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 07:04:53 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 14 -name n2fv_14 -with-ostride 2 -include dft/simd/n2f.h -store-multiple 2 */
29
30 /*
31 * This function contains 74 FP additions, 48 FP multiplications,
32 * (or, 32 additions, 6 multiplications, 42 fused multiply/add),
33 * 51 stack variables, 6 constants, and 35 memory accesses
34 */
35 #include "dft/simd/n2f.h"
36
n2fv_14(const R * ri,const R * ii,R * ro,R * io,stride is,stride os,INT v,INT ivs,INT ovs)37 static void n2fv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 DVK(KP801937735, +0.801937735804838252472204639014890102331838324);
40 DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
41 DVK(KP554958132, +0.554958132087371191422194871006410481067288862);
42 DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
43 DVK(KP692021471, +0.692021471630095869627814897002069140197260599);
44 DVK(KP356895867, +0.356895867892209443894399510021300583399127187);
45 {
46 INT i;
47 const R *xi;
48 R *xo;
49 xi = ri;
50 xo = ro;
51 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(28, is), MAKE_VOLATILE_STRIDE(28, os)) {
52 V T3, TH, Ts, TV, TW, Tt, Tu, TU, Ta, To, Th, Tp, TC, Tx, TK;
53 V TQ, TN, TR, T14, TZ, T1, T2;
54 T1 = LD(&(xi[0]), ivs, &(xi[0]));
55 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
56 T3 = VSUB(T1, T2);
57 TH = VADD(T1, T2);
58 {
59 V T6, TI, T9, TJ, Tn, TP, Tk, TO, Tg, TM, Td, TL;
60 {
61 V T4, T5, Ti, Tj;
62 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
63 T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
64 T6 = VSUB(T4, T5);
65 TI = VADD(T4, T5);
66 {
67 V T7, T8, Tl, Tm;
68 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
69 T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
70 T9 = VSUB(T7, T8);
71 TJ = VADD(T7, T8);
72 Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
73 Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
74 Tn = VSUB(Tl, Tm);
75 TP = VADD(Tl, Tm);
76 }
77 Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
78 Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
79 Tk = VSUB(Ti, Tj);
80 TO = VADD(Ti, Tj);
81 {
82 V Te, Tf, Tb, Tc;
83 Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
84 Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
85 Tg = VSUB(Te, Tf);
86 TM = VADD(Te, Tf);
87 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
88 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
89 Td = VSUB(Tb, Tc);
90 TL = VADD(Tb, Tc);
91 }
92 }
93 Ts = VSUB(T9, T6);
94 TV = VSUB(TL, TM);
95 TW = VSUB(TJ, TI);
96 Tt = VSUB(Tn, Tk);
97 Tu = VSUB(Tg, Td);
98 TU = VSUB(TO, TP);
99 Ta = VADD(T6, T9);
100 To = VADD(Tk, Tn);
101 Th = VADD(Td, Tg);
102 Tp = VFNMS(LDK(KP356895867), Ta, To);
103 TC = VFNMS(LDK(KP356895867), To, Th);
104 Tx = VFNMS(LDK(KP356895867), Th, Ta);
105 TK = VADD(TI, TJ);
106 TQ = VADD(TO, TP);
107 TN = VADD(TL, TM);
108 TR = VFNMS(LDK(KP356895867), TQ, TN);
