Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/q1fv_8.c @ 42:2cd0e3b3e1fd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Current fftw source
author Chris Cannam
date Tue, 18 Oct 2016 13:40:26 +0100
parents
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 16:45:26 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_twidsq_c.native -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -n 8 -dif -name q1fv_8 -include q1f.h */
Chris@42 29
Chris@42 30 /*
Chris@42 31 * This function contains 264 FP additions, 192 FP multiplications,
Chris@42 32 * (or, 184 additions, 112 multiplications, 80 fused multiply/add),
Chris@42 33 * 117 stack variables, 1 constants, and 128 memory accesses
Chris@42 34 */
Chris@42 35 #include "q1f.h"
Chris@42 36
Chris@42 37 static void q1fv_8(R *ri, R *ii, const R *W, stride rs, stride vs, INT mb, INT me, INT ms)
Chris@42 38 {
Chris@42 39 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 40 {
Chris@42 41 INT m;
Chris@42 42 R *x;
Chris@42 43 x = ri;
Chris@42 44 for (m = mb, W = W + (mb * ((TWVL / VL) * 14)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 14), MAKE_VOLATILE_STRIDE(16, rs), MAKE_VOLATILE_STRIDE(16, vs)) {
Chris@42 45 V T42, T43, T1U, T1V, T2Y, T2Z, TT, TS;
Chris@42 46 {
Chris@42 47 V T3, Te, T1E, T1P, Tu, Tp, T25, T20, T2b, T2m, T3M, T2x, T2C, T3X, TA;
Chris@42 48 V TL, T48, T4d, T17, T11, TW, T1i, T2I, T1y, T1t, T2T, T3f, T3q, T34, T39;
Chris@42 49 V T3G, T3B, Ts, Tv, Tf, Ta, T23, T26, T1Q, T1L, T2A, T2D, T2n, T2i, T4b;
Chris@42 50 V T4e, T3Y, T3T, TZ, T12, TM, TH, T35, T2L, T3j, T1w, T1z, T1j, T1e, T36;
Chris@42 51 V T2O, T3C, T3i, T3k;
Chris@42 52 {
Chris@42 53 V T3d, T32, T3e, T3o, T3p, T33;
Chris@42 54 {
Chris@42 55 V T2v, T2w, T3V, T46, T3W;
Chris@42 56 {
Chris@42 57 V T1, T2, Tc, Td, T1C, T1D, T1N, T1O;
Chris@42 58 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@42 59 T2 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@42 60 Tc = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 61 Td = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@42 62 T1C = LD(&(x[WS(vs, 3)]), ms, &(x[WS(vs, 3)]));
Chris@42 63 T1D = LD(&(x[WS(vs, 3) + WS(rs, 4)]), ms, &(x[WS(vs, 3)]));
Chris@42 64 T1N = LD(&(x[WS(vs, 3) + WS(rs, 2)]), ms, &(x[WS(vs, 3)]));
Chris@42 65 T1O = LD(&(x[WS(vs, 3) + WS(rs, 6)]), ms, &(x[WS(vs, 3)]));
Chris@42 66 {
Chris@42 67 V T29, T1Y, T1Z, T2a, T2k, T2l, Tn, To, T3K, T3L;
Chris@42 68 T29 = LD(&(x[WS(vs, 4)]), ms, &(x[WS(vs, 4)]));
Chris@42 69 T3 = VSUB(T1, T2);
Chris@42 70 Tn = VADD(T1, T2);
Chris@42 71 Te = VSUB(Tc, Td);
Chris@42 72 To = VADD(Tc, Td);
Chris@42 73 T1E = VSUB(T1C, T1D);
Chris@42 74 T1Y = VADD(T1C, T1D);
Chris@42 75 T1P = VSUB(T1N, T1O);
Chris@42 76 T1Z = VADD(T1N, T1O);
Chris@42 77 T2a = LD(&(x[WS(vs, 4) + WS(rs, 4)]), ms, &(x[WS(vs, 4)]));
Chris@42 78 T2k = LD(&(x[WS(vs, 4) + WS(rs, 2)]), ms, &(x[WS(vs, 4)]));
Chris@42 79 T2l = LD(&(x[WS(vs, 4) + WS(rs, 6)]), ms, &(x[WS(vs, 4)]));
Chris@42 80 Tu = VSUB(Tn, To);
Chris@42 81 Tp = VADD(Tn, To);
Chris@42 82 T3K = LD(&(x[WS(vs, 7)]), ms, &(x[WS(vs, 7)]));
Chris@42 83 T3L = LD(&(x[WS(vs, 7) + WS(rs, 4)]), ms, &(x[WS(vs, 7)]));
