Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/n1bv_9.c @ 79:91c729825bca pa_catalina

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Update build for AUDIO_COMPONENT_FIX
author Chris Cannam
date Wed, 30 Oct 2019 12:40:34 +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 16:39:03 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_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 */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 46 FP additions, 38 FP multiplications,
Chris@42 32 * (or, 12 additions, 4 multiplications, 34 fused multiply/add),
Chris@42 33 * 68 stack variables, 19 constants, and 18 memory accesses
Chris@42 34 */
Chris@42 35 #include "n1b.h"
Chris@42 36
Chris@42 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)
Chris@42 38 {
Chris@42 39 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
Chris@42 40 DVK(KP907603734, +0.907603734547952313649323976213898122064543220);
Chris@42 41 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
Chris@42 42 DVK(KP666666666, +0.666666666666666666666666666666666666666666667);
Chris@42 43 DVK(KP879385241, +0.879385241571816768108218554649462939872416269);
Chris@42 44 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@42 45 DVK(KP826351822, +0.826351822333069651148283373230685203999624323);
Chris@42 46 DVK(KP347296355, +0.347296355333860697703433253538629592000751354);
Chris@42 47 DVK(KP898197570, +0.898197570222573798468955502359086394667167570);
Chris@42 48 DVK(KP673648177, +0.673648177666930348851716626769314796000375677);
Chris@42 49 DVK(KP420276625, +0.420276625461206169731530603237061658838781920);
Chris@42 50 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 51 DVK(KP586256827, +0.586256827714544512072145703099641959914944179);
Chris@42 52 DVK(KP968908795, +0.968908795874236621082202410917456709164223497);
Chris@42 53 DVK(KP726681596, +0.726681596905677465811651808188092531873167623);
Chris@42 54 DVK(KP439692620, +0.439692620785908384054109277324731469936208134);
Chris@42 55 DVK(KP203604859, +0.203604859554852403062088995281827210665664861);
Chris@42 56 DVK(KP152703644, +0.152703644666139302296566746461370407999248646);
Chris@42 57 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 58 {
Chris@42 59 INT i;
Chris@42 60 const R *xi;
Chris@42 61 R *xo;
Chris@42 62 xi = ii;
Chris@42 63 xo = io;
Chris@42 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)) {
Chris@42 65 V T1, T2, T3, T6, Tf, T7, T8, Tb, Tc, Tp, T4;
Chris@42 66 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@42 67 T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@42 68 T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@42 69 T6 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@42 70 Tf = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@42 71 T7 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@42 72 T8 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@42 73 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@42 74 Tc = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@42 75 Tp = VSUB(T2, T3);
Chris@42 76 T4 = VADD(T2, T3);
Chris@42 77 {
Chris@42 78 V Te, T9, Tg, Td, TF, T5;
Chris@42 79 Te = VSUB(T8, T7);
Chris@42 80 T9 = VADD(T7, T8);
Chris@42 81 Tg = VADD(Tb, Tc);
Chris@42 82 Td = VSUB(Tb, Tc);
Chris@42 83 TF = VADD(T1, T4);
Chris@42 84 T5 = VFNMS(LDK(KP500000000), T4, T1);
Chris@42 85 {
Chris@42 86 V Ta, TH, Th, TG;
Chris@42 87 Ta = VFNMS(LDK(KP500000000), T9, T6);
Chris@42 88 TH = VADD(T6, T9);
Chris@42 89 Th = VFNMS(LDK(KP500000000), Tg, Tf);
Chris@42 90 TG = VADD(Tf, Tg);
Chris@42 91 {
Chris@42 92 V Tr, Tu, Tm, Tv, Ts, Ti, TI, TK;
Chris@42 93 Tr = VFNMS(LDK(KP152703644), Te, Ta);
Chris@42 94 Tu = VFMA(LDK(KP203604859), Ta, Te);
Chris@42 95 Tm = VFNMS(LDK(KP439692620), Td, Ta);
Chris@42 96 Tv = VFNMS(LDK(KP726681596), Td, Th);
