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 Thu Dec 10 07:04:41 EST 2020 */
23
24 #include "dft/codelet-dft.h"
25
26 #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
27
28 /* Generated by: ../../../genfft/gen_notw_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 13 -name n1fv_13 -include dft/simd/n1f.h */
29
30 /*
31 * This function contains 88 FP additions, 63 FP multiplications,
32 * (or, 31 additions, 6 multiplications, 57 fused multiply/add),
33 * 63 stack variables, 23 constants, and 26 memory accesses
34 */
35 #include "dft/simd/n1f.h"
36
n1fv_13(const R * ri,const R * ii,R * ro,R * io,stride is,stride os,INT v,INT ivs,INT ovs)37 static void n1fv_13(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 DVK(KP904176221, +0.904176221990848204433795481776887926501523162);
40 DVK(KP575140729, +0.575140729474003121368385547455453388461001608);
41 DVK(KP957805992, +0.957805992594665126462521754605754580515587217);
42 DVK(KP600477271, +0.600477271932665282925769253334763009352012849);
43 DVK(KP516520780, +0.516520780623489722840901288569017135705033622);
44 DVK(KP581704778, +0.581704778510515730456870384989698884939833902);
45 DVK(KP300462606, +0.300462606288665774426601772289207995520941381);
46 DVK(KP503537032, +0.503537032863766627246873853868466977093348562);
47 DVK(KP251768516, +0.251768516431883313623436926934233488546674281);
48 DVK(KP301479260, +0.301479260047709873958013540496673347309208464);
49 DVK(KP083333333, +0.083333333333333333333333333333333333333333333);
50 DVK(KP859542535, +0.859542535098774820163672132761689612766401925);
51 DVK(KP514918778, +0.514918778086315755491789696138117261566051239);
52 DVK(KP522026385, +0.522026385161275033714027226654165028300441940);
53 DVK(KP853480001, +0.853480001859823990758994934970528322872359049);
54 DVK(KP612264650, +0.612264650376756543746494474777125408779395514);
55 DVK(KP038632954, +0.038632954644348171955506895830342264440241080);
56 DVK(KP302775637, +0.302775637731994646559610633735247973125648287);
57 DVK(KP769338817, +0.769338817572980603471413688209101117038278899);
58 DVK(KP686558370, +0.686558370781754340655719594850823015421401653);
59 DVK(KP226109445, +0.226109445035782405468510155372505010481906348);
60 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
61 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
62 {
63 INT i;
64 const R *xi;
65 R *xo;
66 xi = ri;
67 xo = ro;
68 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(26, is), MAKE_VOLATILE_STRIDE(26, os)) {
69 V T1, TX, TY, To, TH, TR, TU, TB, TE, Tw, TF, TM, TT;
70 T1 = LD(&(xi[0]), ivs, &(xi[0]));
71 {
72 V Tf, TN, Tb, Ty, Tq, T6, Tx, Tr, Ti, Tt, Tl, Tu, Tm, TO, Td;
73 V Te, Tc, Tn;
74 Td = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
75 Te = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
76 Tf = VADD(Td, Te);
77 TN = VSUB(Td, Te);
78 {
79 V T7, T8, T9, Ta;
80 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