Chris@42 84 T25 = VSUB(T1Y, T1Z);
Chris@42 85 T20 = VADD(T1Y, T1Z);
Chris@42 86 T2v = VADD(T29, T2a);
Chris@42 87 T2b = VSUB(T29, T2a);
Chris@42 88 T2w = VADD(T2k, T2l);
Chris@42 89 T2m = VSUB(T2k, T2l);
Chris@42 90 T3V = LD(&(x[WS(vs, 7) + WS(rs, 2)]), ms, &(x[WS(vs, 7)]));
Chris@42 91 T46 = VADD(T3K, T3L);
Chris@42 92 T3M = VSUB(T3K, T3L);
Chris@42 93 T3W = LD(&(x[WS(vs, 7) + WS(rs, 6)]), ms, &(x[WS(vs, 7)]));
Chris@42 94 }
Chris@42 95 }
Chris@42 96 {
Chris@42 97 V T15, TU, T16, T1g, TV, T1h;
Chris@42 98 {
Chris@42 99 V Ty, Tz, TJ, TK, T47;
Chris@42 100 Ty = LD(&(x[WS(vs, 1)]), ms, &(x[WS(vs, 1)]));
Chris@42 101 Tz = LD(&(x[WS(vs, 1) + WS(rs, 4)]), ms, &(x[WS(vs, 1)]));
Chris@42 102 TJ = LD(&(x[WS(vs, 1) + WS(rs, 2)]), ms, &(x[WS(vs, 1)]));
Chris@42 103 T2x = VADD(T2v, T2w);
Chris@42 104 T2C = VSUB(T2v, T2w);
Chris@42 105 TK = LD(&(x[WS(vs, 1) + WS(rs, 6)]), ms, &(x[WS(vs, 1)]));
Chris@42 106 T47 = VADD(T3V, T3W);
Chris@42 107 T3X = VSUB(T3V, T3W);
Chris@42 108 T15 = LD(&(x[WS(vs, 2)]), ms, &(x[WS(vs, 2)]));
Chris@42 109 TA = VSUB(Ty, Tz);
Chris@42 110 TU = VADD(Ty, Tz);
Chris@42 111 T16 = LD(&(x[WS(vs, 2) + WS(rs, 4)]), ms, &(x[WS(vs, 2)]));
Chris@42 112 T1g = LD(&(x[WS(vs, 2) + WS(rs, 2)]), ms, &(x[WS(vs, 2)]));
Chris@42 113 TL = VSUB(TJ, TK);
Chris@42 114 TV = VADD(TJ, TK);
Chris@42 115 T48 = VADD(T46, T47);
Chris@42 116 T4d = VSUB(T46, T47);
Chris@42 117 T1h = LD(&(x[WS(vs, 2) + WS(rs, 6)]), ms, &(x[WS(vs, 2)]));
Chris@42 118 }
Chris@42 119 {
Chris@42 120 V T2G, T1r, T2H, T2R, T1s, T2S;
Chris@42 121 T2G = LD(&(x[WS(vs, 5)]), ms, &(x[WS(vs, 5)]));
Chris@42 122 T17 = VSUB(T15, T16);
Chris@42 123 T1r = VADD(T15, T16);
Chris@42 124 T2H = LD(&(x[WS(vs, 5) + WS(rs, 4)]), ms, &(x[WS(vs, 5)]));
Chris@42 125 T11 = VSUB(TU, TV);
Chris@42 126 TW = VADD(TU, TV);
Chris@42 127 T2R = LD(&(x[WS(vs, 5) + WS(rs, 2)]), ms, &(x[WS(vs, 5)]));
Chris@42 128 T1i = VSUB(T1g, T1h);
Chris@42 129 T1s = VADD(T1g, T1h);
Chris@42 130 T2S = LD(&(x[WS(vs, 5) + WS(rs, 6)]), ms, &(x[WS(vs, 5)]));
Chris@42 131 T3d = LD(&(x[WS(vs, 6)]), ms, &(x[WS(vs, 6)]));
Chris@42 132 T2I = VSUB(T2G, T2H);
Chris@42 133 T32 = VADD(T2G, T2H);
Chris@42 134 T3e = LD(&(x[WS(vs, 6) + WS(rs, 4)]), ms, &(x[WS(vs, 6)]));
Chris@42 135 T3o = LD(&(x[WS(vs, 6) + WS(rs, 2)]), ms, &(x[WS(vs, 6)]));
Chris@42 136 T3p = LD(&(x[WS(vs, 6) + WS(rs, 6)]), ms, &(x[WS(vs, 6)]));
Chris@42 137 T1y = VSUB(T1r, T1s);
Chris@42 138 T1t = VADD(T1r, T1s);
Chris@42 139 T33 = VADD(T2R, T2S);
Chris@42 140 T2T = VSUB(T2R, T2S);
Chris@42 141 }
Chris@42 142 }
Chris@42 143 }
Chris@42 144 {
Chris@42 145 V T2y, T2e, T3Q, T2z, T2h, T49, T3P, T3R;
Chris@42 146 {
Chris@42 147 V T6, Tq, T1I, Tr, T9, T21, T1H, T1J;
Chris@42 148 {
Chris@42 149 V T4, T3z, T3A, T5, T7, T8, T1F, T1G;
Chris@42 150 T4 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 151 T3f = VSUB(T3d, T3e);
Chris@42 152 T3z = VADD(T3d, T3e);
Chris@42 153 T3q = VSUB(T3o, T3p);
Chris@42 154 T3A = VADD(T3o, T3p);
Chris@42 155 T5 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@42 156 T7 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@42 157 T34 = VADD(T32, T33);
Chris@42 158 T39 = VSUB(T32, T33);
