Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/n2sv_8.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 16:41:20 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.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 8 -name n2sv_8 -with-ostride 1 -include n2s.h -store-multiple 4 */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 52 FP additions, 8 FP multiplications,
Chris@42 32 * (or, 44 additions, 0 multiplications, 8 fused multiply/add),
Chris@42 33 * 58 stack variables, 1 constants, and 36 memory accesses
Chris@42 34 */
Chris@42 35 #include "n2s.h"
Chris@42 36
Chris@42 37 static void n2sv_8(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(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 40 {
Chris@42 41 INT i;
Chris@42 42 for (i = v; i > 0; i = i - (2 * VL), ri = ri + ((2 * VL) * ivs), ii = ii + ((2 * VL) * ivs), ro = ro + ((2 * VL) * ovs), io = io + ((2 * VL) * ovs), MAKE_VOLATILE_STRIDE(32, is), MAKE_VOLATILE_STRIDE(32, os)) {
Chris@42 43 V TF, TJ, TD, TR, TS, TT, TU, TV, TW, TE, TX, TY, TK, TI, TZ;
Chris@42 44 V T10, T11, T12;
Chris@42 45 {
Chris@42 46 V Tb, Tn, T3, TC, Ti, TB, T6, To, Tl, Tc, Tw, Tx, T8, T9, Tr;
Chris@42 47 V Ts;
Chris@42 48 {
Chris@42 49 V T1, T2, Tg, Th, T4, T5, Tj, Tk;
Chris@42 50 T1 = LD(&(ri[0]), ivs, &(ri[0]));
Chris@42 51 T2 = LD(&(ri[WS(is, 4)]), ivs, &(ri[0]));
Chris@42 52 Tg = LD(&(ii[0]), ivs, &(ii[0]));
Chris@42 53 Th = LD(&(ii[WS(is, 4)]), ivs, &(ii[0]));
Chris@42 54 T4 = LD(&(ri[WS(is, 2)]), ivs, &(ri[0]));
Chris@42 55 T5 = LD(&(ri[WS(is, 6)]), ivs, &(ri[0]));
Chris@42 56 Tj = LD(&(ii[WS(is, 2)]), ivs, &(ii[0]));
Chris@42 57 Tk = LD(&(ii[WS(is, 6)]), ivs, &(ii[0]));
Chris@42 58 Tb = LD(&(ri[WS(is, 7)]), ivs, &(ri[WS(is, 1)]));
Chris@42 59 Tn = VSUB(T1, T2);
Chris@42 60 T3 = VADD(T1, T2);
Chris@42 61 TC = VSUB(Tg, Th);
Chris@42 62 Ti = VADD(Tg, Th);
Chris@42 63 TB = VSUB(T4, T5);
Chris@42 64 T6 = VADD(T4, T5);
Chris@42 65 To = VSUB(Tj, Tk);
Chris@42 66 Tl = VADD(Tj, Tk);
Chris@42 67 Tc = LD(&(ri[WS(is, 3)]), ivs, &(ri[WS(is, 1)]));
Chris@42 68 Tw = LD(&(ii[WS(is, 7)]), ivs, &(ii[WS(is, 1)]));
Chris@42 69 Tx = LD(&(ii[WS(is, 3)]), ivs, &(ii[WS(is, 1)]));
Chris@42 70 T8 = LD(&(ri[WS(is, 1)]), ivs, &(ri[WS(is, 1)]));
Chris@42 71 T9 = LD(&(ri[WS(is, 5)]), ivs, &(ri[WS(is, 1)]));
