Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.3/dft/simd/common/n1bv_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 07:36:59 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 9 -name n1bv_9 -include n1b.h */
29
30 /*
31 * This function contains 46 FP additions, 38 FP multiplications,
32 * (or, 12 additions, 4 multiplications, 34 fused multiply/add),
33 * 68 stack variables, 19 constants, and 18 memory accesses
34 */
35 #include "n1b.h"
36
37 static void n1bv_9(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
38 {
39 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
40 DVK(KP907603734, +0.907603734547952313649323976213898122064543220);
41 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
42 DVK(KP666666666, +0.666666666666666666666666666666666666666666667);
43 DVK(KP879385241, +0.879385241571816768108218554649462939872416269);
44 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
45 DVK(KP826351822, +0.826351822333069651148283373230685203999624323);
46 DVK(KP347296355, +0.347296355333860697703433253538629592000751354);
47 DVK(KP898197570, +0.898197570222573798468955502359086394667167570);
48 DVK(KP673648177, +0.673648177666930348851716626769314796000375677);
49 DVK(KP420276625, +0.420276625461206169731530603237061658838781920);
50 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
51 DVK(KP586256827, +0.586256827714544512072145703099641959914944179);
52 DVK(KP968908795, +0.968908795874236621082202410917456709164223497);
53 DVK(KP726681596, +0.726681596905677465811651808188092531873167623);
54 DVK(KP439692620, +0.439692620785908384054109277324731469936208134);
55 DVK(KP203604859, +0.203604859554852403062088995281827210665664861);
56 DVK(KP152703644, +0.152703644666139302296566746461370407999248646);
57 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
58 {
59 INT i;
60 const R *xi;
61 R *xo;
62 xi = ii;
63 xo = io;
64 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(18, is), MAKE_VOLATILE_STRIDE(18, os)) {
65 V T1, T2, T3, T6, Tf, T7, T8, Tb, Tc, Tp, T4;
66 T1 = LD(&(xi[0]), ivs, &(xi[0]));
67 T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
68 T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
69 T6 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
70 Tf = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
71 T7 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
72 T8 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
73 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
74 Tc = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
75 Tp = VSUB(T2, T3);
76 T4 = VADD(T2, T3);
77 {
78 V Te, T9, Tg, Td, TF, T5;
79 Te = VSUB(T8, T7);
80 T9 = VADD(T7, T8);
81 Tg = VADD(Tb, Tc);
82 Td = VSUB(Tb, Tc);
83 TF = VADD(T1, T4);
84 T5 = VFNMS(LDK(KP500000000), T4, T1);
85 {
86 V Ta, TH, Th, TG;
87 Ta = VFNMS(LDK(KP500000000), T9, T6);
88 TH = VADD(T6, T9);
89 Th = VFNMS(LDK(KP500000000), Tg, Tf);
90 TG = VADD(Tf, Tg);
91 {
92 V Tr, Tu, Tm, Tv, Ts, Ti, TI, TK;
93 Tr = VFNMS(LDK(KP152703644), Te, Ta);
94 Tu = VFMA(LDK(KP203604859), Ta, Te);
95 Tm = VFNMS(LDK(KP439692620), Td, Ta);
96 Tv = VFNMS(LDK(KP726681596), Td, Th);
