Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.8/dft/simd/common/n1bv_9.c @ 82:d0c2a83c1364

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
date Tue, 19 Nov 2019 14:52:55 +0000
parents
children
rev   line source
Chris@82 1 /*
Chris@82 2 * Copyright (c) 2003, 2007-14 Matteo Frigo
Chris@82 3 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
Chris@82 4 *
Chris@82 5 * This program is free software; you can redistribute it and/or modify
Chris@82 6 * it under the terms of the GNU General Public License as published by
Chris@82 7 * the Free Software Foundation; either version 2 of the License, or
Chris@82 8 * (at your option) any later version.
Chris@82 9 *
Chris@82 10 * This program is distributed in the hope that it will be useful,
Chris@82 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
Chris@82 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Chris@82 13 * GNU General Public License for more details.
Chris@82 14 *
Chris@82 15 * You should have received a copy of the GNU General Public License
Chris@82 16 * along with this program; if not, write to the Free Software
Chris@82 17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
Chris@82 18 *
Chris@82 19 */
Chris@82 20
Chris@82 21 /* This file was automatically generated --- DO NOT EDIT */
Chris@82 22 /* Generated on Thu May 24 08:04:55 EDT 2018 */
Chris@82 23
Chris@82 24 #include "dft/codelet-dft.h"
Chris@82 25
Chris@82 26 #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA)
Chris@82 27
Chris@82 28 /* Generated by: ../../../genfft/gen_notw_c.native -fma -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 9 -name n1bv_9 -include dft/simd/n1b.h */
Chris@82 29
Chris@82 30 /*
Chris@82 31 * This function contains 46 FP additions, 38 FP multiplications,
Chris@82 32 * (or, 12 additions, 4 multiplications, 34 fused multiply/add),
Chris@82 33 * 50 stack variables, 19 constants, and 18 memory accesses
Chris@82 34 */
Chris@82 35 #include "dft/simd/n1b.h"
Chris@82 36
Chris@82 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@82 38 {
Chris@82 39 DVK(KP666666666, +0.666666666666666666666666666666666666666666667);
Chris@82 40 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
Chris@82 41 DVK(KP898197570, +0.898197570222573798468955502359086394667167570);
Chris@82 42 DVK(KP673648177, +0.673648177666930348851716626769314796000375677);
Chris@82 43 DVK(KP879385241, +0.879385241571816768108218554649462939872416269);
Chris@82 44 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@82 45 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
Chris@82 46 DVK(KP826351822, +0.826351822333069651148283373230685203999624323);
Chris@82 47 DVK(KP420276625, +0.420276625461206169731530603237061658838781920);
Chris@82 48 DVK(KP907603734, +0.907603734547952313649323976213898122064543220);
Chris@82 49 DVK(KP347296355, +0.347296355333860697703433253538629592000751354);
Chris@82 50 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82 51 DVK(KP968908795, +0.968908795874236621082202410917456709164223497);
Chris@82 52 DVK(KP726681596, +0.726681596905677465811651808188092531873167623);
Chris@82 53 DVK(KP586256827, +0.586256827714544512072145703099641959914944179);
