Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/dft/simd/common/n1bv_11.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 07:37:00 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 -sign 1 -n 11 -name n1bv_11 -include n1b.h */
29
30 /*
31 * This function contains 70 FP additions, 60 FP multiplications,
32 * (or, 15 additions, 5 multiplications, 55 fused multiply/add),
33 * 67 stack variables, 11 constants, and 22 memory accesses
34 */
35 #include "n1b.h"
36
37 static void n1bv_11(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 DVK(KP959492973, +0.959492973614497389890368057066327699062454848);
40 DVK(KP876768831, +0.876768831002589333891339807079336796764054852);
41 DVK(KP918985947, +0.918985947228994779780736114132655398124909697);
42 DVK(KP989821441, +0.989821441880932732376092037776718787376519372);
43 DVK(KP778434453, +0.778434453334651800608337670740821884709317477);
44 DVK(KP830830026, +0.830830026003772851058548298459246407048009821);
45 DVK(KP372785597, +0.372785597771792209609773152906148328659002598);
46 DVK(KP634356270, +0.634356270682424498893150776899916060542806975);
47 DVK(KP715370323, +0.715370323453429719112414662767260662417897278);
48 DVK(KP342584725, +0.342584725681637509502641509861112333758894680);
49 DVK(KP521108558, +0.521108558113202722944698153526659300680427422);
50 {
51 INT i;
52 const R *xi;
53 R *xo;
54 xi = ii;
55 xo = io;
56 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(22, is), MAKE_VOLATILE_STRIDE(22, os)) {
57 V T1, Tb, T4, Tq, Tg, Tm, T7, Tp, Ta, To, Tc, T11;
58 T1 = LD(&(xi[0]), ivs, &(xi[0]));
59 {
60 V T2, T3, Te, Tf;
61 T2 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
62 T3 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
63 Te = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
64 Tf = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
65 {
66 V T5, T6, T8, T9;
67 T5 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
68 T6 = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
69 T8 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
70 T9 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
71 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
72 T4 = VADD(T2, T3);
73 Tq = VSUB(T2, T3);
74 Tg = VADD(Te, Tf);
75 Tm = VSUB(Te, Tf);
76 T7 = VADD(T5, T6);
77 Tp = VSUB(T5, T6);
78 Ta = VADD(T8, T9);
79 To = VSUB(T8, T9);
80 Tc = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
81 }
82 }
83 T11 = VFMA(LDK(KP521108558), Tm, Tq);
84 {
85 V TA, TS, TE, TW, Td, Tn, Ts, Tw, Tr, Tv, TT, TF;
86 Tr = VFNMS(LDK(KP521108558), Tq, Tp);
87 Tv = VFNMS(LDK(KP342584725), T7, Tg);
88 TA = VFMA(LDK(KP715370323), To, Tq);
