Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/dft/simd/common/n2fv_14.c @ 10:37bf6b4a2645

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW3
author Chris Cannam
date Wed, 20 Mar 2013 15:35:50 +0000
parents
children
comparison
equal deleted inserted replaced
9:c0fb53affa76 10:37bf6b4a2645
1 /*
2 * Copyright (c) 2003, 2007-11 Matteo Frigo
3 * Copyright (c) 2003, 2007-11 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 Sun Nov 25 07:37:22 EST 2012 */
23
24 #include "codelet-dft.h"
25
26 #ifdef HAVE_FMA
27
28 /* Generated by: ../../../genfft/gen_notw_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 14 -name n2fv_14 -with-ostride 2 -include 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 * 65 stack variables, 6 constants, and 35 memory accesses
34 */
35 #include "n2f.h"
36
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(KP900968867, +0.900968867902419126236102319507445051165919162);
40 DVK(KP801937735, +0.801937735804838252472204639014890102331838324);
41 DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
42 DVK(KP692021471, +0.692021471630095869627814897002069140197260599);
43 DVK(KP554958132, +0.554958132087371191422194871006410481067288862);
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 TH, T3, TP, Tn, Ta, Ts, TW, TK, TO, Tk, TM, Tg, TL, Td, T1;
53 V T2;
54 T1 = LD(&(xi[0]), ivs, &(xi[0]));
55 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
56 {
57 V Ti, TI, T6, TJ, T9, Tj, Te, Tf, Tb, Tc;
58 {
59 V T4, T5, T7, T8, Tl, Tm;
60 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
61 T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
62 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
63 T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
64 Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
65 Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
66 Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
67 TH = VADD(T1, T2);
68 T3 = VSUB(T1, T2);
69 TI = VADD(T4, T5);
70 T6 = VSUB(T4, T5);
71 TJ = VADD(T7, T8);
72 T9 = VSUB(T7, T8);
73 TP = VADD(Tl, Tm);
74 Tn = VSUB(Tl, Tm);
75 Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
76 Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
77 Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
78 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
79 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
80 }
81 Ta = VADD(T6, T9);
82 Ts = VSUB(T9, T6);
83 TW = VSUB(TJ, TI);
84 TK = VADD(TI, TJ);
85 TO = VADD(Ti, Tj);
86 Tk = VSUB(Ti, Tj);
87 TM = VADD(Te, Tf);
88 Tg = VSUB(Te, Tf);
89 TL = VADD(Tb, Tc);
90 Td = VSUB(Tb, Tc);
91 }
92 {
93 V T19, T1a, T18, TB, T13, TY, TG, Tw, T11, Tr, T16, TT, Tz, TE, TU;
94 V TQ;
95 TU = VSUB(TO, TP);
96 TQ = VADD(TO, TP);
97 {
98 V Tt, To, TV, TN;
99 Tt = VSUB(Tn, Tk);
100 To = VADD(Tk, Tn);
101 TV = VSUB(TL, TM);
102 TN = VADD(TL, TM);
103 {
104 V Tu, Th, TZ, T17;
105 Tu = VSUB(Tg, Td);
106 Th = VADD(Td, Tg);