Chris@42 159 T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@42 160 T1F = LD(&(x[WS(vs, 3) + WS(rs, 1)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 161 T1G = LD(&(x[WS(vs, 3) + WS(rs, 5)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 162 T3G = VSUB(T3z, T3A);
Chris@42 163 T3B = VADD(T3z, T3A);
Chris@42 164 T6 = VSUB(T4, T5);
Chris@42 165 Tq = VADD(T4, T5);
Chris@42 166 T1I = LD(&(x[WS(vs, 3) + WS(rs, 7)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 167 Tr = VADD(T7, T8);
Chris@42 168 T9 = VSUB(T7, T8);
Chris@42 169 T21 = VADD(T1F, T1G);
Chris@42 170 T1H = VSUB(T1F, T1G);
Chris@42 171 T1J = LD(&(x[WS(vs, 3) + WS(rs, 3)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 172 }
Chris@42 173 {
Chris@42 174 V T2f, T22, T1K, T2g, T2c, T2d, T3N, T3O;
Chris@42 175 T2c = LD(&(x[WS(vs, 4) + WS(rs, 1)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 176 T2d = LD(&(x[WS(vs, 4) + WS(rs, 5)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 177 T2f = LD(&(x[WS(vs, 4) + WS(rs, 7)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 178 Ts = VADD(Tq, Tr);
Chris@42 179 Tv = VSUB(Tr, Tq);
Chris@42 180 Tf = VSUB(T9, T6);
Chris@42 181 Ta = VADD(T6, T9);
Chris@42 182 T22 = VADD(T1I, T1J);
Chris@42 183 T1K = VSUB(T1I, T1J);
Chris@42 184 T2y = VADD(T2c, T2d);
Chris@42 185 T2e = VSUB(T2c, T2d);
Chris@42 186 T2g = LD(&(x[WS(vs, 4) + WS(rs, 3)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 187 T3N = LD(&(x[WS(vs, 7) + WS(rs, 1)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 188 T3O = LD(&(x[WS(vs, 7) + WS(rs, 5)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 189 T3Q = LD(&(x[WS(vs, 7) + WS(rs, 7)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 190 T23 = VADD(T21, T22);
Chris@42 191 T26 = VSUB(T22, T21);
Chris@42 192 T1Q = VSUB(T1K, T1H);
Chris@42 193 T1L = VADD(T1H, T1K);
Chris@42 194 T2z = VADD(T2f, T2g);
Chris@42 195 T2h = VSUB(T2f, T2g);
Chris@42 196 T49 = VADD(T3N, T3O);
Chris@42 197 T3P = VSUB(T3N, T3O);
Chris@42 198 T3R = LD(&(x[WS(vs, 7) + WS(rs, 3)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 199 }
Chris@42 200 }
Chris@42 201 {
Chris@42 202 V TX, TD, T1b, TY, TG, T1u, T1a, T1c;
Chris@42 203 {
Chris@42 204 V TE, T4a, T3S, TF, TB, TC, T18, T19;
Chris@42 205 TB = LD(&(x[WS(vs, 1) + WS(rs, 1)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 206 TC = LD(&(x[WS(vs, 1) + WS(rs, 5)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 207 TE = LD(&(x[WS(vs, 1) + WS(rs, 7)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 208 T2A = VADD(T2y, T2z);
Chris@42 209 T2D = VSUB(T2z, T2y);
Chris@42 210 T2n = VSUB(T2h, T2e);
Chris@42 211 T2i = VADD(T2e, T2h);
Chris@42 212 T4a = VADD(T3Q, T3R);
Chris@42 213 T3S = VSUB(T3Q, T3R);
Chris@42 214 TX = VADD(TB, TC);
Chris@42 215 TD = VSUB(TB, TC);
Chris@42 216 TF = LD(&(x[WS(vs, 1) + WS(rs, 3)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 217 T18 = LD(&(x[WS(vs, 2) + WS(rs, 1)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 218 T19 = LD(&(x[WS(vs, 2) + WS(rs, 5)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 219 T1b = LD(&(x[WS(vs, 2) + WS(rs, 7)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 220 T4b = VADD(T49, T4a);
