Mercurial > hg > js-dsp-test

comparison fft/fftw/fftw-3.3.4/rdft/scalar/r2cb/r2cb_20.c @ 19:26056e866c29

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add FFTW to comparison table
author Chris Cannam
date Tue, 06 Oct 2015 13:08:39 +0100
parents
children
comparison
equal deleted inserted replaced
18:8db794ca3e0b 19:26056e866c29
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 Tue Mar 4 13:50:24 EST 2014 */
23
24 #include "codelet-rdft.h"
25
26 #ifdef HAVE_FMA
27
28 /* Generated by: ../../../genfft/gen_r2cb.native -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 20 -name r2cb_20 -include 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 * 69 stack variables, 5 constants, and 40 memory accesses
34 */
35 #include "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 TY, T1o, T1m, T14, T12, TX, T1n, T1j, TZ, T13;
48 {
49 E Tr, TD, Tl, T5, T1a, T1l, T1d, T1k, TT, T10, TO, T11, TE, TF, Tk;
50 E TI, TC, T1i, To, TG, T16;
51 {
52 E T4, Tq, T1, T2;
53 T4 = Cr[WS(csr, 5)];
54 Tq = Ci[WS(csi, 5)];
55 T1 = Cr[0];
56 T2 = Cr[WS(csr, 10)];
57 {
58 E Ts, T8, T19, TR, T18, Tb, TS, Tv, Tx, Tf, Ty, T1c, TM, T1b, Ti;
59 E Tz, Tt, Tu, TN, TA;
60 {
61 E TP, TQ, T9, Ta;
62 {
63 E T6, T7, Tp, T3;
64 T6 = Cr[WS(csr, 4)];
65 T7 = Cr[WS(csr, 6)];
66 TP = Ci[WS(csi, 4)];
67 Tp = T1 - T2;
68 T3 = T1 + T2;
69 Ts = T6 - T7;
70 T8 = T6 + T7;
71 Tr = FMA(KP2_000000000, Tq, Tp);
72 TD = FNMS(KP2_000000000, Tq, Tp);
73 Tl = FMA(KP2_000000000, T4, T3);
74 T5 = FNMS(KP2_000000000, T4, T3);
75 TQ = Ci[WS(csi, 6)];
76 }
77 T9 = Cr[WS(csr, 9)];
78 Ta = Cr[WS(csr, 1)];
79 Tt = Ci[WS(csi, 9)];
80 T19 = TP + TQ;
81 TR = TP - TQ;
82 T18 = T9 - Ta;
83 Tb = T9 + Ta;
84 Tu = Ci[WS(csi, 1)];
85 }
86 {
87 E TK, TL, Td, Te, Tg, Th;
88 Td = Cr[WS(csr, 8)];
89 Te = Cr[WS(csr, 2)];
90 TK = Ci[WS(csi, 8)];
91 TS = Tt - Tu;
92 Tv = Tt + Tu;
93 Tx = Td - Te;
94 Tf = Td + Te;
95 TL = Ci[WS(csi, 2)];
96 Tg = Cr[WS(csr, 7)];
97 Th = Cr[WS(csr, 3)];
98 Ty = Ci[WS(csi, 7)];
99 T1c = TK + TL;
100 TM = TK - TL;
101 T1b = Tg - Th;
102 Ti = Tg + Th;
103 Tz = Ci[WS(csi, 3)];
104 }
105 T1a = T18 + T19;
106 T1l = T19 - T18;
107 T1d = T1b + T1c;
108 T1k = T1c - T1b;
109 TT = TR - TS;
110 T10 = TS + TR;
111 TN = Tz - Ty;
112 TA = Ty + Tz;
113 TO = TM - TN;
114 T11 = TN + TM;
115 {
116 E Tm, Tc, Tj, Tn, Tw, TB;
117 Tm = T8 + Tb;