81 T8 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
82 T9 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
83 Ta = VADD(T8, T9);
84 Tb = VADD(T7, Ta);
85 Ty = VFMS(LDK(KP500000000), Ta, T7);
86 Tq = VSUB(T8, T9);
87 }
88 {
89 V T2, T3, T4, T5;
90 T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
91 T3 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
92 T4 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
93 T5 = VADD(T3, T4);
94 T6 = VADD(T2, T5);
95 Tx = VFNMS(LDK(KP500000000), T5, T2);
96 Tr = VSUB(T4, T3);
97 }
98 {
99 V Tg, Th, Tj, Tk;
100 Tg = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
101 Th = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
102 Ti = VADD(Tg, Th);
103 Tt = VSUB(Tg, Th);
104 Tj = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
105 Tk = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
106 Tl = VADD(Tj, Tk);
107 Tu = VSUB(Tj, Tk);
108 }
109 Tm = VADD(Ti, Tl);
110 TO = VADD(Tt, Tu);
111 TX = VSUB(T6, Tb);
112 TY = VADD(TN, TO);
113 Tc = VADD(T6, Tb);
114 Tn = VADD(Tf, Tm);
115 To = VADD(Tc, Tn);
116 TH = VSUB(Tc, Tn);
117 {
118 V TP, TQ, Tz, TA;
119 TP = VFNMS(LDK(KP500000000), TO, TN);
120 TQ = VADD(Tr, Tq);
121 TR = VFMA(LDK(KP866025403), TQ, TP);
122 TU = VFNMS(LDK(KP866025403), TQ, TP);
123 Tz = VSUB(Tx, Ty);
124 TA = VFNMS(LDK(KP500000000), Tm, Tf);
125 TB = VADD(Tz, TA);
126 TE = VSUB(Tz, TA);
127 }
128 {
129 V Ts, Tv, TK, TL;
130 Ts = VSUB(Tq, Tr);
131 Tv = VSUB(Tt, Tu);
132 Tw = VADD(Ts, Tv);
133 TF = VSUB(Ts, Tv);
134 TK = VADD(Tx, Ty);
135 TL = VSUB(Ti, Tl);
136 TM = VFMA(LDK(KP866025403), TL, TK);
137 TT = VFNMS(LDK(KP866025403), TL, TK);
138 }
139 }
140 ST(&(xo[0]), VADD(T1, To), ovs, &(xo[0]));
141 {
142 V T1c, T1k, T15, T14, T1e, T1n, TZ, TW, T1f, T1m, TD, T1j, TI, T19, TS;
143 V TV;
144 {
145 V T1a, T1b, T12, T13;
146 T1a = VFNMS(LDK(KP226109445), Tw, TB);
147 T1b = VFMA(LDK(KP686558370), TE, TF);
148 T1c = VFNMS(LDK(KP769338817), T1b, T1a);
149 T1k = VFMA(LDK(KP769338817), T1b, T1a);
150 T15 = VFNMS(LDK(KP302775637), TX, TY);
151 T12 = VFMA(LDK(KP038632954), TM, TR);
152 T13 = VFMA(LDK(KP612264650), TT, TU);
153 T14 = VFNMS(LDK(KP853480001), T13, T12);
154 T1e = VFNMS(LDK(KP522026385), T14, T15);
155 T1n = VFMA(LDK(KP853480001), T13, T12);
156 }
157 TZ = VFMA(LDK(KP302775637), TY, TX);
158 TS = VFNMS(LDK(KP038632954), TR, TM);
159 TV = VFNMS(LDK(KP612264650), TU, TT);
160 TW = VFNMS(LDK(KP853480001), TV, TS);
161 T1f = VFMA(LDK(KP853480001), TV, TS);
162 T1m = VFNMS(LDK(KP522026385), TW, TZ);
163 {
164 V TG, T18, Tp, TC, T17;
165 TG = VFNMS(LDK(KP514918778), TF, TE);
166 T18 = VFNMS(LDK(KP859542535), TG, TH);
167 Tp = VFNMS(LDK(KP083333333), To, T1);
168 TC = VFMA(LDK(KP301479260), TB, Tw);
169 T17 = VFNMS(LDK(KP251768516), TC, Tp);
170 TD = VFMA(LDK(KP503537032), TC, Tp);
171 T1j = VFNMS(LDK(KP300462606), T18, T17);
172 TI = VFMA(LDK(KP581704778), TH, TG);
173 T19 = VFMA(LDK(KP300462606), T18, T17);
174 }
175 {
176 V TJ, T10, T1l, T1o;