107 TZ = VFNMS(LDK(KP356895867), TK, TQ);
108 T17 = VFNMS(LDK(KP554958132), TU, TW);
109 {
110 V Tp, TA, T14, TR;
111 Tp = VFNMS(LDK(KP356895867), Ta, To);
112 TA = VFMA(LDK(KP554958132), Tt, Ts);
113 T19 = VADD(TH, VADD(TK, VADD(TN, TQ)));
114 STM2(&(xo[0]), T19, ovs, &(xo[0]));
115 T14 = VFNMS(LDK(KP356895867), TN, TK);
116 TR = VFNMS(LDK(KP356895867), TQ, TN);
117 {
118 V T12, TX, Tx, TC;
119 T12 = VFMA(LDK(KP554958132), TV, TU);
120 TX = VFMA(LDK(KP554958132), TW, TV);
121 T1a = VADD(T3, VADD(Ta, VADD(Th, To)));
122 STM2(&(xo[14]), T1a, ovs, &(xo[2]));
123 Tx = VFNMS(LDK(KP356895867), Th, Ta);
124 TC = VFNMS(LDK(KP356895867), To, Th);
125 {
126 V TF, Tv, T10, Tq;
127 TF = VFNMS(LDK(KP554958132), Ts, Tu);
128 Tv = VFMA(LDK(KP554958132), Tu, Tt);
129 T10 = VFNMS(LDK(KP692021471), TZ, TN);
130 T18 = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), T17, TV));
131 Tq = VFNMS(LDK(KP692021471), Tp, Th);
132 TB = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), TA, Tu));
133 {
134 V T15, TS, Ty, TD;
135 T15 = VFNMS(LDK(KP692021471), T14, TQ);
136 TS = VFNMS(LDK(KP692021471), TR, TK);
137 T13 = VMUL(LDK(KP974927912), VFMA(LDK(KP801937735), T12, TW));
138 TY = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TX, TU));
139 Ty = VFNMS(LDK(KP692021471), Tx, To);
140 TD = VFNMS(LDK(KP692021471), TC, Ta);
141 TG = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), TF, Tt));
142 Tw = VMUL(LDK(KP974927912), VFNMS(LDK(KP801937735), Tv, Ts));
143 T11 = VFNMS(LDK(KP900968867), T10, TH);
144 Tr = VFNMS(LDK(KP900968867), Tq, T3);
145 T16 = VFNMS(LDK(KP900968867), T15, TH);
146 TT = VFNMS(LDK(KP900968867), TS, TH);
147 Tz = VFNMS(LDK(KP900968867), Ty, T3);
148 TE = VFNMS(LDK(KP900968867), TD, T3);
149 }
150 }
151 }
152 }
153 }
154 }
155 {
156 V T1b, T1c, T1d, T1e;
157 T1b = VFNMSI(T13, T11);
158 STM2(&(xo[24]), T1b, ovs, &(xo[0]));
159 T1c = VFMAI(T13, T11);
160 STM2(&(xo[4]), T1c, ovs, &(xo[0]));
161 T1d = VFMAI(Tw, Tr);
162 STM2(&(xo[18]), T1d, ovs, &(xo[2]));
163 T1e = VFNMSI(Tw, Tr);
164 STM2(&(xo[10]), T1e, ovs, &(xo[2]));
165 {
166 V T1f, T1g, T1h, T1i;
167 T1f = VFNMSI(T18, T16);
168 STM2(&(xo[16]), T1f, ovs, &(xo[0]));
169 STN2(&(xo[16]), T1f, T1d, ovs);
170 T1g = VFMAI(T18, T16);
171 STM2(&(xo[12]), T1g, ovs, &(xo[0]));
172 STN2(&(xo[12]), T1g, T1a, ovs);
173 T1h = VFNMSI(TY, TT);
174 STM2(&(xo[20]), T1h, ovs, &(xo[0]));
175 T1i = VFMAI(TY, TT);
176 STM2(&(xo[8]), T1i, ovs, &(xo[0]));
177 STN2(&(xo[8]), T1i, T1e, ovs);
178 {
179 V T1j, T1k, T1l, T1m;
180 T1j = VFMAI(TB, Tz);
181 STM2(&(xo[2]), T1j, ovs, &(xo[2]));
182 STN2(&(xo[0]), T19, T1j, ovs);
183 T1k = VFNMSI(TB, Tz);
184 STM2(&(xo[26]), T1k, ovs, &(xo[2]));
185 STN2(&(xo[24]), T1b, T1k, ovs);
186 T1l = VFMAI(TG, TE);
187 STM2(&(xo[6]), T1l, ovs, &(xo[2]));
188 STN2(&(xo[4]), T1c, T1l, ovs);
189 T1m = VFNMSI(TG, TE);
190 STM2(&(xo[22]), T1m, ovs, &(xo[2]));
191 STN2(&(xo[20]), T1h, T1m, ovs);
192 }
193 }
