Mercurial > hg > sv-dependency-builds

comparison src/fftw-3.3.8/dft/simd/common/t1buv_9.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 08:05:57 EDT 2018 */
23
24 #include "dft/codelet-dft.h"
25
26 #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
27
28 /* Generated by: ../../../genfft/gen_twiddle_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -n 9 -name t1buv_9 -include dft/simd/t1bu.h -sign 1 */
29
30 /*
31 * This function contains 54 FP additions, 54 FP multiplications,
32 * (or, 20 additions, 20 multiplications, 34 fused multiply/add),
33 * 50 stack variables, 19 constants, and 18 memory accesses
34 */
35 #include "dft/simd/t1bu.h"
36
37 static void t1buv_9(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
38 {
39 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
40 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
41 DVK(KP879385241, +0.879385241571816768108218554649462939872416269);
42 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
43 DVK(KP666666666, +0.666666666666666666666666666666666666666666667);
44 DVK(KP673648177, +0.673648177666930348851716626769314796000375677);
45 DVK(KP898197570, +0.898197570222573798468955502359086394667167570);
46 DVK(KP826351822, +0.826351822333069651148283373230685203999624323);
47 DVK(KP420276625, +0.420276625461206169731530603237061658838781920);
48 DVK(KP907603734, +0.907603734547952313649323976213898122064543220);
49 DVK(KP347296355, +0.347296355333860697703433253538629592000751354);
50 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
51 DVK(KP152703644, +0.152703644666139302296566746461370407999248646);
52 DVK(KP968908795, +0.968908795874236621082202410917456709164223497);
53 DVK(KP203604859, +0.203604859554852403062088995281827210665664861);
54 DVK(KP726681596, +0.726681596905677465811651808188092531873167623);
55 DVK(KP439692620, +0.439692620785908384054109277324731469936208134);
56 DVK(KP586256827, +0.586256827714544512072145703099641959914944179);
57 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
58 {
59 INT m;
60 R *x;
61 x = ii;
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, T6, Tx, TO, TP, Tf, Tp, Tk, Tl, Tq, Tu, TD, TC, TA, Tz;
64 T1 = LD(&(x[0]), ms, &(x[0]));
65 {
66 V T3, T5, T2, T4;
67 T2 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
68 T3 = BYTW(&(W[TWVL * 4]), T2);
69 T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
70 T5 = BYTW(&(W[TWVL * 10]), T4);
71 T6 = VADD(T3, T5);
72 Tx = VSUB(T3, T5);
73 }
74 {
75 V T9, Tn, Tb, Td, Te, Th, Tj, To, T8, Tm;
76 T8 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
77 T9 = BYTW(&(W[TWVL * 2]), T8);
78 Tm = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
79 Tn = BYTW(&(W[0]), Tm);
80 {
81 V Ta, Tc, Tg, Ti;
82 Ta = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
83 Tb = BYTW(&(W[TWVL * 8]), Ta);
84 Tc = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
85 Td = BYTW(&(W[TWVL * 14]), Tc);
86 Te = VADD(Tb, Td);
87 Tg = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