177 TJ = VFNMS(LDK(KP516520780), TI, TD);
178 T10 = VMUL(LDK(KP600477271), VFMA(LDK(KP957805992), TZ, TW));
179 ST(&(xo[WS(os, 5)]), VFNMSI(T10, TJ), ovs, &(xo[WS(os, 1)]));
180 ST(&(xo[WS(os, 8)]), VFMAI(T10, TJ), ovs, &(xo[0]));
181 {
182 V T11, T16, T1p, T1q;
183 T11 = VFMA(LDK(KP516520780), TI, TD);
184 T16 = VMUL(LDK(KP600477271), VFMA(LDK(KP957805992), T15, T14));
185 ST(&(xo[WS(os, 1)]), VFMAI(T16, T11), ovs, &(xo[WS(os, 1)]));
186 ST(&(xo[WS(os, 12)]), VFNMSI(T16, T11), ovs, &(xo[0]));
187 T1p = VFMA(LDK(KP503537032), T1k, T1j);
188 T1q = VMUL(LDK(KP575140729), VFMA(LDK(KP904176221), T1n, T1m));
189 ST(&(xo[WS(os, 2)]), VFNMSI(T1q, T1p), ovs, &(xo[0]));
190 ST(&(xo[WS(os, 11)]), VFMAI(T1q, T1p), ovs, &(xo[WS(os, 1)]));
191 }
192 T1l = VFNMS(LDK(KP503537032), T1k, T1j);
193 T1o = VMUL(LDK(KP575140729), VFNMS(LDK(KP904176221), T1n, T1m));
194 ST(&(xo[WS(os, 6)]), VFNMSI(T1o, T1l), ovs, &(xo[0]));
195 ST(&(xo[WS(os, 7)]), VFMAI(T1o, T1l), ovs, &(xo[WS(os, 1)]));
196 {
197 V T1h, T1i, T1d, T1g;
198 T1h = VFMA(LDK(KP503537032), T1c, T19);
199 T1i = VMUL(LDK(KP575140729), VFNMS(LDK(KP904176221), T1f, T1e));
200 ST(&(xo[WS(os, 3)]), VFMAI(T1i, T1h), ovs, &(xo[WS(os, 1)]));
201 ST(&(xo[WS(os, 10)]), VFNMSI(T1i, T1h), ovs, &(xo[0]));
202 T1d = VFNMS(LDK(KP503537032), T1c, T19);
203 T1g = VMUL(LDK(KP575140729), VFMA(LDK(KP904176221), T1f, T1e));
204 ST(&(xo[WS(os, 4)]), VFNMSI(T1g, T1d), ovs, &(xo[0]));
205 ST(&(xo[WS(os, 9)]), VFMAI(T1g, T1d), ovs, &(xo[WS(os, 1)]));
206 }
207 }
208 }
209 }
210 }
211 VLEAVE();
212 }
213
214 static const kdft_desc desc = { 13, XSIMD_STRING("n1fv_13"), { 31, 6, 57, 0 }, &GENUS, 0, 0, 0, 0 };
215
XSIMD(codelet_n1fv_13)216 void XSIMD(codelet_n1fv_13) (planner *p) { X(kdft_register) (p, n1fv_13, &desc);
217 }
218
219 #else
220
221 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 13 -name n1fv_13 -include dft/simd/n1f.h */
222
223 /*
224 * This function contains 88 FP additions, 34 FP multiplications,
225 * (or, 69 additions, 15 multiplications, 19 fused multiply/add),
226 * 60 stack variables, 20 constants, and 26 memory accesses
227 */
228 #include "dft/simd/n1f.h"
229
n1fv_13(const R * ri,const R * ii,R * ro,R * io,stride is,stride os,INT v,INT ivs,INT ovs)230 static void n1fv_13(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
231 {
232 DVK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
233 DVK(KP083333333, +0.083333333333333333333333333333333333333333333);
234 DVK(KP075902986, +0.075902986037193865983102897245103540356428373);
235 DVK(KP251768516, +0.251768516431883313623436926934233488546674281);
236 DVK(KP132983124, +0.132983124607418643793760531921092974399165133);
237 DVK(KP258260390, +0.258260390311744861420450644284508567852516811);
238 DVK(KP1_732050807, +1.732050807568877293527446341505872366942805254);
239 DVK(KP300238635, +0.300238635966332641462884626667381504676006424);
240 DVK(KP011599105, +0.011599105605768290721655456654083252189827041);