Chris@42 221 T4e = VSUB(T4a, T49);
Chris@42 222 T3Y = VSUB(T3S, T3P);
Chris@42 223 T3T = VADD(T3P, T3S);
Chris@42 224 TY = VADD(TE, TF);
Chris@42 225 TG = VSUB(TE, TF);
Chris@42 226 T1u = VADD(T18, T19);
Chris@42 227 T1a = VSUB(T18, T19);
Chris@42 228 T1c = LD(&(x[WS(vs, 2) + WS(rs, 3)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 229 }
Chris@42 230 {
Chris@42 231 V T2M, T1v, T1d, T2N, T2J, T2K, T3g, T3h;
Chris@42 232 T2J = LD(&(x[WS(vs, 5) + WS(rs, 1)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 233 T2K = LD(&(x[WS(vs, 5) + WS(rs, 5)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 234 T2M = LD(&(x[WS(vs, 5) + WS(rs, 7)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 235 TZ = VADD(TX, TY);
Chris@42 236 T12 = VSUB(TY, TX);
Chris@42 237 TM = VSUB(TG, TD);
Chris@42 238 TH = VADD(TD, TG);
Chris@42 239 T1v = VADD(T1b, T1c);
Chris@42 240 T1d = VSUB(T1b, T1c);
Chris@42 241 T35 = VADD(T2J, T2K);
Chris@42 242 T2L = VSUB(T2J, T2K);
Chris@42 243 T2N = LD(&(x[WS(vs, 5) + WS(rs, 3)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 244 T3g = LD(&(x[WS(vs, 6) + WS(rs, 1)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 245 T3h = LD(&(x[WS(vs, 6) + WS(rs, 5)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 246 T3j = LD(&(x[WS(vs, 6) + WS(rs, 7)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 247 T1w = VADD(T1u, T1v);
Chris@42 248 T1z = VSUB(T1v, T1u);
Chris@42 249 T1j = VSUB(T1d, T1a);
Chris@42 250 T1e = VADD(T1a, T1d);
Chris@42 251 T36 = VADD(T2M, T2N);
Chris@42 252 T2O = VSUB(T2M, T2N);
Chris@42 253 T3C = VADD(T3g, T3h);
Chris@42 254 T3i = VSUB(T3g, T3h);
Chris@42 255 T3k = LD(&(x[WS(vs, 6) + WS(rs, 3)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 256 }
Chris@42 257 }
Chris@42 258 }
Chris@42 259 }
Chris@42 260 {
Chris@42 261 V T3a, T2U, T2P, T3H, T3r, T3m, T13, T27, T3b, T4f;
Chris@42 262 {
Chris@42 263 V T37, T3E, T2B, T24;
Chris@42 264 {
Chris@42 265 V T3D, T3l, Tt, T4c;
Chris@42 266 ST(&(x[0]), VADD(Tp, Ts), ms, &(x[0]));
Chris@42 267 ST(&(x[WS(rs, 2)]), VADD(T1t, T1w), ms, &(x[0]));
Chris@42 268 ST(&(x[WS(rs, 7)]), VADD(T48, T4b), ms, &(x[WS(rs, 1)]));
Chris@42 269 T37 = VADD(T35, T36);
Chris@42 270 T3a = VSUB(T36, T35);
Chris@42 271 T2U = VSUB(T2O, T2L);
Chris@42 272 T2P = VADD(T2L, T2O);
Chris@42 273 T3D = VADD(T3j, T3k);
Chris@42 274 T3l = VSUB(T3j, T3k);
Chris@42 275 ST(&(x[WS(rs, 4)]), VADD(T2x, T2A), ms, &(x[0]));
Chris@42 276 ST(&(x[WS(rs, 3)]), VADD(T20, T23), ms, &(x[WS(rs, 1)]));
Chris@42 277 ST(&(x[WS(rs, 5)]), VADD(T34, T37), ms, &(x[WS(rs, 1)]));
Chris@42 278 ST(&(x[WS(rs, 1)]), VADD(TW, TZ), ms, &(x[WS(rs, 1)]));
Chris@42 279 Tt = BYTWJ(&(W[TWVL * 6]), VSUB(Tp, Ts));
Chris@42 280 T4c = BYTWJ(&(W[TWVL * 6]), VSUB(T48, T4b));
Chris@42 281 T3E = VADD(T3C, T3D);
Chris@42 282 T3H = VSUB(T3D, T3C);
Chris@42 283 T3r = VSUB(T3l, T3i);
Chris@42 284 T3m = VADD(T3i, T3l);
Chris@42 285 T2B = BYTWJ(&(W[TWVL * 6]), VSUB(T2x, T2A));
Chris@42 286 T24 = BYTWJ(&(W[TWVL * 6]), VSUB(T20, T23));
Chris@42 287 ST(&(x[WS(vs, 4)]), Tt, ms, &(x[WS(vs, 4)]));