Chris@42 72 Tr = LD(&(ii[WS(is, 1)]), ivs, &(ii[WS(is, 1)]));
Chris@42 73 Ts = LD(&(ii[WS(is, 5)]), ivs, &(ii[WS(is, 1)]));
Chris@42 74 }
Chris@42 75 {
Chris@42 76 V TL, T7, TP, Tm, Tz, TH, Te, Tf, TO, TQ, TG, Tu, Tp, TA;
Chris@42 77 {
Chris@42 78 V Td, Tv, TN, Ty, Ta, Tq, TM, Tt;
Chris@42 79 TL = VSUB(T3, T6);
Chris@42 80 T7 = VADD(T3, T6);
Chris@42 81 Td = VADD(Tb, Tc);
Chris@42 82 Tv = VSUB(Tb, Tc);
Chris@42 83 TN = VADD(Tw, Tx);
Chris@42 84 Ty = VSUB(Tw, Tx);
Chris@42 85 Ta = VADD(T8, T9);
Chris@42 86 Tq = VSUB(T8, T9);
Chris@42 87 TM = VADD(Tr, Ts);
Chris@42 88 Tt = VSUB(Tr, Ts);
Chris@42 89 TP = VADD(Ti, Tl);
Chris@42 90 Tm = VSUB(Ti, Tl);
Chris@42 91 Tz = VSUB(Tv, Ty);
Chris@42 92 TH = VADD(Tv, Ty);
Chris@42 93 Te = VADD(Ta, Td);
Chris@42 94 Tf = VSUB(Td, Ta);
Chris@42 95 TO = VSUB(TM, TN);
Chris@42 96 TQ = VADD(TM, TN);
Chris@42 97 TG = VSUB(Tt, Tq);
Chris@42 98 Tu = VADD(Tq, Tt);
Chris@42 99 }
Chris@42 100 TF = VSUB(Tn, To);
Chris@42 101 Tp = VADD(Tn, To);
Chris@42 102 TJ = VSUB(TC, TB);
Chris@42 103 TD = VADD(TB, TC);
Chris@42 104 TR = VSUB(Tm, Tf);
Chris@42 105 STM4(&(io[6]), TR, ovs, &(io[0]));
Chris@42 106 TS = VADD(Tf, Tm);
Chris@42 107 STM4(&(io[2]), TS, ovs, &(io[0]));
Chris@42 108 TT = VADD(T7, Te);
Chris@42 109 STM4(&(ro[0]), TT, ovs, &(ro[0]));
Chris@42 110 TU = VSUB(T7, Te);
Chris@42 111 STM4(&(ro[4]), TU, ovs, &(ro[0]));
Chris@42 112 TV = VADD(TP, TQ);
Chris@42 113 STM4(&(io[0]), TV, ovs, &(io[0]));
Chris@42 114 TW = VSUB(TP, TQ);
Chris@42 115 STM4(&(io[4]), TW, ovs, &(io[0]));
Chris@42 116 TE = VSUB(Tz, Tu);
Chris@42 117 TA = VADD(Tu, Tz);
Chris@42 118 TX = VADD(TL, TO);
Chris@42 119 STM4(&(ro[2]), TX, ovs, &(ro[0]));
Chris@42 120 TY = VSUB(TL, TO);
Chris@42 121 STM4(&(ro[6]), TY, ovs, &(ro[0]));
Chris@42 122 TK = VADD(TG, TH);
Chris@42 123 TI = VSUB(TG, TH);
Chris@42 124 TZ = VFMA(LDK(KP707106781), TA, Tp);
Chris@42 125 STM4(&(ro[1]), TZ, ovs, &(ro[1]));
Chris@42 126 T10 = VFNMS(LDK(KP707106781), TA, Tp);
Chris@42 127 STM4(&(ro[5]), T10, ovs, &(ro[1]));
Chris@42 128 }
Chris@42 129 }
Chris@42 130 T11 = VFMA(LDK(KP707106781), TK, TJ);
Chris@42 131 STM4(&(io[1]), T11, ovs, &(io[1]));
Chris@42 132 T12 = VFNMS(LDK(KP707106781), TK, TJ);
Chris@42 133 STM4(&(io[5]), T12, ovs, &(io[1]));
Chris@42 134 {