241 DVK(KP156891391, +0.156891391051584611046832726756003269660212636);
242 DVK(KP256247671, +0.256247671582936600958684654061725059144125175);
243 DVK(KP174138601, +0.174138601152135905005660794929264742616964676);
244 DVK(KP575140729, +0.575140729474003121368385547455453388461001608);
245 DVK(KP503537032, +0.503537032863766627246873853868466977093348562);
246 DVK(KP113854479, +0.113854479055790798974654345867655310534642560);
247 DVK(KP265966249, +0.265966249214837287587521063842185948798330267);
248 DVK(KP387390585, +0.387390585467617292130675966426762851778775217);
249 DVK(KP300462606, +0.300462606288665774426601772289207995520941381);
250 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
251 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
252 {
253 INT i;
254 const R *xi;
255 R *xo;
256 xi = ri;
257 xo = ro;
258 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(26, is), MAKE_VOLATILE_STRIDE(26, os)) {
259 V TW, Tb, Tm, Tu, TC, TR, TX, TK, TU, Tz, TB, TN, TT;
260 TW = LD(&(xi[0]), ivs, &(xi[0]));
261 {
262 V T3, TH, Tl, Tw, Tp, Tg, Tv, To, T6, Tr, T9, Ts, Ta, TI, T1;
263 V T2, Tq, Tt;
264 T1 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
265 T2 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
266 T3 = VSUB(T1, T2);
267 TH = VADD(T1, T2);
268 {
269 V Th, Ti, Tj, Tk;
270 Th = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
271 Ti = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
272 Tj = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
273 Tk = VADD(Ti, Tj);
274 Tl = VADD(Th, Tk);
275 Tw = VSUB(Ti, Tj);
276 Tp = VFNMS(LDK(KP500000000), Tk, Th);
277 }
278 {
279 V Tc, Td, Te, Tf;
280 Tc = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
281 Td = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
282 Te = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
283 Tf = VADD(Td, Te);
284 Tg = VADD(Tc, Tf);
285 Tv = VSUB(Td, Te);
286 To = VFNMS(LDK(KP500000000), Tf, Tc);
287 }
288 {
289 V T4, T5, T7, T8;
290 T4 = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
291 T5 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
292 T6 = VSUB(T4, T5);
293 Tr = VADD(T4, T5);
294 T7 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
295 T8 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
296 T9 = VSUB(T7, T8);
297 Ts = VADD(T7, T8);
298 }
299 Ta = VADD(T6, T9);
300 TI = VADD(Tr, Ts);
301 Tb = VADD(T3, Ta);
302 Tm = VSUB(Tg, Tl);
303 Tq = VSUB(To, Tp);
304 Tt = VMUL(LDK(KP866025403), VSUB(Tr, Ts));
305 Tu = VADD(Tq, Tt);
306 TC = VSUB(Tq, Tt);
307 {
308 V TP, TQ, TG, TJ;
309 TP = VADD(Tg, Tl);
310 TQ = VADD(TH, TI);
311 TR = VMUL(LDK(KP300462606), VSUB(TP, TQ));
312 TX = VADD(TP, TQ);
313 TG = VADD(To, Tp);
314 TJ = VFNMS(LDK(KP500000000), TI, TH);
315 TK = VSUB(TG, TJ);
316 TU = VADD(TG, TJ);
317 }
318 {
319 V Tx, Ty, TL, TM;
320 Tx = VMUL(LDK(KP866025403), VSUB(Tv, Tw));
321 Ty = VFNMS(LDK(KP500000000), Ta, T3);
322 Tz = VSUB(Tx, Ty);
323 TB = VADD(Tx, Ty);
324 TL = VADD(Tv, Tw);
