Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/rdft/scalar/r2cb/r2cb_20.c @ 167:bd3cc4d1df30

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW 3.3.8 source, and a Linux build
author Chris Cannam <cannam@all-day-breakfast.com>
date Tue, 19 Nov 2019 14:52:55 +0000
parents
children
comparison
equal deleted inserted replaced
166:cbd6d7e562c7 167:bd3cc4d1df30
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 May 24 08:07:30 EDT 2018 */
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_r2cb.native -fma -compact -variables 4 -pipeline-latency 4 -sign 1 -n 20 -name r2cb_20 -include rdft/scalar/r2cb.h */
29
30 /*
31 * This function contains 86 FP additions, 44 FP multiplications,
32 * (or, 42 additions, 0 multiplications, 44 fused multiply/add),
33 * 50 stack variables, 5 constants, and 40 memory accesses
34 */
35 #include "rdft/scalar/r2cb.h"
36
37 static void r2cb_20(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
38 {
39 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
40 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
41 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
42 DK(KP618033988, +0.618033988749894848204586834365638117720309180);
43 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
44 {
45 INT i;
46 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(80, rs), MAKE_VOLATILE_STRIDE(80, csr), MAKE_VOLATILE_STRIDE(80, csi)) {
47 E T5, TD, Tl, Tr, TO, T1l, T1d, T10, T1k, TT, T11, T1a, Tc, Tj, Tk;
48 E Tw, TB, TC, Tm, Tn, To, TE, TF, TG;
49 {
50 E T4, Tq, T3, Tp, T1, T2;
51 T4 = Cr[WS(csr, 5)];
52 Tq = Ci[WS(csi, 5)];
53 T1 = Cr[0];
54 T2 = Cr[WS(csr, 10)];
55 T3 = T1 + T2;
56 Tp = T1 - T2;
57 T5 = FNMS(KP2_000000000, T4, T3);
58 TD = FNMS(KP2_000000000, Tq, Tp);
59 Tl = FMA(KP2_000000000, T4, T3);
60 Tr = FMA(KP2_000000000, Tq, Tp);
61 }
62 {
63 E T8, Ts, TR, T19, Tb, T18, Tv, TS, Tf, Tx, TM, T1c, Ti, T1b, TA;
64 E TN;
65 {
66 E T6, T7, TP, TQ;
67 T6 = Cr[WS(csr, 4)];
68 T7 = Cr[WS(csr, 6)];
69 T8 = T6 + T7;
70 Ts = T6 - T7;
71 TP = Ci[WS(csi, 4)];
72 TQ = Ci[WS(csi, 6)];
73 TR = TP - TQ;
74 T19 = TP + TQ;
75 }
76 {
77 E T9, Ta, Tt, Tu;
78 T9 = Cr[WS(csr, 9)];
79 Ta = Cr[WS(csr, 1)];
80 Tb = T9 + Ta;
81 T18 = T9 - Ta;
82 Tt = Ci[WS(csi, 9)];
83 Tu = Ci[WS(csi, 1)];
84 Tv = Tt + Tu;
85 TS = Tt - Tu;
86 }
87 {
88 E Td, Te, TK, TL;
89 Td = Cr[WS(csr, 8)];
90 Te = Cr[WS(csr, 2)];
91 Tf = Td + Te;
92 Tx = Td - Te;
93 TK = Ci[WS(csi, 8)];
94 TL = Ci[WS(csi, 2)];
95 TM = TK - TL;
96 T1c = TK + TL;