Chris@42 97 Ts = VFMA(LDK(KP968908795), Th, Td);
Chris@42 98 Ti = VFNMS(LDK(KP586256827), Th, Te);
Chris@42 99 TI = VADD(TG, TH);
Chris@42 100 TK = VMUL(LDK(KP866025403), VSUB(TG, TH));
Chris@42 101 {
Chris@42 102 V Tt, TA, Tw, Tz, Tj, TJ, To, TE, Tn;
Chris@42 103 Tn = VFNMS(LDK(KP420276625), Tm, Te);
Chris@42 104 Tt = VFNMS(LDK(KP673648177), Ts, Tr);
Chris@42 105 TA = VFMA(LDK(KP673648177), Ts, Tr);
Chris@42 106 Tw = VFMA(LDK(KP898197570), Tv, Tu);
Chris@42 107 Tz = VFNMS(LDK(KP898197570), Tv, Tu);
Chris@42 108 Tj = VFNMS(LDK(KP347296355), Ti, Td);
Chris@42 109 ST(&(xo[0]), VADD(TI, TF), ovs, &(xo[0]));
Chris@42 110 TJ = VFNMS(LDK(KP500000000), TI, TF);
Chris@42 111 To = VFNMS(LDK(KP826351822), Tn, Th);
Chris@42 112 TE = VMUL(LDK(KP984807753), VFMA(LDK(KP879385241), Tp, TA));
Chris@42 113 {
Chris@42 114 V TB, TD, Tx, Tk, Tq, TC, Ty, Tl;
Chris@42 115 TB = VFMA(LDK(KP666666666), TA, Tz);
Chris@42 116 TD = VFMA(LDK(KP852868531), Tw, T5);
Chris@42 117 Tx = VFNMS(LDK(KP500000000), Tw, Tt);
Chris@42 118 Tk = VFNMS(LDK(KP907603734), Tj, Ta);
Chris@42 119 ST(&(xo[WS(os, 6)]), VFNMSI(TK, TJ), ovs, &(xo[0]));
Chris@42 120 ST(&(xo[WS(os, 3)]), VFMAI(TK, TJ), ovs, &(xo[WS(os, 1)]));
Chris@42 121 Tq = VMUL(LDK(KP984807753), VFNMS(LDK(KP879385241), Tp, To));
Chris@42 122 TC = VMUL(LDK(KP866025403), VFNMS(LDK(KP852868531), TB, Tp));
Chris@42 123 ST(&(xo[WS(os, 8)]), VFNMSI(TE, TD), ovs, &(xo[0]));
Chris@42 124 ST(&(xo[WS(os, 1)]), VFMAI(TE, TD), ovs, &(xo[WS(os, 1)]));
Chris@42 125 Ty = VFMA(LDK(KP852868531), Tx, T5);
Chris@42 126 Tl = VFNMS(LDK(KP939692620), Tk, T5);
Chris@42 127 ST(&(xo[WS(os, 5)]), VFNMSI(TC, Ty), ovs, &(xo[WS(os, 1)]));
Chris@42 128 ST(&(xo[WS(os, 4)]), VFMAI(TC, Ty), ovs, &(xo[0]));
Chris@42 129 ST(&(xo[WS(os, 2)]), VFMAI(Tq, Tl), ovs, &(xo[0]));
Chris@42 130 ST(&(xo[WS(os, 7)]), VFNMSI(Tq, Tl), ovs, &(xo[WS(os, 1)]));
Chris@42 131 }
Chris@42 132 }
Chris@42 133 }
Chris@42 134 }
Chris@42 135 }
Chris@42 136 }
Chris@42 137 }
Chris@42 138 VLEAVE();
Chris@42 139 }
Chris@42 140
Chris@42 141 static const kdft_desc desc = { 9, XSIMD_STRING("n1bv_9"), {12, 4, 34, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 142
Chris@42 143 void XSIMD(codelet_n1bv_9) (planner *p) {
Chris@42 144 X(kdft_register) (p, n1bv_9, &desc);
Chris@42 145 }
Chris@42 146
Chris@42 147 #else /* HAVE_FMA */
Chris@42 148
Chris@42 149 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 9 -name n1bv_9 -include n1b.h */
Chris@42 150
Chris@42 151 /*
Chris@42 152 * This function contains 46 FP additions, 26 FP multiplications,
Chris@42 153 * (or, 30 additions, 10 multiplications, 16 fused multiply/add),
Chris@42 154 * 41 stack variables, 14 constants, and 18 memory accesses
Chris@42 155 */
Chris@42 156 #include "n1b.h"
Chris@42 157
Chris@42 158 static void n1bv_9(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@42 159 {
Chris@42 160 DVK(KP342020143, +0.342020143325668733044099614682259580763083368);
Chris@42 161 DVK(KP813797681, +0.813797681349373692844693217248393223289101568);
Chris@42 162 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
Chris@42 163 DVK(KP296198132, +0.296198132726023843175338011893050938967728390);
Chris@42 164 DVK(KP642787609, +0.642787609686539326322643409907263432907559884);
Chris@42 165 DVK(KP663413948, +0.663413948168938396205421319635891297216863310);
Chris@42 166 DVK(KP556670399, +0.556670399226419366452912952047023132968291906);
Chris@42 167 DVK(KP766044443, +0.766044443118978035202392650555416673935832457);
Chris@42 168 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@42 169 DVK(KP150383733, +0.150383733180435296639271897612501926072238258);
Chris@42 170 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
Chris@42 171 DVK(KP173648177, +0.173648177666930348851716626769314796000375677);