Chris@42 135 V T13, T14, T15, T16;
Chris@42 136 T13 = VFMA(LDK(KP707106781), TE, TD);
Chris@42 137 STM4(&(io[3]), T13, ovs, &(io[1]));
Chris@42 138 STN4(&(io[0]), TV, T11, TS, T13, ovs);
Chris@42 139 T14 = VFNMS(LDK(KP707106781), TE, TD);
Chris@42 140 STM4(&(io[7]), T14, ovs, &(io[1]));
Chris@42 141 STN4(&(io[4]), TW, T12, TR, T14, ovs);
Chris@42 142 T15 = VFMA(LDK(KP707106781), TI, TF);
Chris@42 143 STM4(&(ro[3]), T15, ovs, &(ro[1]));
Chris@42 144 STN4(&(ro[0]), TT, TZ, TX, T15, ovs);
Chris@42 145 T16 = VFNMS(LDK(KP707106781), TI, TF);
Chris@42 146 STM4(&(ro[7]), T16, ovs, &(ro[1]));
Chris@42 147 STN4(&(ro[4]), TU, T10, TY, T16, ovs);
Chris@42 148 }
Chris@42 149 }
Chris@42 150 }
Chris@42 151 VLEAVE();
Chris@42 152 }
Chris@42 153
Chris@42 154 static const kdft_desc desc = { 8, XSIMD_STRING("n2sv_8"), {44, 0, 8, 0}, &GENUS, 0, 1, 0, 0 };
Chris@42 155
Chris@42 156 void XSIMD(codelet_n2sv_8) (planner *p) {
Chris@42 157 X(kdft_register) (p, n2sv_8, &desc);
Chris@42 158 }
Chris@42 159
Chris@42 160 #else /* HAVE_FMA */
Chris@42 161
Chris@42 162 /* Generated by: ../../../genfft/gen_notw.native -simd -compact -variables 4 -pipeline-latency 8 -n 8 -name n2sv_8 -with-ostride 1 -include n2s.h -store-multiple 4 */
Chris@42 163
Chris@42 164 /*
Chris@42 165 * This function contains 52 FP additions, 4 FP multiplications,
Chris@42 166 * (or, 52 additions, 4 multiplications, 0 fused multiply/add),
Chris@42 167 * 34 stack variables, 1 constants, and 36 memory accesses
Chris@42 168 */
Chris@42 169 #include "n2s.h"
Chris@42 170
Chris@42 171 static void n2sv_8(const R *ri, const R *ii, R *ro, R *io, stride is, stride os, INT v, INT ivs, INT ovs)
Chris@42 172 {
Chris@42 173 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 174 {
Chris@42 175 INT i;
Chris@42 176 for (i = v; i > 0; i = i - (2 * VL), ri = ri + ((2 * VL) * ivs), ii = ii + ((2 * VL) * ivs), ro = ro + ((2 * VL) * ovs), io = io + ((2 * VL) * ovs), MAKE_VOLATILE_STRIDE(32, is), MAKE_VOLATILE_STRIDE(32, os)) {
Chris@42 177 V T3, Tn, Ti, TC, T6, TB, Tl, To, Td, TN, Tz, TH, Ta, TM, Tu;
Chris@42 178 V TG;
Chris@42 179 {
Chris@42 180 V T1, T2, Tj, Tk;
Chris@42 181 T1 = LD(&(ri[0]), ivs, &(ri[0]));
Chris@42 182 T2 = LD(&(ri[WS(is, 4)]), ivs, &(ri[0]));
Chris@42 183 T3 = VADD(T1, T2);
Chris@42 184 Tn = VSUB(T1, T2);