Chris@42 288 ST(&(x[WS(vs, 4) + WS(rs, 7)]), T4c, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 289 ST(&(x[WS(rs, 6)]), VADD(T3B, T3E), ms, &(x[0]));
Chris@42 290 }
Chris@42 291 {
Chris@42 292 V T38, T1A, Tw, T10, T1x, T3F, T2E, T3I;
Chris@42 293 T10 = BYTWJ(&(W[TWVL * 6]), VSUB(TW, TZ));
Chris@42 294 T1x = BYTWJ(&(W[TWVL * 6]), VSUB(T1t, T1w));
Chris@42 295 T3F = BYTWJ(&(W[TWVL * 6]), VSUB(T3B, T3E));
Chris@42 296 ST(&(x[WS(vs, 4) + WS(rs, 4)]), T2B, ms, &(x[WS(vs, 4)]));
Chris@42 297 ST(&(x[WS(vs, 4) + WS(rs, 3)]), T24, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 298 T38 = BYTWJ(&(W[TWVL * 6]), VSUB(T34, T37));
Chris@42 299 T1A = BYTWJ(&(W[TWVL * 10]), VFNMSI(T1z, T1y));
Chris@42 300 Tw = BYTWJ(&(W[TWVL * 10]), VFNMSI(Tv, Tu));
Chris@42 301 ST(&(x[WS(vs, 4) + WS(rs, 1)]), T10, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 302 ST(&(x[WS(vs, 4) + WS(rs, 2)]), T1x, ms, &(x[WS(vs, 4)]));
Chris@42 303 ST(&(x[WS(vs, 4) + WS(rs, 6)]), T3F, ms, &(x[WS(vs, 4)]));
Chris@42 304 T2E = BYTWJ(&(W[TWVL * 10]), VFNMSI(T2D, T2C));
Chris@42 305 T3I = BYTWJ(&(W[TWVL * 10]), VFNMSI(T3H, T3G));
Chris@42 306 ST(&(x[WS(vs, 4) + WS(rs, 5)]), T38, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 307 ST(&(x[WS(vs, 6) + WS(rs, 2)]), T1A, ms, &(x[WS(vs, 6)]));
Chris@42 308 ST(&(x[WS(vs, 6)]), Tw, ms, &(x[WS(vs, 6)]));
Chris@42 309 T13 = BYTWJ(&(W[TWVL * 10]), VFNMSI(T12, T11));
Chris@42 310 T27 = BYTWJ(&(W[TWVL * 10]), VFNMSI(T26, T25));
Chris@42 311 T3b = BYTWJ(&(W[TWVL * 10]), VFNMSI(T3a, T39));
Chris@42 312 ST(&(x[WS(vs, 6) + WS(rs, 4)]), T2E, ms, &(x[WS(vs, 6)]));
Chris@42 313 ST(&(x[WS(vs, 6) + WS(rs, 6)]), T3I, ms, &(x[WS(vs, 6)]));
Chris@42 314 T4f = BYTWJ(&(W[TWVL * 10]), VFNMSI(T4e, T4d));
Chris@42 315 }
Chris@42 316 }
Chris@42 317 {
Chris@42 318 V Tj, Tk, T2r, T2j, Ti, Th, T2o, T2s, T1M, T1R, T41, T40;
Chris@42 319 {
Chris@42 320 V T3c, T4g, T3J, T2F, Tx, T1B;
Chris@42 321 Tx = BYTWJ(&(W[TWVL * 2]), VFMAI(Tv, Tu));
Chris@42 322 T1B = BYTWJ(&(W[TWVL * 2]), VFMAI(T1z, T1y));
Chris@42 323 ST(&(x[WS(vs, 6) + WS(rs, 1)]), T13, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 324 ST(&(x[WS(vs, 6) + WS(rs, 3)]), T27, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 325 ST(&(x[WS(vs, 6) + WS(rs, 5)]), T3b, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 326 T3c = BYTWJ(&(W[TWVL * 2]), VFMAI(T3a, T39));
Chris@42 327 T4g = BYTWJ(&(W[TWVL * 2]), VFMAI(T4e, T4d));
Chris@42 328 ST(&(x[WS(vs, 6) + WS(rs, 7)]), T4f, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 329 ST(&(x[WS(vs, 2)]), Tx, ms, &(x[WS(vs, 2)]));
Chris@42 330 ST(&(x[WS(vs, 2) + WS(rs, 2)]), T1B, ms, &(x[WS(vs, 2)]));
Chris@42 331 T3J = BYTWJ(&(W[TWVL * 2]), VFMAI(T3H, T3G));
Chris@42 332 T2F = BYTWJ(&(W[TWVL * 2]), VFMAI(T2D, T2C));
Chris@42 333 {
Chris@42 334 V T14, Tb, Tg, T28, T3U, T3Z;
Chris@42 335 T28 = BYTWJ(&(W[TWVL * 2]), VFMAI(T26, T25));
Chris@42 336 ST(&(x[WS(vs, 2) + WS(rs, 5)]), T3c, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 337 ST(&(x[WS(vs, 2) + WS(rs, 7)]), T4g, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 338 T14 = BYTWJ(&(W[TWVL * 2]), VFMAI(T12, T11));