Chris@82 54 DVK(KP152703644, +0.152703644666139302296566746461370407999248646);
Chris@82 55 DVK(KP203604859, +0.203604859554852403062088995281827210665664861);
Chris@82 56 DVK(KP439692620, +0.439692620785908384054109277324731469936208134);
Chris@82 57 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 58 {
Chris@82 59 INT i;
Chris@82 60 const R *xi;
Chris@82 61 R *xo;
Chris@82 62 xi = ii;
Chris@82 63 xo = io;
Chris@82 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@82 65 V T5, TF, Tp, Te, Td, TG, TH, Ta, Tm, Tu, Tr, Th, Ti, Tv, Ts;
Chris@82 66 V TK, TI, TJ;
Chris@82 67 {
Chris@82 68 V T1, T2, T3, T4;
Chris@82 69 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82 70 T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82 71 T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@82 72 T4 = VADD(T2, T3);
Chris@82 73 T5 = VFNMS(LDK(KP500000000), T4, T1);
Chris@82 74 TF = VADD(T1, T4);
Chris@82 75 Tp = VSUB(T2, T3);
Chris@82 76 }
Chris@82 77 {
Chris@82 78 V T6, Tf, T9, Tg;
Chris@82 79 T6 = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82 80 Tf = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82 81 {
Chris@82 82 V T7, T8, Tb, Tc;
Chris@82 83 T7 = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@82 84 T8 = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@82 85 T9 = VADD(T7, T8);
Chris@82 86 Te = VSUB(T8, T7);
Chris@82 87 Tb = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@82 88 Tc = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@82 89 Td = VSUB(Tb, Tc);
Chris@82 90 Tg = VADD(Tb, Tc);
Chris@82 91 }
Chris@82 92 TG = VADD(Tf, Tg);
Chris@82 93 TH = VADD(T6, T9);
Chris@82 94 Ta = VFNMS(LDK(KP500000000), T9, T6);
Chris@82 95 Tm = VFNMS(LDK(KP439692620), Td, Ta);
Chris@82 96 Tu = VFMA(LDK(KP203604859), Ta, Te);
Chris@82 97 Tr = VFNMS(LDK(KP152703644), Te, Ta);
Chris@82 98 Th = VFNMS(LDK(KP500000000), Tg, Tf);
Chris@82 99 Ti = VFNMS(LDK(KP586256827), Th, Te);
Chris@82 100 Tv = VFNMS(LDK(KP726681596), Td, Th);
Chris@82 101 Ts = VFMA(LDK(KP968908795), Th, Td);
Chris@82 102 }
Chris@82 103 TK = VMUL(LDK(KP866025403), VSUB(TG, TH));
Chris@82 104 TI = VADD(TG, TH);
Chris@82 105 TJ = VFNMS(LDK(KP500000000), TI, TF);
Chris@82 106 ST(&(xo[WS(os, 3)]), VFMAI(TK, TJ), ovs, &(xo[WS(os, 1)]));
Chris@82 107 ST(&(xo[0]), VADD(TI, TF), ovs, &(xo[0]));
Chris@82 108 ST(&(xo[WS(os, 6)]), VFNMSI(TK, TJ), ovs, &(xo[0]));
Chris@82 109 {
Chris@82 110 V Tk, To, Tj, Tn, Tl, Tq;
Chris@82 111 Tj = VFNMS(LDK(KP347296355), Ti, Td);
Chris@82 112 Tk = VFNMS(LDK(KP907603734), Tj, Ta);
Chris@82 113 Tn = VFNMS(LDK(KP420276625), Tm, Te);
Chris@82 114 To = VFNMS(LDK(KP826351822), Tn, Th);
Chris@82 115 Tl = VFNMS(LDK(KP939692620), Tk, T5);
Chris@82 116 Tq = VMUL(LDK(KP984807753), VFNMS(LDK(KP879385241), Tp, To));
Chris@82 117 ST(&(xo[WS(os, 7)]), VFNMSI(Tq, Tl), ovs, &(xo[WS(os, 1)]));
Chris@82 118 ST(&(xo[WS(os, 2)]), VFMAI(Tq, Tl), ovs, &(xo[0]));
Chris@82 119 }
Chris@82 120 {
Chris@82 121 V Tx, TD, TB, TE, Ty, TC;
Chris@82 122 {
Chris@82 123 V Tt, Tw, Tz, TA;
Chris@82 124 Tt = VFNMS(LDK(KP673648177), Ts, Tr);
Chris@82 125 Tw = VFMA(LDK(KP898197570), Tv, Tu);
Chris@82 126 Tx = VFNMS(LDK(KP500000000), Tw, Tt);
