Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/dft/simd/common/t1fuv_9.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 07:38:00 EST 2012 */
23
24 #include "codelet-dft.h"
25
26 #ifdef HAVE_FMA
27
28 /* Generated by: ../../../genfft/gen_twiddle_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 9 -name t1fuv_9 -include t1fu.h */
29
30 /*
31 * This function contains 54 FP additions, 54 FP multiplications,
32 * (or, 20 additions, 20 multiplications, 34 fused multiply/add),
33 * 67 stack variables, 19 constants, and 18 memory accesses
34 */
35 #include "t1fu.h"
36
37 static void t1fuv_9(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
38 {
39 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
40 DVK(KP826351822, +0.826351822333069651148283373230685203999624323);
41 DVK(KP879385241, +0.879385241571816768108218554649462939872416269);
42 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
43 DVK(KP666666666, +0.666666666666666666666666666666666666666666667);
44 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
45 DVK(KP907603734, +0.907603734547952313649323976213898122064543220);
46 DVK(KP420276625, +0.420276625461206169731530603237061658838781920);
47 DVK(KP673648177, +0.673648177666930348851716626769314796000375677);
48 DVK(KP898197570, +0.898197570222573798468955502359086394667167570);
49 DVK(KP347296355, +0.347296355333860697703433253538629592000751354);
50 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
51 DVK(KP439692620, +0.439692620785908384054109277324731469936208134);
52 DVK(KP203604859, +0.203604859554852403062088995281827210665664861);
53 DVK(KP152703644, +0.152703644666139302296566746461370407999248646);
54 DVK(KP586256827, +0.586256827714544512072145703099641959914944179);
55 DVK(KP968908795, +0.968908795874236621082202410917456709164223497);
56 DVK(KP726681596, +0.726681596905677465811651808188092531873167623);
57 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
58 {
59 INT m;
60 R *x;
61 x = ri;
62 for (m = mb, W = W + (mb * ((TWVL / VL) * 16)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 16), MAKE_VOLATILE_STRIDE(9, rs)) {
63 V T1, T3, T5, T9, Th, Tb, Td, Tj, Tl, TD, T6;
64 T1 = LD(&(x[0]), ms, &(x[0]));
65 {
66 V T2, T4, T8, Tg;
67 T2 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
68 T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
69 T8 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
70 Tg = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
71 {
72 V Ta, Tc, Ti, Tk;
73 Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
74 Tc = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
75 Ti = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
76 Tk = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
77 T3 = BYTWJ(&(W[TWVL * 4]), T2);
78 T5 = BYTWJ(&(W[TWVL * 10]), T4);
79 T9 = BYTWJ(&(W[0]), T8);
80 Th = BYTWJ(&(W[TWVL * 2]), Tg);
81 Tb = BYTWJ(&(W[TWVL * 6]), Ta);
82 Td = BYTWJ(&(W[TWVL * 12]), Tc);
83 Tj = BYTWJ(&(W[TWVL * 8]), Ti);
84 Tl = BYTWJ(&(W[TWVL * 14]), Tk);
85 }
86 }
87 TD = VSUB(T5, T3);
88 T6 = VADD(T3, T5);
89 {
90 V Tt, Te, Tu, Tm, Tr, T7;
91 Tt = VSUB(Tb, Td);
92 Te = VADD(Tb, Td);
93 Tu = VSUB(Tl, Tj);
94 Tm = VADD(Tj, Tl);
95 Tr = VFNMS(LDK(KP500000000), T6, T1);
96 T7 = VADD(T1, T6);
97 {
98 V Tv, Tf, Ts, Tn;
99 Tv = VFNMS(LDK(KP500000000), Te, T9);
100 Tf = VADD(T9, Te);
101 Ts = VFNMS(LDK(KP500000000), Tm, Th);
102 Tn = VADD(Th, Tm);
103 {
104 V TG, TK, Tw, TJ, TF, TA, To, Tq;
105 TG = VFNMS(LDK(KP726681596), Tt, Tv);