194 }
195 }
196 }
197 }
198 VLEAVE();
199 }
200
201 static const kdft_desc desc = { 14, XSIMD_STRING("n2fv_14"), {32, 6, 42, 0}, &GENUS, 0, 2, 0, 0 };
202
203 void XSIMD(codelet_n2fv_14) (planner *p) {
204 X(kdft_register) (p, n2fv_14, &desc);
205 }
206
207 #else /* HAVE_FMA */
208
209 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 14 -name n2fv_14 -with-ostride 2 -include n2f.h -store-multiple 2 */
210
211 /*
212 * This function contains 74 FP additions, 36 FP multiplications,
213 * (or, 50 additions, 12 multiplications, 24 fused multiply/add),
214 * 39 stack variables, 6 constants, and 35 memory accesses
215 */
216 #include "n2f.h"
217
218 static void n2fv_14(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
219 {
220 DVK(KP222520933, +0.222520933956314404288902564496794759466355569);
221 DVK(KP900968867, +0.900968867902419126236102319507445051165919162);
222 DVK(KP623489801, +0.623489801858733530525004884004239810632274731);
223 DVK(KP433883739, +0.433883739117558120475768332848358754609990728);
224 DVK(KP781831482, +0.781831482468029808708444526674057750232334519);
225 DVK(KP974927912, +0.974927912181823607018131682993931217232785801);
226 {
227 INT i;
228 const R *xi;
229 R *xo;
230 xi = ri;
231 xo = ro;
232 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)) {
233 V T3, Ty, To, TK, Tr, TE, Ta, TJ, Tq, TB, Th, TL, Ts, TH, T1;
234 V T2;
235 T1 = LD(&(xi[0]), ivs, &(xi[0]));
236 T2 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
237 T3 = VSUB(T1, T2);
238 Ty = VADD(T1, T2);
239 {
240 V Tk, TC, Tn, TD;
241 {
242 V Ti, Tj, Tl, Tm;
243 Ti = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
244 Tj = LD(&(xi[WS(is, 13)]), ivs, &(xi[WS(is, 1)]));
245 Tk = VSUB(Ti, Tj);
246 TC = VADD(Ti, Tj);
247 Tl = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
248 Tm = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
249 Tn = VSUB(Tl, Tm);
250 TD = VADD(Tl, Tm);
251 }
252 To = VADD(Tk, Tn);
253 TK = VSUB(TC, TD);
254 Tr = VSUB(Tn, Tk);
255 TE = VADD(TC, TD);
256 }
257 {
258 V T6, Tz, T9, TA;
259 {
260 V T4, T5, T7, T8;
261 T4 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
262 T5 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
263 T6 = VSUB(T4, T5);
264 Tz = VADD(T4, T5);
265 T7 = LD(&(xi[WS(is, 12)]), ivs, &(xi[0]));
266 T8 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
267 T9 = VSUB(T7, T8);
268 TA = VADD(T7, T8);
269 }
270 Ta = VADD(T6, T9);
271 TJ = VSUB(TA, Tz);
272 Tq = VSUB(T9, T6);
273 TB = VADD(Tz, TA);
274 }
275 {
276 V Td, TF, Tg, TG;
277 {
278 V Tb, Tc, Te, Tf;
279 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
280 Tc = LD(&(xi[WS(is, 11)]), ivs, &(xi[WS(is, 1)]));
281 Td = VSUB(Tb, Tc);
282 TF = VADD(Tb, Tc);
283 Te = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
284 Tf = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
285 Tg = VSUB(Te, Tf);
286 TG = VADD(Te, Tf);
287 }
288 Th = VADD(Td, Tg);
