Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/t1fuv_9.c @ 84:08ae793730bd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Add null config files
author Chris Cannam
date Mon, 02 Mar 2020 14:03:47 +0000
parents 2cd0e3b3e1fd
children
rev   line source
Chris@42 1 /*
Chris@42 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@42 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@42 4 *
Chris@42 5 * This program is free software; you can redistribute it and/or modify
Chris@42 6 * it under the terms of the GNU General Public License as published by
Chris@42 7 * the Free Software Foundation; either version 2 of the License, or
Chris@42 8 * (at your option) any later version.
Chris@42 9 *
Chris@42 10 * This program is distributed in the hope that it will be useful,
Chris@42 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@42 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@42 13 * GNU General Public License for more details.
Chris@42 14 *
Chris@42 15 * You should have received a copy of the GNU General Public License
Chris@42 16 * along with this program; if not, write to the Free Software
Chris@42 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@42 18 *
Chris@42 19 */
Chris@42 20
Chris@42 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@42 22 /* Generated on Sat Jul 30 16:41:51 EDT 2016 */
Chris@42 23
Chris@42 24 #include "codelet-dft.h"
Chris@42 25
Chris@42 26 #ifdef HAVE_FMA
Chris@42 27
Chris@42 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 */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 54 FP additions, 54 FP multiplications,
Chris@42 32 * (or, 20 additions, 20 multiplications, 34 fused multiply/add),
Chris@42 33 * 67 stack variables, 19 constants, and 18 memory accesses
Chris@42 34 */
Chris@42 35 #include "t1fu.h"
Chris@42 36
Chris@42 37 static void t1fuv_9(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 38 {
Chris@42 39 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
Chris@42 40 DVK(KP826351822, +0.826351822333069651148283373230685203999624323);
Chris@42 41 DVK(KP879385241, +0.879385241571816768108218554649462939872416269);
Chris@42 42 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@42 43 DVK(KP666666666, +0.666666666666666666666666666666666666666666667);
Chris@42 44 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
Chris@42 45 DVK(KP907603734, +0.907603734547952313649323976213898122064543220);
Chris@42 46 DVK(KP420276625, +0.420276625461206169731530603237061658838781920);
Chris@42 47 DVK(KP673648177, +0.673648177666930348851716626769314796000375677);
Chris@42 48 DVK(KP898197570, +0.898197570222573798468955502359086394667167570);
Chris@42 49 DVK(KP347296355, +0.347296355333860697703433253538629592000751354);
Chris@42 50 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 51 DVK(KP439692620, +0.439692620785908384054109277324731469936208134);
Chris@42 52 DVK(KP203604859, +0.203604859554852403062088995281827210665664861);
Chris@42 53 DVK(KP152703644, +0.152703644666139302296566746461370407999248646);
Chris@42 54 DVK(KP586256827, +0.586256827714544512072145703099641959914944179);
Chris@42 55 DVK(KP968908795, +0.968908795874236621082202410917456709164223497);
Chris@42 56 DVK(KP726681596, +0.726681596905677465811651808188092531873167623);
Chris@42 57 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 58 {
Chris@42 59 INT m;
Chris@42 60 R *x;
Chris@42 61 x = ri;
Chris@42 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)) {