Chris@42 339 Tj = VFNMS(LDK(KP707106781), Ta, T3);
Chris@42 340 Tb = VFMA(LDK(KP707106781), Ta, T3);
Chris@42 341 Tg = VFNMS(LDK(KP707106781), Tf, Te);
Chris@42 342 Tk = VFMA(LDK(KP707106781), Tf, Te);
Chris@42 343 ST(&(x[WS(vs, 2) + WS(rs, 6)]), T3J, ms, &(x[WS(vs, 2)]));
Chris@42 344 ST(&(x[WS(vs, 2) + WS(rs, 4)]), T2F, ms, &(x[WS(vs, 2)]));
Chris@42 345 ST(&(x[WS(vs, 2) + WS(rs, 3)]), T28, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 346 T3U = VFMA(LDK(KP707106781), T3T, T3M);
Chris@42 347 T42 = VFNMS(LDK(KP707106781), T3T, T3M);
Chris@42 348 T43 = VFMA(LDK(KP707106781), T3Y, T3X);
Chris@42 349 T3Z = VFNMS(LDK(KP707106781), T3Y, T3X);
Chris@42 350 ST(&(x[WS(vs, 2) + WS(rs, 1)]), T14, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 351 T2r = VFNMS(LDK(KP707106781), T2i, T2b);
Chris@42 352 T2j = VFMA(LDK(KP707106781), T2i, T2b);
Chris@42 353 Ti = BYTWJ(&(W[TWVL * 12]), VFMAI(Tg, Tb));
Chris@42 354 Th = BYTWJ(&(W[0]), VFNMSI(Tg, Tb));
Chris@42 355 T2o = VFNMS(LDK(KP707106781), T2n, T2m);
Chris@42 356 T2s = VFMA(LDK(KP707106781), T2n, T2m);
Chris@42 357 T1U = VFNMS(LDK(KP707106781), T1L, T1E);
Chris@42 358 T1M = VFMA(LDK(KP707106781), T1L, T1E);
Chris@42 359 T1R = VFNMS(LDK(KP707106781), T1Q, T1P);
Chris@42 360 T1V = VFMA(LDK(KP707106781), T1Q, T1P);
Chris@42 361 T41 = BYTWJ(&(W[TWVL * 12]), VFMAI(T3Z, T3U));
Chris@42 362 T40 = BYTWJ(&(W[0]), VFNMSI(T3Z, T3U));
Chris@42 363 }
Chris@42 364 }
Chris@42 365 {
Chris@42 366 V TQ, TR, T1n, T1o, T3v, T3w;
Chris@42 367 {
Chris@42 368 V T1f, T1k, T3n, TP, TO, T3s, T2Q, T2V;
Chris@42 369 {
Chris@42 370 V TI, T2q, T2p, T1T, T1S, TN;
Chris@42 371 TQ = VFNMS(LDK(KP707106781), TH, TA);
Chris@42 372 TI = VFMA(LDK(KP707106781), TH, TA);
Chris@42 373 ST(&(x[WS(vs, 7)]), Ti, ms, &(x[WS(vs, 7)]));
Chris@42 374 ST(&(x[WS(vs, 1)]), Th, ms, &(x[WS(vs, 1)]));
Chris@42 375 T2q = BYTWJ(&(W[TWVL * 12]), VFMAI(T2o, T2j));
Chris@42 376 T2p = BYTWJ(&(W[0]), VFNMSI(T2o, T2j));
Chris@42 377 T1T = BYTWJ(&(W[TWVL * 12]), VFMAI(T1R, T1M));
Chris@42 378 T1S = BYTWJ(&(W[0]), VFNMSI(T1R, T1M));
Chris@42 379 ST(&(x[WS(vs, 7) + WS(rs, 7)]), T41, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 380 ST(&(x[WS(vs, 1) + WS(rs, 7)]), T40, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 381 TN = VFNMS(LDK(KP707106781), TM, TL);
Chris@42 382 TR = VFMA(LDK(KP707106781), TM, TL);
Chris@42 383 T1n = VFNMS(LDK(KP707106781), T1e, T17);
Chris@42 384 T1f = VFMA(LDK(KP707106781), T1e, T17);
Chris@42 385 ST(&(x[WS(vs, 7) + WS(rs, 4)]), T2q, ms, &(x[WS(vs, 7)]));
Chris@42 386 ST(&(x[WS(vs, 1) + WS(rs, 4)]), T2p, ms, &(x[WS(vs, 1)]));
Chris@42 387 ST(&(x[WS(vs, 7) + WS(rs, 3)]), T1T, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 388 ST(&(x[WS(vs, 1) + WS(rs, 3)]), T1S, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 389 T1k = VFNMS(LDK(KP707106781), T1j, T1i);
Chris@42 390 T1o = VFMA(LDK(KP707106781), T1j, T1i);
Chris@42 391 T3v = VFNMS(LDK(KP707106781), T3m, T3f);
Chris@42 392 T3n = VFMA(LDK(KP707106781), T3m, T3f);
Chris@42 393 TP = BYTWJ(&(W[TWVL * 12]), VFMAI(TN, TI));
Chris@42 394 TO = BYTWJ(&(W[0]), VFNMSI(TN, TI));
Chris@42 395 T3s = VFNMS(LDK(KP707106781), T3r, T3q);