289 TL = VSUB(TF, TG);
290 Ts = VSUB(Tg, Td);
291 TH = VADD(TF, TG);
292 }
293 {
294 V TR, TS, TT, TU, TV, TW;
295 TR = VADD(T3, VADD(Ta, VADD(Th, To)));
296 STM2(&(xo[14]), TR, ovs, &(xo[2]));
297 TS = VADD(Ty, VADD(TB, VADD(TH, TE)));
298 STM2(&(xo[0]), TS, ovs, &(xo[0]));
299 {
300 V Tt, Tp, TP, TQ;
301 Tt = VBYI(VFNMS(LDK(KP781831482), Tr, VFNMS(LDK(KP433883739), Ts, VMUL(LDK(KP974927912), Tq))));
302 Tp = VFMA(LDK(KP623489801), To, VFNMS(LDK(KP900968867), Th, VFNMS(LDK(KP222520933), Ta, T3)));
303 TT = VSUB(Tp, Tt);
304 STM2(&(xo[10]), TT, ovs, &(xo[2]));
305 TU = VADD(Tp, Tt);
306 STM2(&(xo[18]), TU, ovs, &(xo[2]));
307 TP = VBYI(VFMA(LDK(KP974927912), TJ, VFMA(LDK(KP433883739), TL, VMUL(LDK(KP781831482), TK))));
308 TQ = VFMA(LDK(KP623489801), TE, VFNMS(LDK(KP900968867), TH, VFNMS(LDK(KP222520933), TB, Ty)));
309 TV = VADD(TP, TQ);
310 STM2(&(xo[4]), TV, ovs, &(xo[0]));
311 TW = VSUB(TQ, TP);
312 STM2(&(xo[24]), TW, ovs, &(xo[0]));
313 }
314 {
315 V Tv, Tu, TX, TY;
316 Tv = VBYI(VFMA(LDK(KP781831482), Tq, VFMA(LDK(KP974927912), Ts, VMUL(LDK(KP433883739), Tr))));
317 Tu = VFMA(LDK(KP623489801), Ta, VFNMS(LDK(KP900968867), To, VFNMS(LDK(KP222520933), Th, T3)));
318 TX = VSUB(Tu, Tv);
319 STM2(&(xo[26]), TX, ovs, &(xo[2]));
320 STN2(&(xo[24]), TW, TX, ovs);
321 TY = VADD(Tu, Tv);
322 STM2(&(xo[2]), TY, ovs, &(xo[2]));
323 STN2(&(xo[0]), TS, TY, ovs);
324 }
325 {
326 V TM, TI, TZ, T10;
327 TM = VBYI(VFNMS(LDK(KP433883739), TK, VFNMS(LDK(KP974927912), TL, VMUL(LDK(KP781831482), TJ))));
328 TI = VFMA(LDK(KP623489801), TB, VFNMS(LDK(KP900968867), TE, VFNMS(LDK(KP222520933), TH, Ty)));
329 TZ = VSUB(TI, TM);
330 STM2(&(xo[12]), TZ, ovs, &(xo[0]));
331 STN2(&(xo[12]), TZ, TR, ovs);
332 T10 = VADD(TM, TI);
333 STM2(&(xo[16]), T10, ovs, &(xo[0]));
334 STN2(&(xo[16]), T10, TU, ovs);
335 }
336 {
337 V T12, TO, TN, T11;
338 TO = VBYI(VFMA(LDK(KP433883739), TJ, VFNMS(LDK(KP974927912), TK, VMUL(LDK(KP781831482), TL))));
339 TN = VFMA(LDK(KP623489801), TH, VFNMS(LDK(KP222520933), TE, VFNMS(LDK(KP900968867), TB, Ty)));
340 T11 = VSUB(TN, TO);
341 STM2(&(xo[8]), T11, ovs, &(xo[0]));
342 STN2(&(xo[8]), T11, TT, ovs);
343 T12 = VADD(TO, TN);
344 STM2(&(xo[20]), T12, ovs, &(xo[0]));
345 {
346 V Tx, Tw, T13, T14;
347 Tx = VBYI(VFMA(LDK(KP433883739), Tq, VFNMS(LDK(KP781831482), Ts, VMUL(LDK(KP974927912), Tr))));
348 Tw = VFMA(LDK(KP623489801), Th, VFNMS(LDK(KP222520933), To, VFNMS(LDK(KP900968867), Ta, T3)));
349 T13 = VSUB(Tw, Tx);
350 STM2(&(xo[22]), T13, ovs, &(xo[2]));
351 STN2(&(xo[20]), T12, T13, ovs);
352 T14 = VADD(Tw, Tx);
353 STM2(&(xo[6]), T14, ovs, &(xo[2]));
354 STN2(&(xo[4]), TV, T14, ovs);
355 }
356 }
357 }
358 }
359 }
360 VLEAVE();
361 }
362
363 static const kdft_desc desc = { 14, XSIMD_STRING("n2fv_14"), {50, 12, 24, 0}, &GENUS, 0, 2, 0, 0 };
364
365 void XSIMD(codelet_n2fv_14) (planner *p) {
366 X(kdft_register) (p, n2fv_14, &desc);
367 }
368
369 #endif /* HAVE_FMA */