Chris@42 172 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@42 173 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@42 174 {
Chris@42 175 INT i;
Chris@42 176 const R *xi;
Chris@42 177 R *xo;
Chris@42 178 xi = ii;
Chris@42 179 xo = io;
Chris@42 180 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)) {
Chris@42 181 V T5, Ty, Tm, Ti, Tw, Th, Tj, To, Tb, Tv, Ta, Tc, Tn;
Chris@42 182 {
Chris@42 183 V T1, T2, T3, T4;
Chris@42 184 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@42 185 T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@42 186 T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@42 187 T4 = VADD(T2, T3);
Chris@42 188 T5 = VFNMS(LDK(KP500000000), T4, T1);
Chris@42 189 Ty = VADD(T1, T4);
Chris@42 190 Tm = VMUL(LDK(KP866025403), VSUB(T2, T3));
Chris@42 191 }
Chris@42 192 {
Chris@42 193 V Td, Tg, Te, Tf;
Chris@42 194 Td = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@42 195 Te = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@42 196 Tf = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@42 197 Tg = VADD(Te, Tf);
Chris@42 198 Ti = VSUB(Te, Tf);
Chris@42 199 Tw = VADD(Td, Tg);
Chris@42 200 Th = VFNMS(LDK(KP500000000), Tg, Td);
Chris@42 201 Tj = VFNMS(LDK(KP852868531), Ti, VMUL(LDK(KP173648177), Th));
Chris@42 202 To = VFMA(LDK(KP150383733), Ti, VMUL(LDK(KP984807753), Th));
Chris@42 203 }
Chris@42 204 {
Chris@42 205 V T6, T9, T7, T8;
Chris@42 206 T6 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@42 207 T7 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@42 208 T8 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@42 209 T9 = VADD(T7, T8);
Chris@42 210 Tb = VSUB(T7, T8);
Chris@42 211 Tv = VADD(T6, T9);
Chris@42 212 Ta = VFNMS(LDK(KP500000000), T9, T6);
Chris@42 213 Tc = VFNMS(LDK(KP556670399), Tb, VMUL(LDK(KP766044443), Ta));
Chris@42 214 Tn = VFMA(LDK(KP663413948), Tb, VMUL(LDK(KP642787609), Ta));
Chris@42 215 }
Chris@42 216 {
Chris@42 217 V Tx, Tz, TA, Tt, Tu;
Chris@42 218 Tx = VBYI(VMUL(LDK(KP866025403), VSUB(Tv, Tw)));
Chris@42 219 Tz = VADD(Tv, Tw);
Chris@42 220 TA = VFNMS(LDK(KP500000000), Tz, Ty);
Chris@42 221 ST(&(xo[WS(os, 3)]), VADD(Tx, TA), ovs, &(xo[WS(os, 1)]));
Chris@42 222 ST(&(xo[0]), VADD(Ty, Tz), ovs, &(xo[0]));
Chris@42 223 ST(&(xo[WS(os, 6)]), VSUB(TA, Tx), ovs, &(xo[0]));
Chris@42 224 Tt = VFMA(LDK(KP852868531), Tb, VFMA(LDK(KP173648177), Ta, VFMA(LDK(KP296198132), Ti, VFNMS(LDK(KP939692620), Th, T5))));
Chris@42 225 Tu = VBYI(VSUB(VFMA(LDK(KP984807753), Ta, VFMA(LDK(KP813797681), Ti, VFNMS(LDK(KP150383733), Tb, VMUL(LDK(KP342020143), Th)))), Tm));
Chris@42 226 ST(&(xo[WS(os, 7)]), VSUB(Tt, Tu), ovs, &(xo[WS(os, 1)]));
Chris@42 227 ST(&(xo[WS(os, 2)]), VADD(Tt, Tu), ovs, &(xo[0]));
Chris@42 228 {
Chris@42 229 V Tl, Ts, Tq, Tr, Tk, Tp;
Chris@42 230 Tk = VADD(Tc, Tj);
Chris@42 231 Tl = VADD(T5, Tk);
Chris@42 232 Ts = VFMA(LDK(KP866025403), VSUB(To, Tn), VFNMS(LDK(KP500000000), Tk, T5));
Chris@42 233 Tp = VADD(Tn, To);
Chris@42 234 Tq = VBYI(VADD(Tm, Tp));
Chris@42 235 Tr = VBYI(VADD(Tm, VFNMS(LDK(KP500000000), Tp, VMUL(LDK(KP866025403), VSUB(Tc, Tj)))));
Chris@42 236 ST(&(xo[WS(os, 8)]), VSUB(Tl, Tq), ovs, &(xo[0]));
Chris@42 237 ST(&(xo[WS(os, 5)]), VSUB(Ts, Tr), ovs, &(xo[WS(os, 1)]));
Chris@42 238 ST(&(xo[WS(os, 1)]), VADD(Tl, Tq), ovs, &(xo[WS(os, 1)]));
Chris@42 239 ST(&(xo[WS(os, 4)]), VADD(Tr, Ts), ovs, &(xo[0]));
Chris@42 240 }
Chris@42 241 }
Chris@42 242 }
Chris@42 243 }
Chris@42 244 VLEAVE();
Chris@42 245 }
Chris@42 246
Chris@42 247 static const kdft_desc desc = { 9, XSIMD_STRING("n1bv_9"), {30, 10, 16, 0}, &GENUS, 0, 0, 0, 0 };
Chris@42 248
Chris@42 249 void XSIMD(codelet_n1bv_9) (planner *p) {
Chris@42 250 X(kdft_register) (p, n1bv_9, &desc);
Chris@42 251 }
Chris@42 252
Chris@42 253 #endif /* HAVE_FMA */