97 Ts = VFMA(LDK(KP968908795), Th, Td);
98 Ti = VFNMS(LDK(KP586256827), Th, Te);
99 TI = VADD(TG, TH);
100 TK = VMUL(LDK(KP866025403), VSUB(TG, TH));
101 {
102 V Tt, TA, Tw, Tz, Tj, TJ, To, TE, Tn;
103 Tn = VFNMS(LDK(KP420276625), Tm, Te);
104 Tt = VFNMS(LDK(KP673648177), Ts, Tr);
105 TA = VFMA(LDK(KP673648177), Ts, Tr);
106 Tw = VFMA(LDK(KP898197570), Tv, Tu);
107 Tz = VFNMS(LDK(KP898197570), Tv, Tu);
108 Tj = VFNMS(LDK(KP347296355), Ti, Td);
109 ST(&(xo[0]), VADD(TI, TF), ovs, &(xo[0]));
110 TJ = VFNMS(LDK(KP500000000), TI, TF);
111 To = VFNMS(LDK(KP826351822), Tn, Th);
112 TE = VMUL(LDK(KP984807753), VFMA(LDK(KP879385241), Tp, TA));
113 {
114 V TB, TD, Tx, Tk, Tq, TC, Ty, Tl;
115 TB = VFMA(LDK(KP666666666), TA, Tz);
116 TD = VFMA(LDK(KP852868531), Tw, T5);
117 Tx = VFNMS(LDK(KP500000000), Tw, Tt);
118 Tk = VFNMS(LDK(KP907603734), Tj, Ta);
119 ST(&(xo[WS(os, 6)]), VFNMSI(TK, TJ), ovs, &(xo[0]));
120 ST(&(xo[WS(os, 3)]), VFMAI(TK, TJ), ovs, &(xo[WS(os, 1)]));
121 Tq = VMUL(LDK(KP984807753), VFNMS(LDK(KP879385241), Tp, To));
122 TC = VMUL(LDK(KP866025403), VFNMS(LDK(KP852868531), TB, Tp));
123 ST(&(xo[WS(os, 8)]), VFNMSI(TE, TD), ovs, &(xo[0]));
124 ST(&(xo[WS(os, 1)]), VFMAI(TE, TD), ovs, &(xo[WS(os, 1)]));
125 Ty = VFMA(LDK(KP852868531), Tx, T5);
126 Tl = VFNMS(LDK(KP939692620), Tk, T5);
127 ST(&(xo[WS(os, 5)]), VFNMSI(TC, Ty), ovs, &(xo[WS(os, 1)]));
128 ST(&(xo[WS(os, 4)]), VFMAI(TC, Ty), ovs, &(xo[0]));
129 ST(&(xo[WS(os, 2)]), VFMAI(Tq, Tl), ovs, &(xo[0]));
130 ST(&(xo[WS(os, 7)]), VFNMSI(Tq, Tl), ovs, &(xo[WS(os, 1)]));
131 }
132 }
133 }
134 }
135 }
136 }
137 }
138 VLEAVE();
139 }
140
141 static const kdft_desc desc = { 9, XSIMD_STRING("n1bv_9"), {12, 4, 34, 0}, &GENUS, 0, 0, 0, 0 };
142
143 void XSIMD(codelet_n1bv_9) (planner *p) {
144 X(kdft_register) (p, n1bv_9, &desc);
145 }
146
147 #else /* HAVE_FMA */
148
149 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 9 -name n1bv_9 -include n1b.h */
150
151 /*
152 * This function contains 46 FP additions, 26 FP multiplications,
153 * (or, 30 additions, 10 multiplications, 16 fused multiply/add),
154 * 41 stack variables, 14 constants, and 18 memory accesses
155 */
156 #include "n1b.h"
157
158 static void n1bv_9(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
159 {
160 DVK(KP342020143, +0.342020143325668733044099614682259580763083368);
161 DVK(KP813797681, +0.813797681349373692844693217248393223289101568);
162 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
163 DVK(KP296198132, +0.296198132726023843175338011893050938967728390);
164 DVK(KP642787609, +0.642787609686539326322643409907263432907559884);
165 DVK(KP663413948, +0.663413948168938396205421319635891297216863310);
166 DVK(KP556670399, +0.556670399226419366452912952047023132968291906);
167 DVK(KP766044443, +0.766044443118978035202392650555416673935832457);
168 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
169 DVK(KP150383733, +0.150383733180435296639271897612501926072238258);
170 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
171 DVK(KP173648177, +0.173648177666930348851716626769314796000375677);