109 T14 = VFNMS(LDK(KP356895867), TN, TK);
110 TZ = VFNMS(LDK(KP356895867), TK, TQ);
111 }
112 {
113 V T1a, T1b, T19, T1c, T1f, T1i, T1j;
114 T19 = VADD(T3, VADD(Ta, VADD(Th, To)));
115 STM2(&(xo[14]), T19, ovs, &(xo[2]));
116 T1a = VADD(TH, VADD(TK, VADD(TN, TQ)));
117 STM2(&(xo[0]), T1a, ovs, &(xo[0]));
118 {
119 V Tr, Tw, Tq, Tv;
120 Tq = VFNMS(LDK(KP692021471), Tp, Th);
121 Tr = VFNMS(LDK(KP900968867), Tq, T3);
122 Tv = VFMA(LDK(KP554958132), Tu, Tt);
123 Tw = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tv, Ts));
124 T1b = VFNMSI(Tw, Tr);
125 STM2(&(xo[10]), T1b, ovs, &(xo[2]));
126 T1c = VFMAI(Tw, Tr);
127 STM2(&(xo[18]), T1c, ovs, &(xo[2]));
128 }
129 {
130 V T16, T18, T15, T17, T1d, T1e;
131 T15 = VFNMS(LDK(KP692021471), T14, TQ);
132 T16 = VFNMS(LDK(KP900968867), T15, TH);
133 T17 = VFNMS(LDK(KP554958132), TU, TW);
134 T18 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T17, TV));
135 T1d = VFMAI(T18, T16);
136 STM2(&(xo[12]), T1d, ovs, &(xo[0]));
137 STN2(&(xo[12]), T1d, T19, ovs);
138 T1e = VFNMSI(T18, T16);
139 STM2(&(xo[16]), T1e, ovs, &(xo[0]));
140 STN2(&(xo[16]), T1e, T1c, ovs);
141 }
142 {
143 V Tz, TB, Ty, TA, T1g;
144 Ty = VFNMS(LDK(KP692021471), Tx, To);
145 Tz = VFNMS(LDK(KP900968867), Ty, T3);
146 TA = VFMA(LDK(KP554958132), Tt, Ts);
147 TB = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TA, Tu));
148 T1f = VFNMSI(TB, Tz);
149 STM2(&(xo[26]), T1f, ovs, &(xo[2]));
150 T1g = VFMAI(TB, Tz);
151 STM2(&(xo[2]), T1g, ovs, &(xo[2]));
152 STN2(&(xo[0]), T1a, T1g, ovs);
153 }
154 {
155 V TT, TY, TS, TX, T1h;
156 TS = VFNMS(LDK(KP692021471), TR, TK);
157 TT = VFNMS(LDK(KP900968867), TS, TH);
158 TX = VFMA(LDK(KP554958132), TW, TV);
159 TY = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TX, TU));
160 T1h = VFMAI(TY, TT);
161 STM2(&(xo[8]), T1h, ovs, &(xo[0]));
162 STN2(&(xo[8]), T1h, T1b, ovs);
163 T1i = VFNMSI(TY, TT);
164 STM2(&(xo[20]), T1i, ovs, &(xo[0]));
165 }
166 {
167 V T11, T13, T10, T12, T1k;
168 T10 = VFNMS(LDK(KP692021471), TZ, TN);
169 T11 = VFNMS(LDK(KP900968867), T10, TH);
170 T12 = VFMA(LDK(KP554958132), TV, TU);
171 T13 = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), T12, TW));
172 T1j = VFMAI(T13, T11);
173 STM2(&(xo[4]), T1j, ovs, &(xo[0]));
174 T1k = VFNMSI(T13, T11);
175 STM2(&(xo[24]), T1k, ovs, &(xo[0]));
176 STN2(&(xo[24]), T1k, T1f, ovs);
177 }
178 {
179 V TE, TG, TD, TF, T1l, T1m;
180 TD = VFNMS(LDK(KP692021471), TC, Ta);
181 TE = VFNMS(LDK(KP900968867), TD, T3);
182 TF = VFNMS(LDK(KP554958132), Ts, Tu);
183 TG = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TF, Tt));
184 T1l = VFNMSI(TG, TE);
185 STM2(&(xo[22]), T1l, ovs, &(xo[2]));
186 STN2(&(xo[20]), T1i, T1l, ovs);
187 T1m = VFMAI(TG, TE);
188 STM2(&(xo[6]), T1m, ovs, &(xo[2]));
189 STN2(&(xo[4]), T1j, T1m, ovs);
190 }
191 }
192 }
193 }
194 VLEAVE();
195 }
196
197 static const kdft_desc desc = { 14, XSIMD_STRING("n2fv_14"), { 32, 6, 42, 0 }, &GENUS, 0, 2, 0, 0 };