118 Tc = T8 - Tb;
119 Tj = Tf - Ti;
120 Tn = Tf + Ti;
121 TE = Ts - Tv;
122 Tw = Ts + Tv;
123 TB = Tx - TA;
124 TF = Tx + TA;
125 Tk = Tc + Tj;
126 TI = Tc - Tj;
127 TC = Tw + TB;
128 T1i = Tw - TB;
129 TY = Tm - Tn;
130 To = Tm + Tn;
131 }
132 }
133 }
134 R0[WS(rs, 5)] = FMA(KP2_000000000, Tk, T5);
135 R1[WS(rs, 7)] = FMA(KP2_000000000, TC, Tr);
136 TG = TE + TF;
137 T16 = TE - TF;
138 R0[0] = FMA(KP2_000000000, To, Tl);
139 {
140 E TU, TW, T1g, T1e, T15, TV, TJ, TH, T1h, T1f, T17;
141 TU = FNMS(KP618033988, TT, TO);
142 TW = FMA(KP618033988, TO, TT);
143 R1[WS(rs, 2)] = FMA(KP2_000000000, TG, TD);
144 TH = FNMS(KP500000000, Tk, T5);
145 T1g = FNMS(KP618033988, T1a, T1d);
146 T1e = FMA(KP618033988, T1d, T1a);
147 T15 = FNMS(KP500000000, TG, TD);
148 TV = FMA(KP1_118033988, TI, TH);
149 TJ = FNMS(KP1_118033988, TI, TH);
150 T1o = FMA(KP618033988, T1k, T1l);
151 T1m = FNMS(KP618033988, T1l, T1k);
152 R0[WS(rs, 3)] = FNMS(KP1_902113032, TW, TV);
153 R0[WS(rs, 7)] = FMA(KP1_902113032, TW, TV);
154 R0[WS(rs, 1)] = FMA(KP1_902113032, TU, TJ);
155 R0[WS(rs, 9)] = FNMS(KP1_902113032, TU, TJ);
156 T1f = FNMS(KP1_118033988, T16, T15);
157 T17 = FMA(KP1_118033988, T16, T15);
158 T1h = FNMS(KP500000000, TC, Tr);
159 R1[WS(rs, 6)] = FNMS(KP1_902113032, T1g, T1f);
160 R1[WS(rs, 8)] = FMA(KP1_902113032, T1g, T1f);
161 R1[WS(rs, 4)] = FMA(KP1_902113032, T1e, T17);
162 R1[0] = FNMS(KP1_902113032, T1e, T17);
163 T14 = FNMS(KP618033988, T10, T11);
164 T12 = FMA(KP618033988, T11, T10);
165 TX = FNMS(KP500000000, To, Tl);
166 T1n = FMA(KP1_118033988, T1i, T1h);
167 T1j = FNMS(KP1_118033988, T1i, T1h);
168 }
169 }
170 R1[WS(rs, 5)] = FNMS(KP1_902113032, T1o, T1n);
171 R1[WS(rs, 9)] = FMA(KP1_902113032, T1o, T1n);
172 R1[WS(rs, 3)] = FMA(KP1_902113032, T1m, T1j);
173 R1[WS(rs, 1)] = FNMS(KP1_902113032, T1m, T1j);
174 TZ = FMA(KP1_118033988, TY, TX);
175 T13 = FNMS(KP1_118033988, TY, TX);
176 R0[WS(rs, 4)] = FNMS(KP1_902113032, T14, T13);
177 R0[WS(rs, 6)] = FMA(KP1_902113032, T14, T13);
178 R0[WS(rs, 2)] = FMA(KP1_902113032, T12, TZ);
179 R0[WS(rs, 8)] = FNMS(KP1_902113032, T12, TZ);
180 }
181 }
182 }
183
184 static const kr2c_desc desc = { 20, "r2cb_20", {42, 0, 44, 0}, &GENUS };
185
186 void X(codelet_r2cb_20) (planner *p) {
187 X(kr2c_register) (p, r2cb_20, &desc);
188 }
189
190 #else /* HAVE_FMA */