172 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
173 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
174 {
175 INT i;
176 const R *xi;
177 R *xo;
178 xi = ii;
179 xo = io;
180 for (i = v; i > 0; i = i - VL, xi = xi + (VL * ivs), xo = xo + (VL * ovs), MAKE_VOLATILE_STRIDE(18, is), MAKE_VOLATILE_STRIDE(18, os)) {
181 V T5, Ty, Tm, Ti, Tw, Th, Tj, To, Tb, Tv, Ta, Tc, Tn;
182 {
183 V T1, T2, T3, T4;
184 T1 = LD(&(xi[0]), ivs, &(xi[0]));
185 T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
186 T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
187 T4 = VADD(T2, T3);
188 T5 = VFNMS(LDK(KP500000000), T4, T1);
189 Ty = VADD(T1, T4);
190 Tm = VMUL(LDK(KP866025403), VSUB(T2, T3));
191 }
192 {
193 V Td, Tg, Te, Tf;
194 Td = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
195 Te = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
196 Tf = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
197 Tg = VADD(Te, Tf);
198 Ti = VSUB(Te, Tf);
199 Tw = VADD(Td, Tg);
200 Th = VFNMS(LDK(KP500000000), Tg, Td);
201 Tj = VFNMS(LDK(KP852868531), Ti, VMUL(LDK(KP173648177), Th));
202 To = VFMA(LDK(KP150383733), Ti, VMUL(LDK(KP984807753), Th));
203 }
204 {
205 V T6, T9, T7, T8;
206 T6 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
207 T7 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
208 T8 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
209 T9 = VADD(T7, T8);
210 Tb = VSUB(T7, T8);
211 Tv = VADD(T6, T9);
212 Ta = VFNMS(LDK(KP500000000), T9, T6);
213 Tc = VFNMS(LDK(KP556670399), Tb, VMUL(LDK(KP766044443), Ta));
214 Tn = VFMA(LDK(KP663413948), Tb, VMUL(LDK(KP642787609), Ta));
215 }
216 {
217 V Tx, Tz, TA, Tt, Tu;
218 Tx = VBYI(VMUL(LDK(KP866025403), VSUB(Tv, Tw)));
219 Tz = VADD(Tv, Tw);
220 TA = VFNMS(LDK(KP500000000), Tz, Ty);
221 ST(&(xo[WS(os, 3)]), VADD(Tx, TA), ovs, &(xo[WS(os, 1)]));
222 ST(&(xo[0]), VADD(Ty, Tz), ovs, &(xo[0]));
223 ST(&(xo[WS(os, 6)]), VSUB(TA, Tx), ovs, &(xo[0]));
224 Tt = VFMA(LDK(KP852868531), Tb, VFMA(LDK(KP173648177), Ta, VFMA(LDK(KP296198132), Ti, VFNMS(LDK(KP939692620), Th, T5))));
225 Tu = VBYI(VSUB(VFMA(LDK(KP984807753), Ta, VFMA(LDK(KP813797681), Ti, VFNMS(LDK(KP150383733), Tb, VMUL(LDK(KP342020143), Th)))), Tm));
226 ST(&(xo[WS(os, 7)]), VSUB(Tt, Tu), ovs, &(xo[WS(os, 1)]));
227 ST(&(xo[WS(os, 2)]), VADD(Tt, Tu), ovs, &(xo[0]));
228 {
229 V Tl, Ts, Tq, Tr, Tk, Tp;
230 Tk = VADD(Tc, Tj);
231 Tl = VADD(T5, Tk);
232 Ts = VFMA(LDK(KP866025403), VSUB(To, Tn), VFNMS(LDK(KP500000000), Tk, T5));
233 Tp = VADD(Tn, To);
234 Tq = VBYI(VADD(Tm, Tp));
235 Tr = VBYI(VADD(Tm, VFNMS(LDK(KP500000000), Tp, VMUL(LDK(KP866025403), VSUB(Tc, Tj)))));
236 ST(&(xo[WS(os, 8)]), VSUB(Tl, Tq), ovs, &(xo[0]));
237 ST(&(xo[WS(os, 5)]), VSUB(Ts, Tr), ovs, &(xo[WS(os, 1)]));
238 ST(&(xo[WS(os, 1)]), VADD(Tl, Tq), ovs, &(xo[WS(os, 1)]));
239 ST(&(xo[WS(os, 4)]), VADD(Tr, Ts), ovs, &(xo[0]));
240 }
241 }
242 }
243 }
244 VLEAVE();
245 }
246
247 static const kdft_desc desc = { 9, XSIMD_STRING("n1bv_9"), {30, 10, 16, 0}, &GENUS, 0, 0, 0, 0 };
248
249 void XSIMD(codelet_n1bv_9) (planner *p) {
250 X(kdft_register) (p, n1bv_9, &desc);
251 }
252
253 #endif /* HAVE_FMA */