Chris@42 185 {
Chris@42 186 V Tg, Th, T4, T5;
Chris@42 187 Tg = LD(&(ii[0]), ivs, &(ii[0]));
Chris@42 188 Th = LD(&(ii[WS(is, 4)]), ivs, &(ii[0]));
Chris@42 189 Ti = VADD(Tg, Th);
Chris@42 190 TC = VSUB(Tg, Th);
Chris@42 191 T4 = LD(&(ri[WS(is, 2)]), ivs, &(ri[0]));
Chris@42 192 T5 = LD(&(ri[WS(is, 6)]), ivs, &(ri[0]));
Chris@42 193 T6 = VADD(T4, T5);
Chris@42 194 TB = VSUB(T4, T5);
Chris@42 195 }
Chris@42 196 Tj = LD(&(ii[WS(is, 2)]), ivs, &(ii[0]));
Chris@42 197 Tk = LD(&(ii[WS(is, 6)]), ivs, &(ii[0]));
Chris@42 198 Tl = VADD(Tj, Tk);
Chris@42 199 To = VSUB(Tj, Tk);
Chris@42 200 {
Chris@42 201 V Tb, Tc, Tv, Tw, Tx, Ty;
Chris@42 202 Tb = LD(&(ri[WS(is, 7)]), ivs, &(ri[WS(is, 1)]));
Chris@42 203 Tc = LD(&(ri[WS(is, 3)]), ivs, &(ri[WS(is, 1)]));
Chris@42 204 Tv = VSUB(Tb, Tc);
Chris@42 205 Tw = LD(&(ii[WS(is, 7)]), ivs, &(ii[WS(is, 1)]));
Chris@42 206 Tx = LD(&(ii[WS(is, 3)]), ivs, &(ii[WS(is, 1)]));
Chris@42 207 Ty = VSUB(Tw, Tx);
Chris@42 208 Td = VADD(Tb, Tc);
Chris@42 209 TN = VADD(Tw, Tx);
Chris@42 210 Tz = VSUB(Tv, Ty);
Chris@42 211 TH = VADD(Tv, Ty);
Chris@42 212 }
Chris@42 213 {
Chris@42 214 V T8, T9, Tq, Tr, Ts, Tt;
Chris@42 215 T8 = LD(&(ri[WS(is, 1)]), ivs, &(ri[WS(is, 1)]));
Chris@42 216 T9 = LD(&(ri[WS(is, 5)]), ivs, &(ri[WS(is, 1)]));
Chris@42 217 Tq = VSUB(T8, T9);
Chris@42 218 Tr = LD(&(ii[WS(is, 1)]), ivs, &(ii[WS(is, 1)]));
Chris@42 219 Ts = LD(&(ii[WS(is, 5)]), ivs, &(ii[WS(is, 1)]));
Chris@42 220 Tt = VSUB(Tr, Ts);
Chris@42 221 Ta = VADD(T8, T9);
Chris@42 222 TM = VADD(Tr, Ts);
Chris@42 223 Tu = VADD(Tq, Tt);
Chris@42 224 TG = VSUB(Tt, Tq);
Chris@42 225 }
Chris@42 226 }
Chris@42 227 {
Chris@42 228 V TR, TS, TT, TU, TV, TW, TX, TY;
Chris@42 229 {
Chris@42 230 V T7, Te, TP, TQ;
Chris@42 231 T7 = VADD(T3, T6);
Chris@42 232 Te = VADD(Ta, Td);
Chris@42 233 TR = VSUB(T7, Te);
Chris@42 234 STM4(&(ro[4]), TR, ovs, &(ro[0]));
Chris@42 235 TS = VADD(T7, Te);
Chris@42 236 STM4(&(ro[0]), TS, ovs, &(ro[0]));
Chris@42 237 TP = VADD(Ti, Tl);
Chris@42 238 TQ = VADD(TM, TN);
Chris@42 239 TT = VSUB(TP, TQ);
Chris@42 240 STM4(&(io[4]), TT, ovs, &(io[0]));
Chris@42 241 TU = VADD(TP, TQ);
Chris@42 242 STM4(&(io[0]), TU, ovs, &(io[0]));
Chris@42 243 }
Chris@42 244 {
Chris@42 245 V Tf, Tm, TL, TO;
Chris@42 246 Tf = VSUB(Td, Ta);