191
192 /* Generated by: ../../../genfft/gen_r2cb.native -compact -variables 4 -pipeline-latency 4 -sign 1 -n 20 -name r2cb_20 -include r2cb.h */
193
194 /*
195 * This function contains 86 FP additions, 30 FP multiplications,
196 * (or, 70 additions, 14 multiplications, 16 fused multiply/add),
197 * 50 stack variables, 5 constants, and 40 memory accesses
198 */
199 #include "r2cb.h"
200
201 static void r2cb_20(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs)
202 {
203 DK(KP1_118033988, +1.118033988749894848204586834365638117720309180);
204 DK(KP500000000, +0.500000000000000000000000000000000000000000000);
205 DK(KP1_902113032, +1.902113032590307144232878666758764286811397268);
206 DK(KP1_175570504, +1.175570504584946258337411909278145537195304875);
207 DK(KP2_000000000, +2.000000000000000000000000000000000000000000000);
208 {
209 INT i;
210 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)) {
211 E T6, TF, Tm, Tt, TQ, T1n, T1f, T12, T1m, TV, T13, T1c, Td, Tk, Tl;
212 E Ty, TD, TE, Tn, To, Tp, TG, TH, TI;
213 {
214 E T5, Ts, T3, Tq;
215 {
216 E T4, Tr, T1, T2;
217 T4 = Cr[WS(csr, 5)];
218 T5 = KP2_000000000 * T4;
219 Tr = Ci[WS(csi, 5)];
220 Ts = KP2_000000000 * Tr;
221 T1 = Cr[0];
222 T2 = Cr[WS(csr, 10)];
223 T3 = T1 + T2;
224 Tq = T1 - T2;
225 }
226 T6 = T3 - T5;
227 TF = Tq - Ts;
228 Tm = T3 + T5;
229 Tt = Tq + Ts;
230 }
231 {
232 E T9, Tu, TO, T1b, Tc, T1a, Tx, TP, Tg, Tz, TT, T1e, Tj, T1d, TC;
233 E TU;
234 {
235 E T7, T8, TM, TN;
236 T7 = Cr[WS(csr, 4)];
237 T8 = Cr[WS(csr, 6)];
238 T9 = T7 + T8;
239 Tu = T7 - T8;
240 TM = Ci[WS(csi, 4)];
241 TN = Ci[WS(csi, 6)];
242 TO = TM - TN;
243 T1b = TM + TN;
244 }
245 {
246 E Ta, Tb, Tv, Tw;
247 Ta = Cr[WS(csr, 9)];
248 Tb = Cr[WS(csr, 1)];
249 Tc = Ta + Tb;
250 T1a = Ta - Tb;
251 Tv = Ci[WS(csi, 9)];
252 Tw = Ci[WS(csi, 1)];
253 Tx = Tv + Tw;
254 TP = Tv - Tw;
255 }
256 {
257 E Te, Tf, TR, TS;
258 Te = Cr[WS(csr, 8)];
259 Tf = Cr[WS(csr, 2)];
260 Tg = Te + Tf;
261 Tz = Te - Tf;
262 TR = Ci[WS(csi, 8)];
263 TS = Ci[WS(csi, 2)];
264 TT = TR - TS;
265 T1e = TR + TS;
266 }
267 {
268 E Th, Ti, TA, TB;
269 Th = Cr[WS(csr, 7)];
270 Ti = Cr[WS(csr, 3)];
271 Tj = Th + Ti;
272 T1d = Th - Ti;
273 TA = Ci[WS(csi, 7)];
274 TB = Ci[WS(csi, 3)];
275 TC = TA + TB;
276 TU = TB - TA;
277 }
278 TQ = TO - TP;
279 T1n = T1e - T1d;