97 }
98 {
99 E Tg, Th, Ty, Tz;
100 Tg = Cr[WS(csr, 7)];
101 Th = Cr[WS(csr, 3)];
102 Ti = Tg + Th;
103 T1b = Tg - Th;
104 Ty = Ci[WS(csi, 7)];
105 Tz = Ci[WS(csi, 3)];
106 TA = Ty + Tz;
107 TN = Tz - Ty;
108 }
109 TO = TM - TN;
110 T1l = T19 - T18;
111 T1d = T1b + T1c;
112 T10 = TS + TR;
113 T1k = T1c - T1b;
114 TT = TR - TS;
115 T11 = TN + TM;
116 T1a = T18 + T19;
117 Tc = T8 - Tb;
118 Tj = Tf - Ti;
119 Tk = Tc + Tj;
120 Tw = Ts + Tv;
121 TB = Tx - TA;
122 TC = Tw + TB;
123 Tm = T8 + Tb;
124 Tn = Tf + Ti;
125 To = Tm + Tn;
126 TE = Ts - Tv;
127 TF = Tx + TA;
128 TG = TE + TF;
129 }
130 R0[WS(rs, 5)] = FMA(KP2_000000000, Tk, T5);
131 R1[WS(rs, 7)] = FMA(KP2_000000000, TC, Tr);
132 R1[WS(rs, 2)] = FMA(KP2_000000000, TG, TD);
133 R0[0] = FMA(KP2_000000000, To, Tl);
134 {
135 E TU, TW, TJ, TV, TH, TI;
136 TU = FNMS(KP618033988, TT, TO);
137 TW = FMA(KP618033988, TO, TT);
138 TH = FNMS(KP500000000, Tk, T5);
139 TI = Tc - Tj;
140 TJ = FNMS(KP1_118033988, TI, TH);
141 TV = FMA(KP1_118033988, TI, TH);
142 R0[WS(rs, 9)] = FNMS(KP1_902113032, TU, TJ);
143 R0[WS(rs, 7)] = FMA(KP1_902113032, TW, TV);
144 R0[WS(rs, 1)] = FMA(KP1_902113032, TU, TJ);
145 R0[WS(rs, 3)] = FNMS(KP1_902113032, TW, TV);
146 }
147 {
148 E T1e, T1g, T17, T1f, T15, T16;
149 T1e = FMA(KP618033988, T1d, T1a);
150 T1g = FNMS(KP618033988, T1a, T1d);
151 T15 = FNMS(KP500000000, TG, TD);
152 T16 = TE - TF;
153 T17 = FMA(KP1_118033988, T16, T15);
154 T1f = FNMS(KP1_118033988, T16, T15);
155 R1[0] = FNMS(KP1_902113032, T1e, T17);
156 R1[WS(rs, 8)] = FMA(KP1_902113032, T1g, T1f);
157 R1[WS(rs, 4)] = FMA(KP1_902113032, T1e, T17);
158 R1[WS(rs, 6)] = FNMS(KP1_902113032, T1g, T1f);
159 }
160 {
161 E T1m, T1o, T1j, T1n, T1h, T1i;
162 T1m = FNMS(KP618033988, T1l, T1k);
163 T1o = FMA(KP618033988, T1k, T1l);
164 T1h = FNMS(KP500000000, TC, Tr);
165 T1i = Tw - TB;
166 T1j = FNMS(KP1_118033988, T1i, T1h);
167 T1n = FMA(KP1_118033988, T1i, T1h);
168 R1[WS(rs, 1)] = FNMS(KP1_902113032, T1m, T1j);
169 R1[WS(rs, 9)] = FMA(KP1_902113032, T1o, T1n);
170 R1[WS(rs, 3)] = FMA(KP1_902113032, T1m, T1j);
171 R1[WS(rs, 5)] = FNMS(KP1_902113032, T1o, T1n);
172 }
173 {
174 E T12, T14, TZ, T13, TX, TY;
175 T12 = FMA(KP618033988, T11, T10);
176 T14 = FNMS(KP618033988, T10, T11);
177 TX = FNMS(KP500000000, To, Tl);
178 TY = Tm - Tn;
179 TZ = FMA(KP1_118033988, TY, TX);
180 T13 = FNMS(KP1_118033988, TY, TX);
181 R0[WS(rs, 8)] = FNMS(KP1_902113032, T12, TZ);
182 R0[WS(rs, 6)] = FMA(KP1_902113032, T14, T13);
