv - Tm; |
259 | 0 | Cr[WS(csr, 4)] = Td - Tw; |
260 | 0 | Ci[WS(csi, 7)] = TR + TW; |
261 | 0 | Cr[WS(csr, 3)] = Td + Tw; |
262 | 0 | Ci[0] = TR - TW; |
263 | 0 | } |
264 | 0 | { |
265 | 0 | E Tx, TY, TA, TX, Ty, Tz; |
266 | 0 | Tx = T5 + Tc; |
267 | 0 | TY = TV - TS; |
268 | 0 | Ty = FNMS(KP195090322, Tl, KP980785280 * Ti); |
269 | 0 | Tz = FMA(KP980785280, Tr, KP195090322 * Tu); |
270 | 0 | TA = Ty + Tz; |
271 | 0 | TX = Tz - Ty; |
272 | 0 | Cr[WS(csr, 7)] = Tx - TA; |
273 | 0 | Ci[WS(csi, 3)] = TX + TY; |
274 | 0 | Cr[0] = Tx + TA; |
275 | 0 | Ci[WS(csi, 4)] = TX - TY; |
276 | 0 | } |
277 | 0 | { |
278 | 0 | E TF, T12, TM, TZ, TI, TL; |
279 | 0 | TF = TB + TE; |
280 | 0 | T12 = T10 - T11; |
281 | 0 | TI = FMA(KP831469612, TG, KP555570233 * TH); |
282 | 0 | TL = FMA(KP831469612, TJ, KP555570233 * TK); |
283 | 0 | TM = TI - TL; |
284 | 0 | TZ = TI + TL; |
285 | 0 | Cr[WS(csr, 6)] = TF - TM; |
286 | 0 | Ci[WS(csi, 2)] = T12 - TZ; |
287 | 0 | Cr[WS(csr, 1)] = TF + TM; |
288 | 0 | Ci[WS(csi, 5)] = -(TZ + T12); |
289 | 0 | } |
290 | 0 | { |
291 | 0 | E TN, T14, TQ, T13, TO, TP; |
292 | 0 | TN = TB - TE; |
293 | 0 | T14 = T10 + T11; |
294 | 0 | TO = FNMS(KP555570233, TJ, KP831469612 * TK); |
295 | 0 | TP = FNMS(KP555570233, TG, KP831469612 * TH); |
296 | 0 | TQ = TO - TP; |
297 | 0 | T13 = TP + TO; |
298 | 0 | Cr[WS(csr, 5)] = TN - TQ; |
299 | 0 | Ci[WS(csi, 1)] = T13 + T14; |
300 | 0 | Cr[WS(csr, 2)] = TN + TQ; |
301 | 0 | Ci[WS(csi, 6)] = T13 - T14; |
302 | 0 | } |
303 | 0 | } |
304 | 0 | } |
305 | 0 | } |
306 | | |
307 | | static const kr2c_desc desc = { 16, "r2cfII_16", { 54, 18, 12, 0 }, &GENUS }; |
308 | | |
309 | 1 | void X(codelet_r2cfII_16) (planner *p) { X(kr2c_register) (p, r2cfII_16, &desc); |
310 | 1 | } |
311 | | |
312 | | #endif |