280 T1f = T1d + T1e;
281 T12 = TP + TO;
282 T1m = T1b - T1a;
283 TV = TT - TU;
284 T13 = TU + TT;
285 T1c = T1a + T1b;
286 Td = T9 - Tc;
287 Tk = Tg - Tj;
288 Tl = Td + Tk;
289 Ty = Tu + Tx;
290 TD = Tz - TC;
291 TE = Ty + TD;
292 Tn = T9 + Tc;
293 To = Tg + Tj;
294 Tp = Tn + To;
295 TG = Tu - Tx;
296 TH = Tz + TC;
297 TI = TG + TH;
298 }
299 R0[WS(rs, 5)] = FMA(KP2_000000000, Tl, T6);
300 R1[WS(rs, 7)] = FMA(KP2_000000000, TE, Tt);
301 R1[WS(rs, 2)] = FMA(KP2_000000000, TI, TF);
302 R0[0] = FMA(KP2_000000000, Tp, Tm);
303 {
304 E TW, TY, TL, TX, TJ, TK;
305 TW = FNMS(KP1_902113032, TV, KP1_175570504 * TQ);
306 TY = FMA(KP1_902113032, TQ, KP1_175570504 * TV);
307 TJ = FNMS(KP500000000, Tl, T6);
308 TK = KP1_118033988 * (Td - Tk);
309 TL = TJ - TK;
310 TX = TK + TJ;
311 R0[WS(rs, 1)] = TL - TW;
312 R0[WS(rs, 7)] = TX + TY;
313 R0[WS(rs, 9)] = TL + TW;
314 R0[WS(rs, 3)] = TX - TY;
315 }
316 {
317 E T1g, T1i, T19, T1h, T17, T18;
318 T1g = FNMS(KP1_902113032, T1f, KP1_175570504 * T1c);
319 T1i = FMA(KP1_902113032, T1c, KP1_175570504 * T1f);
320 T17 = FNMS(KP500000000, TI, TF);
321 T18 = KP1_118033988 * (TG - TH);
322 T19 = T17 - T18;
323 T1h = T18 + T17;
324 R1[WS(rs, 8)] = T19 - T1g;
325 R1[WS(rs, 4)] = T1h + T1i;
326 R1[WS(rs, 6)] = T19 + T1g;
327 R1[0] = T1h - T1i;
328 }
329 {
330 E T1o, T1q, T1l, T1p, T1j, T1k;
331 T1o = FNMS(KP1_902113032, T1n, KP1_175570504 * T1m);
332 T1q = FMA(KP1_902113032, T1m, KP1_175570504 * T1n);
333 T1j = FNMS(KP500000000, TE, Tt);
334 T1k = KP1_118033988 * (Ty - TD);
335 T1l = T1j - T1k;
336 T1p = T1k + T1j;
337 R1[WS(rs, 3)] = T1l - T1o;
338 R1[WS(rs, 9)] = T1p + T1q;
339 R1[WS(rs, 1)] = T1l + T1o;
340 R1[WS(rs, 5)] = T1p - T1q;
341 }
342 {
343 E T14, T16, T11, T15, TZ, T10;
344 T14 = FNMS(KP1_902113032, T13, KP1_175570504 * T12);
345 T16 = FMA(KP1_902113032, T12, KP1_175570504 * T13);
346 TZ = FNMS(KP500000000, Tp, Tm);
347 T10 = KP1_118033988 * (Tn - To);
348 T11 = TZ - T10;
349 T15 = T10 + TZ;
350 R0[WS(rs, 6)] = T11 - T14;
351 R0[WS(rs, 2)] = T15 + T16;
352 R0[WS(rs, 4)] = T11 + T14;
353 R0[WS(rs, 8)] = T15 - T16;
354 }
355 }
356 }
357 }
358
359 static const kr2c_desc desc = { 20, "r2cb_20", {70, 14, 16, 0}, &GENUS };
360
361 void X(codelet_r2cb_20) (planner *p) {
362 X(kr2c_register) (p, r2cb_20, &desc);
363 }
364
365 #endif /* HAVE_FMA */