Chris@42 396 T3w = VFMA(LDK(KP707106781), T3r, T3q);
Chris@42 397 }
Chris@42 398 T2Y = VFNMS(LDK(KP707106781), T2P, T2I);
Chris@42 399 T2Q = VFMA(LDK(KP707106781), T2P, T2I);
Chris@42 400 T2V = VFNMS(LDK(KP707106781), T2U, T2T);
Chris@42 401 T2Z = VFMA(LDK(KP707106781), T2U, T2T);
Chris@42 402 {
Chris@42 403 V T3u, T3t, T2X, T2W, T1m, T1l;
Chris@42 404 T1m = BYTWJ(&(W[TWVL * 12]), VFMAI(T1k, T1f));
Chris@42 405 T1l = BYTWJ(&(W[0]), VFNMSI(T1k, T1f));
Chris@42 406 ST(&(x[WS(vs, 7) + WS(rs, 1)]), TP, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 407 ST(&(x[WS(vs, 1) + WS(rs, 1)]), TO, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 408 T3u = BYTWJ(&(W[TWVL * 12]), VFMAI(T3s, T3n));
Chris@42 409 T3t = BYTWJ(&(W[0]), VFNMSI(T3s, T3n));
Chris@42 410 T2X = BYTWJ(&(W[TWVL * 12]), VFMAI(T2V, T2Q));
Chris@42 411 T2W = BYTWJ(&(W[0]), VFNMSI(T2V, T2Q));
Chris@42 412 ST(&(x[WS(vs, 7) + WS(rs, 2)]), T1m, ms, &(x[WS(vs, 7)]));
Chris@42 413 ST(&(x[WS(vs, 1) + WS(rs, 2)]), T1l, ms, &(x[WS(vs, 1)]));
Chris@42 414 ST(&(x[WS(vs, 7) + WS(rs, 6)]), T3u, ms, &(x[WS(vs, 7)]));
Chris@42 415 ST(&(x[WS(vs, 1) + WS(rs, 6)]), T3t, ms, &(x[WS(vs, 1)]));
Chris@42 416 ST(&(x[WS(vs, 7) + WS(rs, 5)]), T2X, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 417 ST(&(x[WS(vs, 1) + WS(rs, 5)]), T2W, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 418 }
Chris@42 419 }
Chris@42 420 {
Chris@42 421 V T2u, T2t, T3y, T3x;
Chris@42 422 {
Chris@42 423 V T1q, T1p, Tm, Tl;
Chris@42 424 T1q = BYTWJ(&(W[TWVL * 4]), VFMAI(T1o, T1n));
Chris@42 425 T1p = BYTWJ(&(W[TWVL * 8]), VFNMSI(T1o, T1n));
Chris@42 426 Tm = BYTWJ(&(W[TWVL * 4]), VFMAI(Tk, Tj));
Chris@42 427 Tl = BYTWJ(&(W[TWVL * 8]), VFNMSI(Tk, Tj));
Chris@42 428 ST(&(x[WS(vs, 3) + WS(rs, 2)]), T1q, ms, &(x[WS(vs, 3)]));
Chris@42 429 ST(&(x[WS(vs, 5) + WS(rs, 2)]), T1p, ms, &(x[WS(vs, 5)]));
Chris@42 430 T2u = BYTWJ(&(W[TWVL * 4]), VFMAI(T2s, T2r));
Chris@42 431 T2t = BYTWJ(&(W[TWVL * 8]), VFNMSI(T2s, T2r));
Chris@42 432 T3y = BYTWJ(&(W[TWVL * 4]), VFMAI(T3w, T3v));
Chris@42 433 T3x = BYTWJ(&(W[TWVL * 8]), VFNMSI(T3w, T3v));
Chris@42 434 ST(&(x[WS(vs, 3)]), Tm, ms, &(x[WS(vs, 3)]));
Chris@42 435 ST(&(x[WS(vs, 5)]), Tl, ms, &(x[WS(vs, 5)]));
Chris@42 436 }
Chris@42 437 ST(&(x[WS(vs, 3) + WS(rs, 4)]), T2u, ms, &(x[WS(vs, 3)]));
Chris@42 438 ST(&(x[WS(vs, 5) + WS(rs, 4)]), T2t, ms, &(x[WS(vs, 5)]));
Chris@42 439 ST(&(x[WS(vs, 3) + WS(rs, 6)]), T3y, ms, &(x[WS(vs, 3)]));
Chris@42 440 ST(&(x[WS(vs, 5) + WS(rs, 6)]), T3x, ms, &(x[WS(vs, 5)]));
Chris@42 441 TT = BYTWJ(&(W[TWVL * 4]), VFMAI(TR, TQ));
Chris@42 442 TS = BYTWJ(&(W[TWVL * 8]), VFNMSI(TR, TQ));
Chris@42 443 }
Chris@42 444 }
Chris@42 445 }
Chris@42 446 }
Chris@42 447 }
Chris@42 448 {
Chris@42 449 V T31, T30, T45, T44, T1X, T1W;
Chris@42 450 T1X = BYTWJ(&(W[TWVL * 4]), VFMAI(T1V, T1U));
Chris@42 451 T1W = BYTWJ(&(W[TWVL * 8]), VFNMSI(T1V, T1U));
Chris@42 452 ST(&(x[WS(vs, 3) + WS(rs, 1)]), TT, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 453 ST(&(x[WS(vs, 5) + WS(rs, 1)]), TS, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 454 T31 = BYTWJ(&(W[TWVL * 4]), VFMAI(T2Z, T2Y));