198
XSIMD(codelet_n2fv_14)199 void XSIMD(codelet_n2fv_14) (planner *p) { X(kdft_register) (p, n2fv_14, &desc);
200 }
201
202 #else
203
204 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 14 -name n2fv_14 -with-ostride 2 -include dft/simd/n2f.h -store-multiple 2 */
205
206 /*
207 * This function contains 74 FP additions, 36 FP multiplications,
208 * (or, 50 additions, 12 multiplications, 24 fused multiply/add),
209 * 39 stack variables, 6 constants, and 35 memory accesses
210 */
211 #include "dft/simd/n2f.h"
212
n2fv_14(const R * ri,const R * ii,R * ro,R * io,stride is,stride os,INT v,INT ivs,INT ovs)213 static void n2fv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
214 {
215 DVK(KP222520933, +0.222520933956314404288902564496794759466355569);
216 DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
217 DVK(KP623489801, +0.623489801858733530525004884004239810632274731);
218 DVK(KP433883739, +0.433883739117558120475768332848358754609990728);
219 DVK(KP781831482, +0.781831482468029808708444526674057750232334519);
220 DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
221 {
222 INT i;
223 const R *xi;
224 R *xo;
225 xi = ri;
226 xo = ro;
227 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(28, is), MAKE_VOLATILE_STRIDE(28, os)) {
228 V T3, Ty, To, TK, Tr, TE, Ta, TJ, Tq, TB, Th, TL, Ts, TH, T1;
229 V T2;
230 T1 = LD(&(xi[0]), ivs, &(xi[0]));
231 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
232 T3 = VSUB(T1, T2);
233 Ty = VADD(T1, T2);
234 {
235 V Tk, TC, Tn, TD;
236 {
237 V Ti, Tj, Tl, Tm;
238 Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
239 Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
240 Tk = VSUB(Ti, Tj);
241 TC = VADD(Ti, Tj);
242 Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
243 Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
244 Tn = VSUB(Tl, Tm);
245 TD = VADD(Tl, Tm);
246 }
247 To = VADD(Tk, Tn);
248 TK = VSUB(TC, TD);
249 Tr = VSUB(Tn, Tk);
250 TE = VADD(TC, TD);
251 }
252 {
253 V T6, Tz, T9, TA;
254 {
255 V T4, T5, T7, T8;
256 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
257 T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
258 T6 = VSUB(T4, T5);
259 Tz = VADD(T4, T5);
260 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
261 T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
262 T9 = VSUB(T7, T8);
263 TA = VADD(T7, T8);
264 }
265 Ta = VADD(T6, T9);
266 TJ = VSUB(TA, Tz);
267 Tq = VSUB(T9, T6);
268 TB = VADD(Tz, TA);
269 }
270 {
271 V Td, TF, Tg, TG;
272 {
273 V Tb, Tc, Te, Tf;
274 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
275 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
276 Td = VSUB(Tb, Tc);
277 TF = VADD(Tb, Tc);
278 Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
279 Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
280 Tg = VSUB(Te, Tf);
281 TG = VADD(Te, Tf);
282 }
283 Th = VADD(Td, Tg);