106 TK = VFMA(LDK(KP968908795), Tv, Tt);
107 Tw = VFNMS(LDK(KP586256827), Tv, Tu);
108 TJ = VFNMS(LDK(KP152703644), Tu, Ts);
109 TF = VFMA(LDK(KP203604859), Ts, Tu);
110 TA = VFNMS(LDK(KP439692620), Tt, Ts);
111 To = VADD(Tf, Tn);
112 Tq = VMUL(LDK(KP866025403), VSUB(Tn, Tf));
113 {
114 V TQ, TH, TL, TN, TB, Tp, Ty, TI, Tx;
115 Tx = VFNMS(LDK(KP347296355), Tw, Tt);
116 TQ = VFNMS(LDK(KP898197570), TG, TF);
117 TH = VFMA(LDK(KP898197570), TG, TF);
118 TL = VFMA(LDK(KP673648177), TK, TJ);
119 TN = VFNMS(LDK(KP673648177), TK, TJ);
120 TB = VFNMS(LDK(KP420276625), TA, Tu);
121 ST(&(x[0]), VADD(T7, To), ms, &(x[0]));
122 Tp = VFNMS(LDK(KP500000000), To, T7);
123 Ty = VFNMS(LDK(KP907603734), Tx, Ts);
124 TI = VFMA(LDK(KP852868531), TH, Tr);
125 {
126 V TO, TR, TM, TC, Tz, TP, TS, TE;
127 TO = VFNMS(LDK(KP500000000), TH, TN);
128 TR = VFMA(LDK(KP666666666), TL, TQ);
129 TM = VMUL(LDK(KP984807753), VFNMS(LDK(KP879385241), TD, TL));
130 TC = VFNMS(LDK(KP826351822), TB, Tv);
131 ST(&(x[WS(rs, 6)]), VFNMSI(Tq, Tp), ms, &(x[0]));
132 ST(&(x[WS(rs, 3)]), VFMAI(Tq, Tp), ms, &(x[WS(rs, 1)]));
133 Tz = VFNMS(LDK(KP939692620), Ty, Tr);
134 TP = VFMA(LDK(KP852868531), TO, Tr);
135 TS = VMUL(LDK(KP866025403), VFMA(LDK(KP852868531), TR, TD));
136 ST(&(x[WS(rs, 8)]), VFMAI(TM, TI), ms, &(x[0]));
137 ST(&(x[WS(rs, 1)]), VFNMSI(TM, TI), ms, &(x[WS(rs, 1)]));
138 TE = VMUL(LDK(KP984807753), VFMA(LDK(KP879385241), TD, TC));
139 ST(&(x[WS(rs, 4)]), VFMAI(TS, TP), ms, &(x[0]));
140 ST(&(x[WS(rs, 5)]), VFNMSI(TS, TP), ms, &(x[WS(rs, 1)]));
141 ST(&(x[WS(rs, 7)]), VFMAI(TE, Tz), ms, &(x[WS(rs, 1)]));
142 ST(&(x[WS(rs, 2)]), VFNMSI(TE, Tz), ms, &(x[0]));
143 }
144 }
145 }
146 }
147 }
148 }
149 }
150 VLEAVE();
151 }
152
153 static const tw_instr twinstr[] = {
154 VTW(0, 1),
155 VTW(0, 2),
156 VTW(0, 3),
157 VTW(0, 4),
158 VTW(0, 5),
159 VTW(0, 6),
160 VTW(0, 7),
161 VTW(0, 8),
162 {TW_NEXT, VL, 0}
163 };
164
165 static const ct_desc desc = { 9, XSIMD_STRING("t1fuv_9"), twinstr, &GENUS, {20, 20, 34, 0}, 0, 0, 0 };
166
167 void XSIMD(codelet_t1fuv_9) (planner *p) {
168 X(kdft_dit_register) (p, t1fuv_9, &desc);
169 }
170 #else /* HAVE_FMA */
171
172 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 9 -name t1fuv_9 -include t1fu.h */
173
174 /*
175 * This function contains 54 FP additions, 42 FP multiplications,
176 * (or, 38 additions, 26 multiplications, 16 fused multiply/add),
177 * 38 stack variables, 14 constants, and 18 memory accesses
178 */
179 #include "t1fu.h"
180
181 static void t1fuv_9(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
182 {
183 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
184 DVK(KP296198132, +0.296198132726023843175338011893050938967728390);
185 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
186 DVK(KP173648177, +0.173648177666930348851716626769314796000375677);
187 DVK(KP556670399, +0.556670399226419366452912952047023132968291906);
188 DVK(KP766044443, +0.766044443118978035202392650555416673935832457);
189 DVK(KP642787609, +0.642787609686539326322643409907263432907559884);
190 DVK(KP663413948, +0.663413948168938396205421319635891297216863310);
191 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
192 DVK(KP150383733, +0.150383733180435296639271897612501926072238258);
193 DVK(KP342020143, +0.342020143325668733044099614682259580763083368);
194 DVK(KP813797681, +0.813797681349373692844693217248393223289101568);
195 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
196 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
197 {