89 TS = VFMA(LDK(KP521108558), To, Tm);
90 TE = VFNMS(LDK(KP342584725), T4, Ta);
91 TW = VFNMS(LDK(KP342584725), Ta, T7);
92 Td = VADD(Tb, Tc);
93 Tn = VSUB(Tb, Tc);
94 Ts = VFNMS(LDK(KP715370323), Tr, To);
95 Tw = VFNMS(LDK(KP634356270), Tv, T4);
96 TT = VFNMS(LDK(KP715370323), TS, Tp);
97 TF = VFNMS(LDK(KP634356270), TE, Tg);
98 {
99 V Tu, TV, TD, TL, T14, TP, TZ, Tj, Tz, TI, TB, TJ, TM;
100 TB = VFMA(LDK(KP372785597), Tn, TA);
101 TJ = VFNMS(LDK(KP521108558), Tp, Tn);
102 {
103 V T12, TN, TX, Th;
104 T12 = VFMA(LDK(KP715370323), T11, Tn);
105 ST(&(xo[0]), VADD(Tg, VADD(Td, VADD(Ta, VADD(T7, VADD(T4, T1))))), ovs, &(xo[0]));
106 TN = VFNMS(LDK(KP342584725), Td, T4);
107 TX = VFNMS(LDK(KP634356270), TW, Td);
108 Th = VFNMS(LDK(KP342584725), Tg, Td);
109 {
110 V Tt, Tx, TU, TG;
111 Tt = VFNMS(LDK(KP830830026), Ts, Tn);
112 Tx = VFNMS(LDK(KP778434453), Tw, Ta);
113 TU = VFMA(LDK(KP830830026), TT, Tq);
114 TG = VFNMS(LDK(KP778434453), TF, Td);
115 {
116 V TC, TK, T13, TO;
117 TC = VFNMS(LDK(KP830830026), TB, Tm);
118 TK = VFMA(LDK(KP715370323), TJ, Tm);
119 T13 = VFMA(LDK(KP830830026), T12, Tp);
120 TO = VFNMS(LDK(KP634356270), TN, T7);
121 {
122 V TY, Ti, Ty, TH;
123 TY = VFNMS(LDK(KP778434453), TX, T4);
124 Ti = VFNMS(LDK(KP634356270), Th, Ta);
125 Tu = VMUL(LDK(KP989821441), VFNMS(LDK(KP918985947), Tt, Tm));
126 Ty = VFNMS(LDK(KP876768831), Tx, Td);
127 TV = VMUL(LDK(KP989821441), VFNMS(LDK(KP918985947), TU, Tn));
128 TH = VFNMS(LDK(KP876768831), TG, T7);
129 TD = VMUL(LDK(KP989821441), VFMA(LDK(KP918985947), TC, Tp));
130 TL = VFNMS(LDK(KP830830026), TK, To);
131 T14 = VMUL(LDK(KP989821441), VFMA(LDK(KP918985947), T13, To));
132 TP = VFNMS(LDK(KP778434453), TO, Tg);
133 TZ = VFNMS(LDK(KP876768831), TY, Tg);
134 Tj = VFNMS(LDK(KP778434453), Ti, T7);
135 Tz = VFNMS(LDK(KP959492973), Ty, T1);
136 TI = VFNMS(LDK(KP959492973), TH, T1);
137 }
138 }
139 }
140 }
141 TM = VMUL(LDK(KP989821441), VFNMS(LDK(KP918985947), TL, Tq));
142 {
143 V TQ, T10, Tk, TR, Tl;
144 TQ = VFNMS(LDK(KP876768831), TP, Ta);
145 T10 = VFNMS(LDK(KP959492973), TZ, T1);
146 Tk = VFNMS(LDK(KP876768831), Tj, T4);
147 ST(&(xo[WS(os, 7)]), VFMAI(TD, Tz), ovs, &(xo[WS(os, 1)]));
148 ST(&(xo[WS(os, 4)]), VFNMSI(TD, Tz), ovs, &(xo[0]));
149 ST(&(xo[WS(os, 8)]), VFNMSI(TM, TI), ovs, &(xo[0]));
150 ST(&(xo[WS(os, 3)]), VFMAI(TM, TI), ovs, &(xo[WS(os, 1)]));
151 TR = VFNMS(LDK(KP959492973), TQ, T1);
152 ST(&(xo[WS(os, 10)]), VFNMSI(T14, T10), ovs, &(xo[0]));
153 ST(&(xo[WS(os, 1)]), VFMAI(T14, T10), ovs, &(xo[WS(os, 1)]));
154 Tl = VFNMS(LDK(KP959492973), Tk, T1);
155 ST(&(xo[WS(os, 9)]), VFMAI(TV, TR), ovs, &(xo[WS(os, 1)]));
156 ST(&(xo[WS(os, 2)]), VFNMSI(TV, TR), ovs, &(xo[0]));