88 Th = BYTW(&(W[TWVL * 6]), Tg);
89 Ti = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
90 Tj = BYTW(&(W[TWVL * 12]), Ti);
91 To = VADD(Th, Tj);
92 }
93 TO = VADD(Tn, To);
94 TP = VADD(T9, Te);
95 Tf = VFNMS(LDK(KP500000000), Te, T9);
96 Tp = VFNMS(LDK(KP500000000), To, Tn);
97 Tk = VSUB(Th, Tj);
98 Tl = VSUB(Td, Tb);
99 Tq = VFNMS(LDK(KP586256827), Tp, Tl);
100 Tu = VFNMS(LDK(KP439692620), Tk, Tf);
101 TD = VFNMS(LDK(KP726681596), Tk, Tp);
102 TC = VFMA(LDK(KP203604859), Tf, Tl);
103 TA = VFMA(LDK(KP968908795), Tp, Tk);
104 Tz = VFNMS(LDK(KP152703644), Tl, Tf);
105 }
106 {
107 V TS, TN, TQ, TR;
108 TS = VMUL(LDK(KP866025403), VSUB(TO, TP));
109 TN = VADD(T1, T6);
110 TQ = VADD(TO, TP);
111 TR = VFNMS(LDK(KP500000000), TQ, TN);
112 ST(&(x[WS(rs, 3)]), VFMAI(TS, TR), ms, &(x[WS(rs, 1)]));
113 ST(&(x[0]), VADD(TQ, TN), ms, &(x[0]));
114 ST(&(x[WS(rs, 6)]), VFNMSI(TS, TR), ms, &(x[0]));
115 }
116 {
117 V Ts, Tw, TJ, TM, T7, TF, TL, Tr, Tv;
118 Tr = VFNMS(LDK(KP347296355), Tq, Tk);
119 Ts = VFNMS(LDK(KP907603734), Tr, Tf);
120 Tv = VFNMS(LDK(KP420276625), Tu, Tl);
121 Tw = VFNMS(LDK(KP826351822), Tv, Tp);
122 {
123 V TH, TI, TE, TB;
124 TH = VFNMS(LDK(KP898197570), TD, TC);
125 TI = VFMA(LDK(KP673648177), TA, Tz);
126 TJ = VFMA(LDK(KP666666666), TI, TH);
127 TM = VMUL(LDK(KP984807753), VFMA(LDK(KP879385241), Tx, TI));
128 T7 = VFNMS(LDK(KP500000000), T6, T1);
129 TE = VFMA(LDK(KP898197570), TD, TC);
130 TB = VFNMS(LDK(KP673648177), TA, Tz);
131 TF = VFNMS(LDK(KP500000000), TE, TB);
132 TL = VFMA(LDK(KP852868531), TE, T7);
133 }
134 ST(&(x[WS(rs, 1)]), VFMAI(TM, TL), ms, &(x[WS(rs, 1)]));
135 ST(&(x[WS(rs, 8)]), VFNMSI(TM, TL), ms, &(x[0]));
136 {
137 V Tt, Ty, TG, TK;
138 Tt = VFNMS(LDK(KP939692620), Ts, T7);
139 Ty = VMUL(LDK(KP984807753), VFNMS(LDK(KP879385241), Tx, Tw));
140 ST(&(x[WS(rs, 7)]), VFNMSI(Ty, Tt), ms, &(x[WS(rs, 1)]));
141 ST(&(x[WS(rs, 2)]), VFMAI(Ty, Tt), ms, &(x[0]));
142 TG = VFMA(LDK(KP852868531), TF, T7);
143 TK = VMUL(LDK(KP866025403), VFNMS(LDK(KP852868531), TJ, Tx));
144 ST(&(x[WS(rs, 4)]), VFMAI(TK, TG), ms, &(x[0]));
145 ST(&(x[WS(rs, 5)]), VFNMSI(TK, TG), ms, &(x[WS(rs, 1)]));
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("t1buv_9"), twinstr, &GENUS, {20, 20, 34, 0}, 0, 0, 0 };
166
167 void XSIMD(codelet_t1buv_9) (planner *p) {
168 X(kdft_dit_register) (p, t1buv_9, &desc);
169 }
170 #else
171
172 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 9 -name t1buv_9 -include dft/simd/t1bu.h -sign 1 */
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 "dft/simd/t1bu.h"
180
181 static void t1buv_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(KP150383733, +0.150383733180435296639271897612501926072238258);
192 DVK(KP342020143, +0.342020143325668733044099614682259580763083368);
193 DVK(KP813797681, +0.813797681349373692844693217248393223289101568);
194 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
195 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
196 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