198 INT m;
199 R *x;
200 x = ri;
201 for (m = mb, W = W + (mb * ((TWVL / VL) * 16)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 16), MAKE_VOLATILE_STRIDE(9, rs)) {
202 V T1, T6, TA, Tt, Tf, Ts, Tw, Tn, Tv;
203 T1 = LD(&(x[0]), ms, &(x[0]));
204 {
205 V T3, T5, T2, T4;
206 T2 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
207 T3 = BYTWJ(&(W[TWVL * 4]), T2);
208 T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
209 T5 = BYTWJ(&(W[TWVL * 10]), T4);
210 T6 = VADD(T3, T5);
211 TA = VMUL(LDK(KP866025403), VSUB(T5, T3));
212 }
213 {
214 V T9, Td, Tb, T8, Tc, Ta, Te;
215 T8 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
216 T9 = BYTWJ(&(W[0]), T8);
217 Tc = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
218 Td = BYTWJ(&(W[TWVL * 12]), Tc);
219 Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
220 Tb = BYTWJ(&(W[TWVL * 6]), Ta);
221 Tt = VSUB(Td, Tb);
222 Te = VADD(Tb, Td);
223 Tf = VADD(T9, Te);
224 Ts = VFNMS(LDK(KP500000000), Te, T9);
225 }
226 {
227 V Th, Tl, Tj, Tg, Tk, Ti, Tm;
228 Tg = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
229 Th = BYTWJ(&(W[TWVL * 2]), Tg);
230 Tk = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
231 Tl = BYTWJ(&(W[TWVL * 14]), Tk);
232 Ti = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
233 Tj = BYTWJ(&(W[TWVL * 8]), Ti);
234 Tw = VSUB(Tl, Tj);
235 Tm = VADD(Tj, Tl);
236 Tn = VADD(Th, Tm);
237 Tv = VFNMS(LDK(KP500000000), Tm, Th);
238 }
239 {
240 V Tq, T7, To, Tp;
241 Tq = VBYI(VMUL(LDK(KP866025403), VSUB(Tn, Tf)));
242 T7 = VADD(T1, T6);
243 To = VADD(Tf, Tn);
244 Tp = VFNMS(LDK(KP500000000), To, T7);
245 ST(&(x[0]), VADD(T7, To), ms, &(x[0]));
246 ST(&(x[WS(rs, 3)]), VADD(Tp, Tq), ms, &(x[WS(rs, 1)]));
247 ST(&(x[WS(rs, 6)]), VSUB(Tp, Tq), ms, &(x[0]));
248 }
249 {
250 V TI, TB, TC, TD, Tu, Tx, Ty, Tr, TH;
251 TI = VBYI(VSUB(VFNMS(LDK(KP342020143), Tv, VFNMS(LDK(KP150383733), Tt, VFNMS(LDK(KP984807753), Ts, VMUL(LDK(KP813797681), Tw)))), TA));
252 TB = VFNMS(LDK(KP642787609), Ts, VMUL(LDK(KP663413948), Tt));
253 TC = VFNMS(LDK(KP984807753), Tv, VMUL(LDK(KP150383733), Tw));
254 TD = VADD(TB, TC);
255 Tu = VFMA(LDK(KP766044443), Ts, VMUL(LDK(KP556670399), Tt));
256 Tx = VFMA(LDK(KP173648177), Tv, VMUL(LDK(KP852868531), Tw));
257 Ty = VADD(Tu, Tx);
258 Tr = VFNMS(LDK(KP500000000), T6, T1);
259 TH = VFMA(LDK(KP173648177), Ts, VFNMS(LDK(KP296198132), Tw, VFNMS(LDK(KP939692620), Tv, VFNMS(LDK(KP852868531), Tt, Tr))));
260 ST(&(x[WS(rs, 7)]), VSUB(TH, TI), ms, &(x[WS(rs, 1)]));
261 ST(&(x[WS(rs, 2)]), VADD(TH, TI), ms, &(x[0]));
262 {
263 V Tz, TE, TF, TG;
264 Tz = VADD(Tr, Ty);
265 TE = VBYI(VADD(TA, TD));
266 ST(&(x[WS(rs, 8)]), VSUB(Tz, TE), ms, &(x[0]));
267 ST(&(x[WS(rs, 1)]), VADD(TE, Tz), ms, &(x[WS(rs, 1)]));
268 TF = VFMA(LDK(KP866025403), VSUB(TB, TC), VFNMS(LDK(KP500000000), Ty, Tr));
269 TG = VBYI(VADD(TA, VFNMS(LDK(KP500000000), TD, VMUL(LDK(KP866025403), VSUB(Tx, Tu)))));
270 ST(&(x[WS(rs, 5)]), VSUB(TF, TG), ms, &(x[WS(rs, 1)]));
271 ST(&(x[WS(rs, 4)]), VADD(TF, TG), ms, &(x[0]));
272 }
273 }
274 }
275 }
276 VLEAVE();
277 }
278
279 static const tw_instr twinstr[] = {
280 VTW(0, 1),
281 VTW(0, 2),
282 VTW(0, 3),
283 VTW(0, 4),
284 VTW(0, 5),
285 VTW(0, 6),
286 VTW(0, 7),
287 VTW(0, 8),
288 {TW_NEXT, VL, 0}
289 };
290
291 static const ct_desc desc = { 9, XSIMD_STRING("t1fuv_9"), twinstr, &GENUS, {38, 26, 16, 0}, 0, 0, 0 };
292
293 void XSIMD(codelet_t1fuv_9) (planner *p) {
294 X(kdft_dit_register) (p, t1fuv_9, &desc);
295 }
296 #endif /* HAVE_FMA */