Chris@42 63 V T1, T3, T5, T9, Th, Tb, Td, Tj, Tl, TD, T6;
Chris@42 64 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@42 65 {
Chris@42 66 V T2, T4, T8, Tg;
Chris@42 67 T2 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@42 68 T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@42 69 T8 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 70 Tg = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 71 {
Chris@42 72 V Ta, Tc, Ti, Tk;
Chris@42 73 Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@42 74 Tc = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@42 75 Ti = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@42 76 Tk = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@42 77 T3 = BYTWJ(&(W[TWVL * 4]), T2);
Chris@42 78 T5 = BYTWJ(&(W[TWVL * 10]), T4);
Chris@42 79 T9 = BYTWJ(&(W[0]), T8);
Chris@42 80 Th = BYTWJ(&(W[TWVL * 2]), Tg);
Chris@42 81 Tb = BYTWJ(&(W[TWVL * 6]), Ta);
Chris@42 82 Td = BYTWJ(&(W[TWVL * 12]), Tc);
Chris@42 83 Tj = BYTWJ(&(W[TWVL * 8]), Ti);
Chris@42 84 Tl = BYTWJ(&(W[TWVL * 14]), Tk);
Chris@42 85 }
Chris@42 86 }
Chris@42 87 TD = VSUB(T5, T3);
Chris@42 88 T6 = VADD(T3, T5);
Chris@42 89 {
Chris@42 90 V Tt, Te, Tu, Tm, Tr, T7;
Chris@42 91 Tt = VSUB(Tb, Td);
Chris@42 92 Te = VADD(Tb, Td);
Chris@42 93 Tu = VSUB(Tl, Tj);
Chris@42 94 Tm = VADD(Tj, Tl);
Chris@42 95 Tr = VFNMS(LDK(KP500000000), T6, T1);
Chris@42 96 T7 = VADD(T1, T6);
Chris@42 97 {
Chris@42 98 V Tv, Tf, Ts, Tn;
Chris@42 99 Tv = VFNMS(LDK(KP500000000), Te, T9);
Chris@42 100 Tf = VADD(T9, Te);
Chris@42 101 Ts = VFNMS(LDK(KP500000000), Tm, Th);
Chris@42 102 Tn = VADD(Th, Tm);
Chris@42 103 {
Chris@42 104 V TG, TK, Tw, TJ, TF, TA, To, Tq;
Chris@42 105 TG = VFNMS(LDK(KP726681596), Tt, Tv);
Chris@42 106 TK = VFMA(LDK(KP968908795), Tv, Tt);
Chris@42 107 Tw = VFNMS(LDK(KP586256827), Tv, Tu);
Chris@42 108 TJ = VFNMS(LDK(KP152703644), Tu, Ts);
Chris@42 109 TF = VFMA(LDK(KP203604859), Ts, Tu);
Chris@42 110 TA = VFNMS(LDK(KP439692620), Tt, Ts);
Chris@42 111 To = VADD(Tf, Tn);
Chris@42 112 Tq = VMUL(LDK(KP866025403), VSUB(Tn, Tf));
Chris@42 113 {
Chris@42 114 V TQ, TH, TL, TN, TB, Tp, Ty, TI, Tx;
Chris@42 115 Tx = VFNMS(LDK(KP347296355), Tw, Tt);
Chris@42 116 TQ = VFNMS(LDK(KP898197570), TG, TF);
Chris@42 117 TH = VFMA(LDK(KP898197570), TG, TF);
Chris@42 118 TL = VFMA(LDK(KP673648177), TK, TJ);
Chris@42 119 TN = VFNMS(LDK(KP673648177), TK, TJ);
Chris@42 120 TB = VFNMS(LDK(KP420276625), TA, Tu);
Chris@42 121 ST(&(x[0]), VADD(T7, To), ms, &(x[0]));
Chris@42 122 Tp = VFNMS(LDK(KP500000000), To, T7);
Chris@42 123 Ty = VFNMS(LDK(KP907603734), Tx, Ts);
Chris@42 124 TI = VFMA(LDK(KP852868531), TH, Tr);
Chris@42 125 {
Chris@42 126 V TO, TR, TM, TC, Tz, TP, TS, TE;
Chris@42 127 TO = VFNMS(LDK(KP500000000), TH, TN);
Chris@42 128 TR = VFMA(LDK(KP666666666), TL, TQ);
Chris@42 129 TM = VMUL(LDK(KP984807753), VFNMS(LDK(KP879385241), TD, TL));
Chris@42 130 TC = VFNMS(LDK(KP826351822), TB, Tv);
Chris@42 131 ST(&(x[WS(rs, 6)]), VFNMSI(Tq, Tp), ms, &(x[0]));
Chris@42 132 ST(&(x[WS(rs, 3)]), VFMAI(Tq, Tp), ms, &(x[WS(rs, 1)]));
Chris@42 133 Tz = VFNMS(LDK(KP939692620), Ty, Tr);
Chris@42 134 TP = VFMA(LDK(KP852868531), TO, Tr);
Chris@42 135 TS = VMUL(LDK(KP866025403), VFMA(LDK(KP852868531), TR, TD));
Chris@42 136 ST(&(x[WS(rs, 8)]), VFMAI(TM, TI), ms, &(x[0]));
Chris@42 137 ST(&(x[WS(rs, 1)]), VFNMSI(TM, TI), ms, &(x[WS(rs, 1)]));
Chris@42 138 TE = VMUL(LDK(KP984807753), VFMA(LDK(KP879385241), TD, TC));