183 R0[WS(rs, 2)] = FMA(KP1_902113032, T12, TZ);
184 R0[WS(rs, 4)] = FNMS(KP1_902113032, T14, T13);
185 }
186 }
187 }
188 }
189
190 static const kr2c_desc desc = { 20, "r2cb_20", {42, 0, 44, 0}, &GENUS };
191
192 void X(codelet_r2cb_20) (planner *p) {
193 X(kr2c_register) (p, r2cb_20, &desc);
194 }
195
196 #else
197
198 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 20 -name r2cb_20 -include rdft/scalar/r2cb.h */
199
200 /*
201 * This function contains 86 FP additions, 30 FP multiplications,
202 * (or, 70 additions, 14 multiplications, 16 fused multiply/add),
203 * 50 stack variables, 5 constants, and 40 memory accesses
204 */
205 #include "rdft/scalar/r2cb.h"
206
207 static void r2cb_20(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
208 {
209 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
210 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
211 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
212 DK(KP1_175570504, +1.175570504584946258337411909278145537195304875);
213 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
214 {
215 INT i;
216 for (i = v; i > 0; i = i - 1, R0 = R0 + ovs, R1 = R1 + ovs, Cr = Cr + ivs, Ci = Ci + ivs, MAKE_VOLATILE_STRIDE(80, rs), MAKE_VOLATILE_STRIDE(80, csr), MAKE_VOLATILE_STRIDE(80, csi)) {
217 E T6, TF, Tm, Tt, TQ, T1n, T1f, T12, T1m, TV, T13, T1c, Td, Tk, Tl;
218 E Ty, TD, TE, Tn, To, Tp, TG, TH, TI;
219 {
220 E T5, Ts, T3, Tq;
221 {
222 E T4, Tr, T1, T2;
223 T4 = Cr[WS(csr, 5)];
224 T5 = KP2_000000000 * T4;
225 Tr = Ci[WS(csi, 5)];
226 Ts = KP2_000000000 * Tr;
227 T1 = Cr[0];
228 T2 = Cr[WS(csr, 10)];
229 T3 = T1 + T2;
230 Tq = T1 - T2;
231 }
232 T6 = T3 - T5;
233 TF = Tq - Ts;
234 Tm = T3 + T5;
235 Tt = Tq + Ts;
236 }
237 {
238 E T9, Tu, TO, T1b, Tc, T1a, Tx, TP, Tg, Tz, TT, T1e, Tj, T1d, TC;
239 E TU;
240 {
241 E T7, T8, TM, TN;
242 T7 = Cr[WS(csr, 4)];
243 T8 = Cr[WS(csr, 6)];
244 T9 = T7 + T8;
245 Tu = T7 - T8;
246 TM = Ci[WS(csi, 4)];
247 TN = Ci[WS(csi, 6)];
248 TO = TM - TN;
249 T1b = TM + TN;
250 }
251 {
252 E Ta, Tb, Tv, Tw;
253 Ta = Cr[WS(csr, 9)];
254 Tb = Cr[WS(csr, 1)];
255 Tc = Ta + Tb;
256 T1a = Ta - Tb;
257 Tv = Ci[WS(csi, 9)];
258 Tw = Ci[WS(csi, 1)];
259 Tx = Tv + Tw;
260 TP = Tv - Tw;
261 }
262 {
263 E Te, Tf, TR, TS;
264 Te = Cr[WS(csr, 8)];
265 Tf = Cr[WS(csr, 2)];
266 Tg = Te + Tf;
267 Tz = Te - Tf;
268 TR = Ci[WS(csi, 8)];
269 TS = Ci[WS(csi, 2)];
270 TT = TR - TS;
271 T1e = TR + TS;
272 }
273 {
274 E Th, Ti, TA, TB;
275 Th = Cr[WS(csr, 7)];