325 TM = VSUB(T6, T9);
326 TN = VSUB(TL, TM);
327 TT = VADD(TL, TM);
328 }
329 }
330 ST(&(xo[0]), VADD(TW, TX), ovs, &(xo[0]));
331 {
332 V T19, T1n, T14, T13, T1f, T1k, Tn, TE, T1e, T1j, TS, T1m, TZ, T1c, TA;
333 V TD;
334 {
335 V T17, T18, T11, T12;
336 T17 = VFMA(LDK(KP387390585), TN, VMUL(LDK(KP265966249), TK));
337 T18 = VFNMS(LDK(KP503537032), TU, VMUL(LDK(KP113854479), TT));
338 T19 = VSUB(T17, T18);
339 T1n = VADD(T17, T18);
340 T14 = VFMA(LDK(KP575140729), Tm, VMUL(LDK(KP174138601), Tb));
341 T11 = VFNMS(LDK(KP156891391), TB, VMUL(LDK(KP256247671), TC));
342 T12 = VFMA(LDK(KP011599105), Tz, VMUL(LDK(KP300238635), Tu));
343 T13 = VSUB(T11, T12);
344 T1f = VADD(T14, T13);
345 T1k = VMUL(LDK(KP1_732050807), VADD(T11, T12));
346 }
347 Tn = VFNMS(LDK(KP174138601), Tm, VMUL(LDK(KP575140729), Tb));
348 TA = VFNMS(LDK(KP300238635), Tz, VMUL(LDK(KP011599105), Tu));
349 TD = VFMA(LDK(KP256247671), TB, VMUL(LDK(KP156891391), TC));
350 TE = VSUB(TA, TD);
351 T1e = VMUL(LDK(KP1_732050807), VADD(TD, TA));
352 T1j = VSUB(Tn, TE);
353 {
354 V TO, T1b, TV, TY, T1a;
355 TO = VFNMS(LDK(KP132983124), TN, VMUL(LDK(KP258260390), TK));
356 T1b = VSUB(TR, TO);
357 TV = VFMA(LDK(KP251768516), TT, VMUL(LDK(KP075902986), TU));
358 TY = VFNMS(LDK(KP083333333), TX, TW);
359 T1a = VSUB(TY, TV);
360 TS = VFMA(LDK(KP2_000000000), TO, TR);
361 T1m = VADD(T1b, T1a);
362 TZ = VFMA(LDK(KP2_000000000), TV, TY);
363 T1c = VSUB(T1a, T1b);
364 }
365 {
366 V TF, T10, T1l, T1o;
367 TF = VBYI(VFMA(LDK(KP2_000000000), TE, Tn));
368 T10 = VADD(TS, TZ);
369 ST(&(xo[WS(os, 1)]), VADD(TF, T10), ovs, &(xo[WS(os, 1)]));
370 ST(&(xo[WS(os, 12)]), VSUB(T10, TF), ovs, &(xo[0]));
371 {
372 V T15, T16, T1p, T1q;
373 T15 = VBYI(VFMS(LDK(KP2_000000000), T13, T14));
374 T16 = VSUB(TZ, TS);
375 ST(&(xo[WS(os, 5)]), VADD(T15, T16), ovs, &(xo[WS(os, 1)]));
376 ST(&(xo[WS(os, 8)]), VSUB(T16, T15), ovs, &(xo[0]));
377 T1p = VADD(T1n, T1m);
378 T1q = VBYI(VADD(T1j, T1k));
379 ST(&(xo[WS(os, 4)]), VSUB(T1p, T1q), ovs, &(xo[0]));
380 ST(&(xo[WS(os, 9)]), VADD(T1q, T1p), ovs, &(xo[WS(os, 1)]));
381 }
382 T1l = VBYI(VSUB(T1j, T1k));
383 T1o = VSUB(T1m, T1n);
384 ST(&(xo[WS(os, 3)]), VADD(T1l, T1o), ovs, &(xo[WS(os, 1)]));
385 ST(&(xo[WS(os, 10)]), VSUB(T1o, T1l), ovs, &(xo[0]));
386 {
387 V T1h, T1i, T1d, T1g;
388 T1h = VBYI(VSUB(T1e, T1f));
389 T1i = VSUB(T1c, T19);
390 ST(&(xo[WS(os, 6)]), VADD(T1h, T1i), ovs, &(xo[0]));
391 ST(&(xo[WS(os, 7)]), VSUB(T1i, T1h), ovs, &(xo[WS(os, 1)]));
392 T1d = VADD(T19, T1c);
393 T1g = VBYI(VADD(T1e, T1f));
394 ST(&(xo[WS(os, 2)]), VSUB(T1d, T1g), ovs, &(xo[0]));
395 ST(&(xo[WS(os, 11)]), VADD(T1g, T1d), ovs, &(xo[WS(os, 1)]));
396 }
397 }
398 }
399 }
400 }
401 VLEAVE();
402 }
403
404 static const kdft_desc desc = { 13, XSIMD_STRING("n1fv_13"), { 69, 15, 19, 0 }, &GENUS, 0, 0, 0, 0 };
405
XSIMD(codelet_n1fv_13)406 void XSIMD(codelet_n1fv_13) (planner *p) { X(kdft_register) (p, n1fv_13, &desc);
407 }
408
409 #endif
410