Chris@82 127 TD = VFMA(LDK(KP852868531), Tw, T5);
Chris@82 128 Tz = VFNMS(LDK(KP898197570), Tv, Tu);
Chris@82 129 TA = VFMA(LDK(KP673648177), Ts, Tr);
Chris@82 130 TB = VFMA(LDK(KP666666666), TA, Tz);
Chris@82 131 TE = VMUL(LDK(KP984807753), VFMA(LDK(KP879385241), Tp, TA));
Chris@82 132 }
Chris@82 133 ST(&(xo[WS(os, 1)]), VFMAI(TE, TD), ovs, &(xo[WS(os, 1)]));
Chris@82 134 ST(&(xo[WS(os, 8)]), VFNMSI(TE, TD), ovs, &(xo[0]));
Chris@82 135 Ty = VFMA(LDK(KP852868531), Tx, T5);
Chris@82 136 TC = VMUL(LDK(KP866025403), VFNMS(LDK(KP852868531), TB, Tp));
Chris@82 137 ST(&(xo[WS(os, 4)]), VFMAI(TC, Ty), ovs, &(xo[0]));
Chris@82 138 ST(&(xo[WS(os, 5)]), VFNMSI(TC, Ty), ovs, &(xo[WS(os, 1)]));
Chris@82 139 }
Chris@82 140 }
Chris@82 141 }
Chris@82 142 VLEAVE();
Chris@82 143 }
Chris@82 144
Chris@82 145 static const kdft_desc desc = { 9, XSIMD_STRING("n1bv_9"), {12, 4, 34, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 146
Chris@82 147 void XSIMD(codelet_n1bv_9) (planner *p) {
Chris@82 148 X(kdft_register) (p, n1bv_9, &desc);
Chris@82 149 }
Chris@82 150
Chris@82 151 #else
Chris@82 152
Chris@82 153 /* Generated by: ../../../genfft/gen_notw_c.native -simd -compact -variables 4 -pipeline-latency 8 -sign 1 -n 9 -name n1bv_9 -include dft/simd/n1b.h */
Chris@82 154
Chris@82 155 /*
Chris@82 156 * This function contains 46 FP additions, 26 FP multiplications,
Chris@82 157 * (or, 30 additions, 10 multiplications, 16 fused multiply/add),
Chris@82 158 * 41 stack variables, 14 constants, and 18 memory accesses
Chris@82 159 */
Chris@82 160 #include "dft/simd/n1b.h"
Chris@82 161
Chris@82 162 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@82 163 {
Chris@82 164 DVK(KP342020143, +0.342020143325668733044099614682259580763083368);
Chris@82 165 DVK(KP813797681, +0.813797681349373692844693217248393223289101568);
Chris@82 166 DVK(KP939692620, +0.939692620785908384054109277324731469936208134);
Chris@82 167 DVK(KP296198132, +0.296198132726023843175338011893050938967728390);
Chris@82 168 DVK(KP642787609, +0.642787609686539326322643409907263432907559884);
Chris@82 169 DVK(KP663413948, +0.663413948168938396205421319635891297216863310);
Chris@82 170 DVK(KP556670399, +0.556670399226419366452912952047023132968291906);
Chris@82 171 DVK(KP766044443, +0.766044443118978035202392650555416673935832457);
Chris@82 172 DVK(KP984807753, +0.984807753012208059366743024589523013670643252);
Chris@82 173 DVK(KP150383733, +0.150383733180435296639271897612501926072238258);
Chris@82 174 DVK(KP852868531, +0.852868531952443209628250963940074071936020296);
Chris@82 175 DVK(KP173648177, +0.173648177666930348851716626769314796000375677);
Chris@82 176 DVK(KP866025403, +0.866025403784438646763723170752936183471402627);
Chris@82 177 DVK(KP500000000, +0.500000000000000000000000000000000000000000000);
Chris@82 178 {
Chris@82 179 INT i;
Chris@82 180 const R *xi;
Chris@82 181 R *xo;
Chris@82 182 xi = ii;
Chris@82 183 xo = io;
Chris@82 184 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@82 185 V T5, Ty, Tm, Ti, Tw, Th, Tj, To, Tb, Tv, Ta, Tc, Tn;
Chris@82 186 {
Chris@82 187 V T1, T2, T3, T4;
Chris@82 188 T1 = LD(&(xi[0]), ivs, &(xi[0]));
Chris@82 189 T2 = LD(&(xi[WS(is, 3)]), ivs, &(xi[WS(is, 1)]));