Chris@42 247 Tm = VSUB(Ti, Tl);
Chris@42 248 TV = VADD(Tf, Tm);
Chris@42 249 STM4(&(io[2]), TV, ovs, &(io[0]));
Chris@42 250 TW = VSUB(Tm, Tf);
Chris@42 251 STM4(&(io[6]), TW, ovs, &(io[0]));
Chris@42 252 TL = VSUB(T3, T6);
Chris@42 253 TO = VSUB(TM, TN);
Chris@42 254 TX = VSUB(TL, TO);
Chris@42 255 STM4(&(ro[6]), TX, ovs, &(ro[0]));
Chris@42 256 TY = VADD(TL, TO);
Chris@42 257 STM4(&(ro[2]), TY, ovs, &(ro[0]));
Chris@42 258 }
Chris@42 259 {
Chris@42 260 V TZ, T10, T11, T12;
Chris@42 261 {
Chris@42 262 V Tp, TA, TJ, TK;
Chris@42 263 Tp = VADD(Tn, To);
Chris@42 264 TA = VMUL(LDK(KP707106781), VADD(Tu, Tz));
Chris@42 265 TZ = VSUB(Tp, TA);
Chris@42 266 STM4(&(ro[5]), TZ, ovs, &(ro[1]));
Chris@42 267 T10 = VADD(Tp, TA);
Chris@42 268 STM4(&(ro[1]), T10, ovs, &(ro[1]));
Chris@42 269 TJ = VSUB(TC, TB);
Chris@42 270 TK = VMUL(LDK(KP707106781), VADD(TG, TH));
Chris@42 271 T11 = VSUB(TJ, TK);
Chris@42 272 STM4(&(io[5]), T11, ovs, &(io[1]));
Chris@42 273 T12 = VADD(TJ, TK);
Chris@42 274 STM4(&(io[1]), T12, ovs, &(io[1]));
Chris@42 275 }
Chris@42 276 {
Chris@42 277 V TD, TE, T13, T14;
Chris@42 278 TD = VADD(TB, TC);
Chris@42 279 TE = VMUL(LDK(KP707106781), VSUB(Tz, Tu));
Chris@42 280 T13 = VSUB(TD, TE);
Chris@42 281 STM4(&(io[7]), T13, ovs, &(io[1]));
Chris@42 282 STN4(&(io[4]), TT, T11, TW, T13, ovs);
Chris@42 283 T14 = VADD(TD, TE);
Chris@42 284 STM4(&(io[3]), T14, ovs, &(io[1]));
Chris@42 285 STN4(&(io[0]), TU, T12, TV, T14, ovs);
Chris@42 286 }
Chris@42 287 {
Chris@42 288 V TF, TI, T15, T16;
Chris@42 289 TF = VSUB(Tn, To);
Chris@42 290 TI = VMUL(LDK(KP707106781), VSUB(TG, TH));
Chris@42 291 T15 = VSUB(TF, TI);
Chris@42 292 STM4(&(ro[7]), T15, ovs, &(ro[1]));
Chris@42 293 STN4(&(ro[4]), TR, TZ, TX, T15, ovs);
Chris@42 294 T16 = VADD(TF, TI);
Chris@42 295 STM4(&(ro[3]), T16, ovs, &(ro[1]));
Chris@42 296 STN4(&(ro[0]), TS, T10, TY, T16, ovs);
Chris@42 297 }
Chris@42 298 }
Chris@42 299 }
Chris@42 300 }
Chris@42 301 }
Chris@42 302 VLEAVE();
Chris@42 303 }
Chris@42 304
Chris@42 305 static const kdft_desc desc = { 8, XSIMD_STRING("n2sv_8"), {52, 4, 0, 0}, &GENUS, 0, 1, 0, 0 };
Chris@42 306
Chris@42 307 void XSIMD(codelet_n2sv_8) (planner *p) {
Chris@42 308 X(kdft_register) (p, n2sv_8, &desc);
Chris@42 309 }
Chris@42 310
Chris@42 311 #endif /* HAVE_FMA */