276 Ti = Cr[WS(csr, 3)];
277 Tj = Th + Ti;
278 T1d = Th - Ti;
279 TA = Ci[WS(csi, 7)];
280 TB = Ci[WS(csi, 3)];
281 TC = TA + TB;
282 TU = TB - TA;
283 }
284 TQ = TO - TP;
285 T1n = T1e - T1d;
286 T1f = T1d + T1e;
287 T12 = TP + TO;
288 T1m = T1b - T1a;
289 TV = TT - TU;
290 T13 = TU + TT;
291 T1c = T1a + T1b;
292 Td = T9 - Tc;
293 Tk = Tg - Tj;
294 Tl = Td + Tk;
295 Ty = Tu + Tx;
296 TD = Tz - TC;
297 TE = Ty + TD;
298 Tn = T9 + Tc;
299 To = Tg + Tj;
300 Tp = Tn + To;
301 TG = Tu - Tx;
302 TH = Tz + TC;
303 TI = TG + TH;
304 }
305 R0[WS(rs, 5)] = FMA(KP2_000000000, Tl, T6);
306 R1[WS(rs, 7)] = FMA(KP2_000000000, TE, Tt);
307 R1[WS(rs, 2)] = FMA(KP2_000000000, TI, TF);
308 R0[0] = FMA(KP2_000000000, Tp, Tm);
309 {
310 E TW, TY, TL, TX, TJ, TK;
311 TW = FNMS(KP1_902113032, TV, KP1_175570504 * TQ);
312 TY = FMA(KP1_902113032, TQ, KP1_175570504 * TV);
313 TJ = FNMS(KP500000000, Tl, T6);
314 TK = KP1_118033988 * (Td - Tk);
315 TL = TJ - TK;
316 TX = TK + TJ;
317 R0[WS(rs, 1)] = TL - TW;
318 R0[WS(rs, 7)] = TX + TY;
319 R0[WS(rs, 9)] = TL + TW;
320 R0[WS(rs, 3)] = TX - TY;
321 }
322 {
323 E T1g, T1i, T19, T1h, T17, T18;
324 T1g = FNMS(KP1_902113032, T1f, KP1_175570504 * T1c);
325 T1i = FMA(KP1_902113032, T1c, KP1_175570504 * T1f);
326 T17 = FNMS(KP500000000, TI, TF);
327 T18 = KP1_118033988 * (TG - TH);
328 T19 = T17 - T18;
329 T1h = T18 + T17;
330 R1[WS(rs, 8)] = T19 - T1g;
331 R1[WS(rs, 4)] = T1h + T1i;
332 R1[WS(rs, 6)] = T19 + T1g;
333 R1[0] = T1h - T1i;
334 }
335 {
336 E T1o, T1q, T1l, T1p, T1j, T1k;
337 T1o = FNMS(KP1_902113032, T1n, KP1_175570504 * T1m);
338 T1q = FMA(KP1_902113032, T1m, KP1_175570504 * T1n);
339 T1j = FNMS(KP500000000, TE, Tt);
340 T1k = KP1_118033988 * (Ty - TD);
341 T1l = T1j - T1k;
342 T1p = T1k + T1j;
343 R1[WS(rs, 3)] = T1l - T1o;
344 R1[WS(rs, 9)] = T1p + T1q;
345 R1[WS(rs, 1)] = T1l + T1o;
346 R1[WS(rs, 5)] = T1p - T1q;
347 }
348 {
349 E T14, T16, T11, T15, TZ, T10;
350 T14 = FNMS(KP1_902113032, T13, KP1_175570504 * T12);
351 T16 = FMA(KP1_902113032, T12, KP1_175570504 * T13);
352 TZ = FNMS(KP500000000, Tp, Tm);
353 T10 = KP1_118033988 * (Tn - To);
354 T11 = TZ - T10;
355 T15 = T10 + TZ;
356 R0[WS(rs, 6)] = T11 - T14;
357 R0[WS(rs, 2)] = T15 + T16;
358 R0[WS(rs, 4)] = T11 + T14;
359 R0[WS(rs, 8)] = T15 - T16;
360 }
361 }
362 }
363 }
364
365 static const kr2c_desc desc = { 20, "r2cb_20", {70, 14, 16, 0}, &GENUS };
366
367 void X(codelet_r2cb_20) (planner *p) {
368 X(kr2c_register) (p, r2cb_20, &desc);
369 }
370
371 #endif