284 TL = VSUB(TF, TG);
285 Ts = VSUB(Tg, Td);
286 TH = VADD(TF, TG);
287 }
288 {
289 V TR, TS, TT, TU, TV, TW;
290 TR = VADD(T3, VADD(Ta, VADD(Th, To)));
291 STM2(&(xo[14]), TR, ovs, &(xo[2]));
292 TS = VADD(Ty, VADD(TB, VADD(TH, TE)));
293 STM2(&(xo[0]), TS, ovs, &(xo[0]));
294 {
295 V Tt, Tp, TP, TQ;
296 Tt = VBYI(VFNMS(LDK(KP781831482), Tr, VFNMS(LDK(KP433883739), Ts, VMUL(LDK(KP974927912), Tq))));
297 Tp = VFMA(LDK(KP623489801), To, VFNMS(LDK(KP900968867), Th, VFNMS(LDK(KP222520933), Ta, T3)));
298 TT = VSUB(Tp, Tt);
299 STM2(&(xo[10]), TT, ovs, &(xo[2]));
300 TU = VADD(Tp, Tt);
301 STM2(&(xo[18]), TU, ovs, &(xo[2]));
302 TP = VBYI(VFMA(LDK(KP974927912), TJ, VFMA(LDK(KP433883739), TL, VMUL(LDK(KP781831482), TK))));
303 TQ = VFMA(LDK(KP623489801), TE, VFNMS(LDK(KP900968867), TH, VFNMS(LDK(KP222520933), TB, Ty)));
304 TV = VADD(TP, TQ);
305 STM2(&(xo[4]), TV, ovs, &(xo[0]));
306 TW = VSUB(TQ, TP);
307 STM2(&(xo[24]), TW, ovs, &(xo[0]));
308 }
309 {
310 V Tv, Tu, TX, TY;
311 Tv = VBYI(VFMA(LDK(KP781831482), Tq, VFMA(LDK(KP974927912), Ts, VMUL(LDK(KP433883739), Tr))));
312 Tu = VFMA(LDK(KP623489801), Ta, VFNMS(LDK(KP900968867), To, VFNMS(LDK(KP222520933), Th, T3)));
313 TX = VSUB(Tu, Tv);
314 STM2(&(xo[26]), TX, ovs, &(xo[2]));
315 STN2(&(xo[24]), TW, TX, ovs);
316 TY = VADD(Tu, Tv);
317 STM2(&(xo[2]), TY, ovs, &(xo[2]));
318 STN2(&(xo[0]), TS, TY, ovs);
319 }
320 {
321 V TM, TI, TZ, T10;
322 TM = VBYI(VFNMS(LDK(KP433883739), TK, VFNMS(LDK(KP974927912), TL, VMUL(LDK(KP781831482), TJ))));
323 TI = VFMA(LDK(KP623489801), TB, VFNMS(LDK(KP900968867), TE, VFNMS(LDK(KP222520933), TH, Ty)));
324 TZ = VSUB(TI, TM);
325 STM2(&(xo[12]), TZ, ovs, &(xo[0]));
326 STN2(&(xo[12]), TZ, TR, ovs);
327 T10 = VADD(TM, TI);
328 STM2(&(xo[16]), T10, ovs, &(xo[0]));
329 STN2(&(xo[16]), T10, TU, ovs);
330 }
331 {
332 V T12, TO, TN, T11;
333 TO = VBYI(VFMA(LDK(KP433883739), TJ, VFNMS(LDK(KP974927912), TK, VMUL(LDK(KP781831482), TL))));
334 TN = VFMA(LDK(KP623489801), TH, VFNMS(LDK(KP222520933), TE, VFNMS(LDK(KP900968867), TB, Ty)));
335 T11 = VSUB(TN, TO);
336 STM2(&(xo[8]), T11, ovs, &(xo[0]));
337 STN2(&(xo[8]), T11, TT, ovs);
338 T12 = VADD(TO, TN);
339 STM2(&(xo[20]), T12, ovs, &(xo[0]));
340 {
341 V Tx, Tw, T13, T14;
342 Tx = VBYI(VFMA(LDK(KP433883739), Tq, VFNMS(LDK(KP781831482), Ts, VMUL(LDK(KP974927912), Tr))));
343 Tw = VFMA(LDK(KP623489801), Th, VFNMS(LDK(KP222520933), To, VFNMS(LDK(KP900968867), Ta, T3)));
344 T13 = VSUB(Tw, Tx);
345 STM2(&(xo[22]), T13, ovs, &(xo[2]));
346 STN2(&(xo[20]), T12, T13, ovs);
347 T14 = VADD(Tw, Tx);
348 STM2(&(xo[6]), T14, ovs, &(xo[2]));
349 STN2(&(xo[4]), TV, T14, ovs);
350 }
351 }
352 }
353 }
354 }
355 VLEAVE();
356 }
357
358 static const kdft_desc desc = { 14, XSIMD_STRING("n2fv_14"), { 50, 12, 24, 0 }, &GENUS, 0, 2, 0, 0 };
359
XSIMD(codelet_n2fv_14)360 void XSIMD(codelet_n2fv_14) (planner *p) { X(kdft_register) (p, n2fv_14, &desc);
361 }
362
363 #endif
364