157 ST(&(xo[WS(os, 6)]), VFNMSI(Tu, Tl), ovs, &(xo[0]));
158 ST(&(xo[WS(os, 5)]), VFMAI(Tu, Tl), ovs, &(xo[WS(os, 1)]));
159 }
160 }
161 }
162 }
163 }
164 VLEAVE();
165 }
166
167 static const kdft_desc desc = { 11, XSIMD_STRING("n1bv_11"), {15, 5, 55, 0}, &GENUS, 0, 0, 0, 0 };
168
169 void XSIMD(codelet_n1bv_11) (planner *p) {
170 X(kdft_register) (p, n1bv_11, &desc);
171 }
172
173 #else /* HAVE_FMA */
174
175 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 11 -name n1bv_11 -include n1b.h */
176
177 /*
178 * This function contains 70 FP additions, 50 FP multiplications,
179 * (or, 30 additions, 10 multiplications, 40 fused multiply/add),
180 * 32 stack variables, 10 constants, and 22 memory accesses
181 */
182 #include "n1b.h"
183
184 static void n1bv_11(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
185 {
186 DVK(KP959492973, +0.959492973614497389890368057066327699062454848);
187 DVK(KP654860733, +0.654860733945285064056925072466293553183791199);
188 DVK(KP142314838, +0.142314838273285140443792668616369668791051361);
189 DVK(KP415415013, +0.415415013001886425529274149229623203524004910);
190 DVK(KP841253532, +0.841253532831181168861811648919367717513292498);
191 DVK(KP540640817, +0.540640817455597582107635954318691695431770608);
192 DVK(KP909631995, +0.909631995354518371411715383079028460060241051);
193 DVK(KP989821441, +0.989821441880932732376092037776718787376519372);
194 DVK(KP755749574, +0.755749574354258283774035843972344420179717445);
195 DVK(KP281732556, +0.281732556841429697711417915346616899035777899);
196 {
197 INT i;
198 const R *xi;
199 R *xo;
200 xi = ii;
201 xo = io;
202 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(22, is), MAKE_VOLATILE_STRIDE(22, os)) {
203 V Th, T3, Tm, Tf, Ti, Tc, Tj, T9, Tk, T6, Tl, Ta, Tb, Ts, Tt;
204 Th = LD(&(xi[0]), ivs, &(xi[0]));
205 {
206 V T1, T2, Td, Te;
207 T1 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
208 T2 = LD(&(xi[WS(is, 10)]), ivs, &(xi[0]));
209 T3 = VSUB(T1, T2);
210 Tm = VADD(T1, T2);
211 Td = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
212 Te = LD(&(xi[WS(is, 9)]), ivs, &(xi[WS(is, 1)]));
213 Tf = VSUB(Td, Te);
214 Ti = VADD(Td, Te);
215 }
216 Ta = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
217 Tb = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
218 Tc = VSUB(Ta, Tb);
219 Tj = VADD(Ta, Tb);
220 {
221 V T7, T8, T4, T5;
222 T7 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
223 T8 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
224 T9 = VSUB(T7, T8);
225 Tk = VADD(T7, T8);
226 T4 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
227 T5 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