Chris@42 139 ST(&(x[WS(rs, 4)]), VFMAI(TS, TP), ms, &(x[0]));
Chris@42 140 ST(&(x[WS(rs, 5)]), VFNMSI(TS, TP), ms, &(x[WS(rs, 1)]));
Chris@42 141 ST(&(x[WS(rs, 7)]), VFMAI(TE, Tz), ms, &(x[WS(rs, 1)]));
Chris@42 142 ST(&(x[WS(rs, 2)]), VFNMSI(TE, Tz), ms, &(x[0]));
Chris@42 143 }
Chris@42 144 }
Chris@42 145 }
Chris@42 146 }
Chris@42 147 }
Chris@42 148 }
Chris@42 149 }
Chris@42 150 VLEAVE();
Chris@42 151 }
Chris@42 152
Chris@42 153 static const tw_instr twinstr[] = {
Chris@42 154 VTW(0, 1),
Chris@42 155 VTW(0, 2),
Chris@42 156 VTW(0, 3),
Chris@42 157 VTW(0, 4),
Chris@42 158 VTW(0, 5),
Chris@42 159 VTW(0, 6),
Chris@42 160 VTW(0, 7),
Chris@42 161 VTW(0, 8),
Chris@42 162 {TW_NEXT, VL, 0}
Chris@42 163 };
Chris@42 164
Chris@42 165 static const ct_desc desc = { 9, XSIMD_STRING("t1fuv_9"), twinstr, &GENUS, {20, 20, 34, 0}, 0, 0, 0 };
Chris@42 166
Chris@42 167 void XSIMD(codelet_t1fuv_9) (planner *p) {
Chris@42 168 X(kdft_dit_register) (p, t1fuv_9, &desc);
Chris@42 169 }
Chris@42 170 #else /* HAVE_FMA */
Chris@42 171
Chris@42 172 /* Generated by: ../../../genfft/gen_twiddle_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 9 -name t1fuv_9 -include t1fu.h */
Chris@42 173
Chris@42 174 /*
Chris@42 175 * This function contains 54 FP additions, 42 FP multiplications,
Chris@42 176 * (or, 38 additions, 26 multiplications, 16 fused multiply/add),
Chris@42 177 * 38 stack variables, 14 constants, and 18 memory accesses
Chris@42 178 */
Chris@42 179 #include "t1fu.h"
Chris@42 180
Chris@42 181 static void t1fuv_9(R *ri, R *ii, const R *W, stride rs, INT mb, INT me, INT ms)
Chris@42 182 {
Chris@42 183 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
Chris@42 184 DVK(KP296198132, +0.296198132726023843175338011893050938967728390);
Chris@42 185 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
Chris@42 186 DVK(KP173648177, +0.173648177666930348851716626769314796000375677);
Chris@42 187 DVK(KP556670399, +0.556670399226419366452912952047023132968291906);
Chris@42 188 DVK(KP766044443, +0.766044443118978035202392650555416673935832457);
Chris@42 189 DVK(KP642787609, +0.642787609686539326322643409907263432907559884);
Chris@42 190 DVK(KP663413948, +0.663413948168938396205421319635891297216863310);
Chris@42 191 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@42 192 DVK(KP150383733, +0.150383733180435296639271897612501926072238258);
Chris@42 193 DVK(KP342020143, +0.342020143325668733044099614682259580763083368);
Chris@42 194 DVK(KP813797681, +0.813797681349373692844693217248393223289101568);
Chris@42 195 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 196 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 197 {
Chris@42 198 INT m;
Chris@42 199 R *x;
Chris@42 200 x = ri;
Chris@42 201 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)) {
Chris@42 202 V T1, T6, TA, Tt, Tf, Ts, Tw, Tn, Tv;
Chris@42 203 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@42 204 {
Chris@42 205 V T3, T5, T2, T4;
Chris@42 206 T2 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@42 207 T3 = BYTWJ(&(W[TWVL * 4]), T2);
Chris@42 208 T4 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@42 209 T5 = BYTWJ(&(W[TWVL * 10]), T4);
Chris@42 210 T6 = VADD(T3, T5);
Chris@42 211 TA = VMUL(LDK(KP866025403), VSUB(T5, T3));
Chris@42 212 }
Chris@42 213 {
Chris@42 214 V T9, Td, Tb, T8, Tc, Ta, Te;
Chris@42 215 T8 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 216 T9 = BYTWJ(&(W[0]), T8);