Chris@82 190 T3 = LD(&(xi[WS(is, 6)]), ivs, &(xi[0]));
Chris@82 191 T4 = VADD(T2, T3);
Chris@82 192 T5 = VFNMS(LDK(KP500000000), T4, T1);
Chris@82 193 Ty = VADD(T1, T4);
Chris@82 194 Tm = VMUL(LDK(KP866025403), VSUB(T2, T3));
Chris@82 195 }
Chris@82 196 {
Chris@82 197 V Td, Tg, Te, Tf;
Chris@82 198 Td = LD(&(xi[WS(is, 2)]), ivs, &(xi[0]));
Chris@82 199 Te = LD(&(xi[WS(is, 5)]), ivs, &(xi[WS(is, 1)]));
Chris@82 200 Tf = LD(&(xi[WS(is, 8)]), ivs, &(xi[0]));
Chris@82 201 Tg = VADD(Te, Tf);
Chris@82 202 Ti = VSUB(Te, Tf);
Chris@82 203 Tw = VADD(Td, Tg);
Chris@82 204 Th = VFNMS(LDK(KP500000000), Tg, Td);
Chris@82 205 Tj = VFNMS(LDK(KP852868531), Ti, VMUL(LDK(KP173648177), Th));
Chris@82 206 To = VFMA(LDK(KP150383733), Ti, VMUL(LDK(KP984807753), Th));
Chris@82 207 }
Chris@82 208 {
Chris@82 209 V T6, T9, T7, T8;
Chris@82 210 T6 = LD(&(xi[WS(is, 1)]), ivs, &(xi[WS(is, 1)]));
Chris@82 211 T7 = LD(&(xi[WS(is, 4)]), ivs, &(xi[0]));
Chris@82 212 T8 = LD(&(xi[WS(is, 7)]), ivs, &(xi[WS(is, 1)]));
Chris@82 213 T9 = VADD(T7, T8);
Chris@82 214 Tb = VSUB(T7, T8);
Chris@82 215 Tv = VADD(T6, T9);
Chris@82 216 Ta = VFNMS(LDK(KP500000000), T9, T6);
Chris@82 217 Tc = VFNMS(LDK(KP556670399), Tb, VMUL(LDK(KP766044443), Ta));
Chris@82 218 Tn = VFMA(LDK(KP663413948), Tb, VMUL(LDK(KP642787609), Ta));
Chris@82 219 }
Chris@82 220 {
Chris@82 221 V Tx, Tz, TA, Tt, Tu;
Chris@82 222 Tx = VBYI(VMUL(LDK(KP866025403), VSUB(Tv, Tw)));
Chris@82 223 Tz = VADD(Tv, Tw);
Chris@82 224 TA = VFNMS(LDK(KP500000000), Tz, Ty);
Chris@82 225 ST(&(xo[WS(os, 3)]), VADD(Tx, TA), ovs, &(xo[WS(os, 1)]));
Chris@82 226 ST(&(xo[0]), VADD(Ty, Tz), ovs, &(xo[0]));
Chris@82 227 ST(&(xo[WS(os, 6)]), VSUB(TA, Tx), ovs, &(xo[0]));
Chris@82 228 Tt = VFMA(LDK(KP852868531), Tb, VFMA(LDK(KP173648177), Ta, VFMA(LDK(KP296198132), Ti, VFNMS(LDK(KP939692620), Th, T5))));
Chris@82 229 Tu = VBYI(VSUB(VFMA(LDK(KP984807753), Ta, VFMA(LDK(KP813797681), Ti, VFNMS(LDK(KP150383733), Tb, VMUL(LDK(KP342020143), Th)))), Tm));
Chris@82 230 ST(&(xo[WS(os, 7)]), VSUB(Tt, Tu), ovs, &(xo[WS(os, 1)]));
Chris@82 231 ST(&(xo[WS(os, 2)]), VADD(Tt, Tu), ovs, &(xo[0]));
Chris@82 232 {
Chris@82 233 V Tl, Ts, Tq, Tr, Tk, Tp;
Chris@82 234 Tk = VADD(Tc, Tj);
Chris@82 235 Tl = VADD(T5, Tk);
Chris@82 236 Ts = VFMA(LDK(KP866025403), VSUB(To, Tn), VFNMS(LDK(KP500000000), Tk, T5));
Chris@82 237 Tp = VADD(Tn, To);
Chris@82 238 Tq = VBYI(VADD(Tm, Tp));
Chris@82 239 Tr = VBYI(VADD(Tm, VFNMS(LDK(KP500000000), Tp, VMUL(LDK(KP866025403), VSUB(Tc, Tj)))));
Chris@82 240 ST(&(xo[WS(os, 8)]), VSUB(Tl, Tq), ovs, &(xo[0]));
Chris@82 241 ST(&(xo[WS(os, 5)]), VSUB(Ts, Tr), ovs, &(xo[WS(os, 1)]));
Chris@82 242 ST(&(xo[WS(os, 1)]), VADD(Tl, Tq), ovs, &(xo[WS(os, 1)]));
Chris@82 243 ST(&(xo[WS(os, 4)]), VADD(Tr, Ts), ovs, &(xo[0]));
Chris@82 244 }
Chris@82 245 }
Chris@82 246 }
Chris@82 247 }
Chris@82 248 VLEAVE();
Chris@82 249 }
Chris@82 250
Chris@82 251 static const kdft_desc desc = { 9, XSIMD_STRING("n1bv_9"), {30, 10, 16, 0}, &GENUS, 0, 0, 0, 0 };
Chris@82 252
Chris@82 253 void XSIMD(codelet_n1bv_9) (planner *p) {
Chris@82 254 X(kdft_register) (p, n1bv_9, &desc);
Chris@82 255 }
Chris@82 256
Chris@82 257 #endif