228 T6 = VSUB(T4, T5);
229 Tl = VADD(T4, T5);
230 }
231 ST(&(xo[0]), VADD(Th, VADD(Tm, VADD(Ti, VADD(Tl, VADD(Tj, Tk))))), ovs, &(xo[0]));
232 {
233 V Tg, Tn, Tu, Tv;
234 Tg = VBYI(VFMA(LDK(KP281732556), T3, VFMA(LDK(KP755749574), T6, VFNMS(LDK(KP909631995), Tc, VFNMS(LDK(KP540640817), Tf, VMUL(LDK(KP989821441), T9))))));
235 Tn = VFMA(LDK(KP841253532), Ti, VFMA(LDK(KP415415013), Tj, VFNMS(LDK(KP142314838), Tk, VFNMS(LDK(KP654860733), Tl, VFNMS(LDK(KP959492973), Tm, Th)))));
236 ST(&(xo[WS(os, 5)]), VADD(Tg, Tn), ovs, &(xo[WS(os, 1)]));
237 ST(&(xo[WS(os, 6)]), VSUB(Tn, Tg), ovs, &(xo[0]));
238 Tu = VBYI(VFMA(LDK(KP755749574), T3, VFMA(LDK(KP540640817), T6, VFNMS(LDK(KP909631995), T9, VFNMS(LDK(KP989821441), Tf, VMUL(LDK(KP281732556), Tc))))));
239 Tv = VFMA(LDK(KP841253532), Tl, VFMA(LDK(KP415415013), Tk, VFNMS(LDK(KP959492973), Tj, VFNMS(LDK(KP142314838), Ti, VFNMS(LDK(KP654860733), Tm, Th)))));
240 ST(&(xo[WS(os, 4)]), VADD(Tu, Tv), ovs, &(xo[0]));
241 ST(&(xo[WS(os, 7)]), VSUB(Tv, Tu), ovs, &(xo[WS(os, 1)]));
242 }
243 Ts = VBYI(VFMA(LDK(KP909631995), T3, VFNMS(LDK(KP540640817), T9, VFNMS(LDK(KP989821441), Tc, VFNMS(LDK(KP281732556), T6, VMUL(LDK(KP755749574), Tf))))));
244 Tt = VFMA(LDK(KP415415013), Tm, VFMA(LDK(KP841253532), Tk, VFNMS(LDK(KP142314838), Tj, VFNMS(LDK(KP959492973), Tl, VFNMS(LDK(KP654860733), Ti, Th)))));
245 ST(&(xo[WS(os, 2)]), VADD(Ts, Tt), ovs, &(xo[0]));
246 ST(&(xo[WS(os, 9)]), VSUB(Tt, Ts), ovs, &(xo[WS(os, 1)]));
247 {
248 V Tq, Tr, To, Tp;
249 Tq = VBYI(VFMA(LDK(KP540640817), T3, VFMA(LDK(KP909631995), Tf, VFMA(LDK(KP989821441), T6, VFMA(LDK(KP755749574), Tc, VMUL(LDK(KP281732556), T9))))));
250 Tr = VFMA(LDK(KP841253532), Tm, VFMA(LDK(KP415415013), Ti, VFNMS(LDK(KP959492973), Tk, VFNMS(LDK(KP654860733), Tj, VFNMS(LDK(KP142314838), Tl, Th)))));
251 ST(&(xo[WS(os, 1)]), VADD(Tq, Tr), ovs, &(xo[WS(os, 1)]));
252 ST(&(xo[WS(os, 10)]), VSUB(Tr, Tq), ovs, &(xo[0]));
253 To = VBYI(VFMA(LDK(KP989821441), T3, VFMA(LDK(KP540640817), Tc, VFNMS(LDK(KP909631995), T6, VFNMS(LDK(KP281732556), Tf, VMUL(LDK(KP755749574), T9))))));
254 Tp = VFMA(LDK(KP415415013), Tl, VFMA(LDK(KP841253532), Tj, VFNMS(LDK(KP654860733), Tk, VFNMS(LDK(KP959492973), Ti, VFNMS(LDK(KP142314838), Tm, Th)))));
255 ST(&(xo[WS(os, 3)]), VADD(To, Tp), ovs, &(xo[WS(os, 1)]));
256 ST(&(xo[WS(os, 8)]), VSUB(Tp, To), ovs, &(xo[0]));
257 }
258 }
259 }
260 VLEAVE();
261 }
262
263 static const kdft_desc desc = { 11, XSIMD_STRING("n1bv_11"), {30, 10, 40, 0}, &GENUS, 0, 0, 0, 0 };
264
265 void XSIMD(codelet_n1bv_11) (planner *p) {
266 X(kdft_register) (p, n1bv_11, &desc);
267 }
268
269 #endif /* HAVE_FMA */