Chris@42 217 Tc = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@42 218 Td = BYTWJ(&(W[TWVL * 12]), Tc);
Chris@42 219 Ta = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@42 220 Tb = BYTWJ(&(W[TWVL * 6]), Ta);
Chris@42 221 Tt = VSUB(Td, Tb);
Chris@42 222 Te = VADD(Tb, Td);
Chris@42 223 Tf = VADD(T9, Te);
Chris@42 224 Ts = VFNMS(LDK(KP500000000), Te, T9);
Chris@42 225 }
Chris@42 226 {
Chris@42 227 V Th, Tl, Tj, Tg, Tk, Ti, Tm;
Chris@42 228 Tg = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 229 Th = BYTWJ(&(W[TWVL * 2]), Tg);
Chris@42 230 Tk = LD(&(x[WS(rs, 8)]), ms, &(x[0]));
Chris@42 231 Tl = BYTWJ(&(W[TWVL * 14]), Tk);
Chris@42 232 Ti = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@42 233 Tj = BYTWJ(&(W[TWVL * 8]), Ti);
Chris@42 234 Tw = VSUB(Tl, Tj);
Chris@42 235 Tm = VADD(Tj, Tl);
Chris@42 236 Tn = VADD(Th, Tm);
Chris@42 237 Tv = VFNMS(LDK(KP500000000), Tm, Th);
Chris@42 238 }
Chris@42 239 {
Chris@42 240 V Tq, T7, To, Tp;
Chris@42 241 Tq = VBYI(VMUL(LDK(KP866025403), VSUB(Tn, Tf)));
Chris@42 242 T7 = VADD(T1, T6);
Chris@42 243 To = VADD(Tf, Tn);
Chris@42 244 Tp = VFNMS(LDK(KP500000000), To, T7);
Chris@42 245 ST(&(x[0]), VADD(T7, To), ms, &(x[0]));
Chris@42 246 ST(&(x[WS(rs, 3)]), VADD(Tp, Tq), ms, &(x[WS(rs, 1)]));
Chris@42 247 ST(&(x[WS(rs, 6)]), VSUB(Tp, Tq), ms, &(x[0]));
Chris@42 248 }
Chris@42 249 {
Chris@42 250 V TI, TB, TC, TD, Tu, Tx, Ty, Tr, TH;
Chris@42 251 TI = VBYI(VSUB(VFNMS(LDK(KP342020143), Tv, VFNMS(LDK(KP150383733), Tt, VFNMS(LDK(KP984807753), Ts, VMUL(LDK(KP813797681), Tw)))), TA));
Chris@42 252 TB = VFNMS(LDK(KP642787609), Ts, VMUL(LDK(KP663413948), Tt));
Chris@42 253 TC = VFNMS(LDK(KP984807753), Tv, VMUL(LDK(KP150383733), Tw));
Chris@42 254 TD = VADD(TB, TC);
Chris@42 255 Tu = VFMA(LDK(KP766044443), Ts, VMUL(LDK(KP556670399), Tt));
Chris@42 256 Tx = VFMA(LDK(KP173648177), Tv, VMUL(LDK(KP852868531), Tw));
Chris@42 257 Ty = VADD(Tu, Tx);
Chris@42 258 Tr = VFNMS(LDK(KP500000000), T6, T1);
Chris@42 259 TH = VFMA(LDK(KP173648177), Ts, VFNMS(LDK(KP296198132), Tw, VFNMS(LDK(KP939692620), Tv, VFNMS(LDK(KP852868531), Tt, Tr))));
Chris@42 260 ST(&(x[WS(rs, 7)]), VSUB(TH, TI), ms, &(x[WS(rs, 1)]));
Chris@42 261 ST(&(x[WS(rs, 2)]), VADD(TH, TI), ms, &(x[0]));
Chris@42 262 {
Chris@42 263 V Tz, TE, TF, TG;
Chris@42 264 Tz = VADD(Tr, Ty);
Chris@42 265 TE = VBYI(VADD(TA, TD));
Chris@42 266 ST(&(x[WS(rs, 8)]), VSUB(Tz, TE), ms, &(x[0]));
Chris@42 267 ST(&(x[WS(rs, 1)]), VADD(TE, Tz), ms, &(x[WS(rs, 1)]));
Chris@42 268 TF = VFMA(LDK(KP866025403), VSUB(TB, TC), VFNMS(LDK(KP500000000), Ty, Tr));
Chris@42 269 TG = VBYI(VADD(TA, VFNMS(LDK(KP500000000), TD, VMUL(LDK(KP866025403), VSUB(Tx, Tu)))));
Chris@42 270 ST(&(x[WS(rs, 5)]), VSUB(TF, TG), ms, &(x[WS(rs, 1)]));
Chris@42 271 ST(&(x[WS(rs, 4)]), VADD(TF, TG), ms, &(x[0]));
Chris@42 272 }
Chris@42 273 }
Chris@42 274 }
Chris@42 275 }
Chris@42 276 VLEAVE();
Chris@42 277 }
Chris@42 278
Chris@42 279 static const tw_instr twinstr[] = {
Chris@42 280 VTW(0, 1),
Chris@42 281 VTW(0, 2),
Chris@42 282 VTW(0, 3),
Chris@42 283 VTW(0, 4),
Chris@42 284 VTW(0, 5),
Chris@42 285 VTW(0, 6),
Chris@42 286 VTW(0, 7),
Chris@42 287 VTW(0, 8),
Chris@42 288 {TW_NEXT, VL, 0}
Chris@42 289 };
Chris@42 290
Chris@42 291 static const ct_desc desc = { 9, XSIMD_STRING("t1fuv_9"), twinstr, &GENUS, {38, 26, 16, 0}, 0, 0, 0 };
Chris@42 292
Chris@42 293 void XSIMD(codelet_t1fuv_9) (planner *p) {
Chris@42 294 X(kdft_dit_register) (p, t1fuv_9, &desc);
Chris@42 295 }
Chris@42 296 #endif /* HAVE_FMA */