197 {
198 INT m;
199 R *x;
200 x = ii;
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, Tu, Tg, Tf, TD, Tq, Tp, TE;
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 = BYTW(&(W[TWVL * 4]), T2);
208 T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
209 T5 = BYTW(&(W[TWVL * 10]), T4);
210 T6 = VADD(T3, T5);
211 Tu = VMUL(LDK(KP866025403), VSUB(T3, T5));
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 = BYTW(&(W[0]), T8);
217 Tc = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
218 Td = BYTW(&(W[TWVL * 12]), Tc);
219 Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
220 Tb = BYTW(&(W[TWVL * 6]), Ta);
221 Tg = VSUB(Tb, Td);
222 Te = VADD(Tb, Td);
223 Tf = VFNMS(LDK(KP500000000), Te, T9);
224 TD = VADD(T9, Te);
225 }
226 {
227 V Tj, Tn, Tl, Ti, Tm, Tk, To;
228 Ti = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
229 Tj = BYTW(&(W[TWVL * 2]), Ti);
230 Tm = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
231 Tn = BYTW(&(W[TWVL * 14]), Tm);
232 Tk = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
233 Tl = BYTW(&(W[TWVL * 8]), Tk);
234 Tq = VSUB(Tl, Tn);
235 To = VADD(Tl, Tn);
236 Tp = VFNMS(LDK(KP500000000), To, Tj);
237 TE = VADD(Tj, To);
238 }
239 {
240 V TF, TG, TH, TI;
241 TF = VBYI(VMUL(LDK(KP866025403), VSUB(TD, TE)));
242 TG = VADD(T1, T6);
243 TH = VADD(TD, TE);
244 TI = VFNMS(LDK(KP500000000), TH, TG);
245 ST(&(x[WS(rs, 3)]), VADD(TF, TI), ms, &(x[WS(rs, 1)]));
246 ST(&(x[0]), VADD(TG, TH), ms, &(x[0]));
247 ST(&(x[WS(rs, 6)]), VSUB(TI, TF), ms, &(x[0]));
248 }
249 {
250 V TC, Tv, Tw, Tx, Th, Tr, Ts, T7, TB;
251 TC = VBYI(VSUB(VFMA(LDK(KP984807753), Tf, VFMA(LDK(KP813797681), Tq, VFNMS(LDK(KP150383733), Tg, VMUL(LDK(KP342020143), Tp)))), Tu));
252 Tv = VFMA(LDK(KP663413948), Tg, VMUL(LDK(KP642787609), Tf));
253 Tw = VFMA(LDK(KP150383733), Tq, VMUL(LDK(KP984807753), Tp));
254 Tx = VADD(Tv, Tw);
255 Th = VFNMS(LDK(KP556670399), Tg, VMUL(LDK(KP766044443), Tf));
256 Tr = VFNMS(LDK(KP852868531), Tq, VMUL(LDK(KP173648177), Tp));
257 Ts = VADD(Th, Tr);
258 T7 = VFNMS(LDK(KP500000000), T6, T1);
259 TB = VFMA(LDK(KP852868531), Tg, VFMA(LDK(KP173648177), Tf, VFMA(LDK(KP296198132), Tq, VFNMS(LDK(KP939692620), Tp, T7))));
260 ST(&(x[WS(rs, 7)]), VSUB(TB, TC), ms, &(x[WS(rs, 1)]));
261 ST(&(x[WS(rs, 2)]), VADD(TB, TC), ms, &(x[0]));
262 {
263 V Tt, Ty, Tz, TA;
264 Tt = VADD(T7, Ts);
265 Ty = VBYI(VADD(Tu, Tx));
266 ST(&(x[WS(rs, 8)]), VSUB(Tt, Ty), ms, &(x[0]));
267 ST(&(x[WS(rs, 1)]), VADD(Tt, Ty), ms, &(x[WS(rs, 1)]));
268 Tz = VBYI(VADD(Tu, VFNMS(LDK(KP500000000), Tx, VMUL(LDK(KP866025403), VSUB(Th, Tr)))));
269 TA = VFMA(LDK(KP866025403), VSUB(Tw, Tv), VFNMS(LDK(KP500000000), Ts, T7));
270 ST(&(x[WS(rs, 4)]), VADD(Tz, TA), ms, &(x[0]));
271 ST(&(x[WS(rs, 5)]), VSUB(TA, Tz), ms, &(x[WS(rs, 1)]));
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("t1buv_9"), twinstr, &GENUS, {38, 26, 16, 0}, 0, 0, 0 };
292
293 void XSIMD(codelet_t1buv_9) (planner *p) {
294 X(kdft_dit_register) (p, t1buv_9, &desc);
295 }
296 #endif