Chris@42 455 T30 = BYTWJ(&(W[TWVL * 8]), VFNMSI(T2Z, T2Y));
Chris@42 456 T45 = BYTWJ(&(W[TWVL * 4]), VFMAI(T43, T42));
Chris@42 457 T44 = BYTWJ(&(W[TWVL * 8]), VFNMSI(T43, T42));
Chris@42 458 ST(&(x[WS(vs, 3) + WS(rs, 3)]), T1X, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 459 ST(&(x[WS(vs, 5) + WS(rs, 3)]), T1W, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 460 ST(&(x[WS(vs, 3) + WS(rs, 5)]), T31, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 461 ST(&(x[WS(vs, 5) + WS(rs, 5)]), T30, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 462 ST(&(x[WS(vs, 3) + WS(rs, 7)]), T45, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 463 ST(&(x[WS(vs, 5) + WS(rs, 7)]), T44, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 464 }
Chris@42 465 }
Chris@42 466 }
Chris@42 467 VLEAVE();
Chris@42 468 }
Chris@42 469
Chris@42 470 static const tw_instr twinstr[] = {
Chris@42 471 VTW(0, 1),
Chris@42 472 VTW(0, 2),
Chris@42 473 VTW(0, 3),
Chris@42 474 VTW(0, 4),
Chris@42 475 VTW(0, 5),
Chris@42 476 VTW(0, 6),
Chris@42 477 VTW(0, 7),
Chris@42 478 {TW_NEXT, VL, 0}
Chris@42 479 };
Chris@42 480
Chris@42 481 static const ct_desc desc = { 8, XSIMD_STRING("q1fv_8"), twinstr, &GENUS, {184, 112, 80, 0}, 0, 0, 0 };
Chris@42 482
Chris@42 483 void XSIMD(codelet_q1fv_8) (planner *p) {
Chris@42 484 X(kdft_difsq_register) (p, q1fv_8, &desc);
Chris@42 485 }
Chris@42 486 #else /* HAVE_FMA */
Chris@42 487
Chris@42 488 /* Generated by: ../../../genfft/gen_twidsq_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 8 -dif -name q1fv_8 -include q1f.h */
Chris@42 489
Chris@42 490 /*
Chris@42 491 * This function contains 264 FP additions, 128 FP multiplications,
Chris@42 492 * (or, 264 additions, 128 multiplications, 0 fused multiply/add),
Chris@42 493 * 77 stack variables, 1 constants, and 128 memory accesses
Chris@42 494 */
Chris@42 495 #include "q1f.h"
Chris@42 496
Chris@42 497 static void q1fv_8(R *ri, R *ii, const R *W, stride rs, stride vs, INT mb, INT me, INT ms)
Chris@42 498 {
Chris@42 499 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 500 {
Chris@42 501 INT m;
Chris@42 502 R *x;
Chris@42 503 x = ri;
Chris@42 504 for (m = mb, W = W + (mb * ((TWVL / VL) * 14)); m < me; m = m + VL, x = x + (VL * ms), W = W + (TWVL * 14), MAKE_VOLATILE_STRIDE(16, rs), MAKE_VOLATILE_STRIDE(16, vs)) {
Chris@42 505 V T3, Tu, Tf, Tp, T1E, T25, T1Q, T20, T2b, T2C, T2n, T2x, T3M, T4d, T3Y;
Chris@42 506 V T48, TA, T11, TM, TW, T17, T1y, T1j, T1t, T2I, T39, T2U, T34, T3f, T3G;
Chris@42 507 V T3r, T3B, Ta, Tv, Tc, Ts, T1L, T26, T1N, T23, T2i, T2D, T2k, T2A, T3T;
Chris@42 508 V T4e, T3V, T4b, TH, T12, TJ, TZ, T1e, T1z, T1g, T1w, T2P, T3a, T2R, T37;
Chris@42 509 V T3m, T3H, T3o, T3E, T28, T14;
Chris@42 510 {
Chris@42 511 V T1, T2, Tn, Td, Te, To;
Chris@42 512 T1 = LD(&(x[0]), ms, &(x[0]));
Chris@42 513 T2 = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@42 514 Tn = VADD(T1, T2);
Chris@42 515 Td = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 516 Te = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@42 517 To = VADD(Td, Te);
Chris@42 518 T3 = VSUB(T1, T2);
Chris@42 519 Tu = VSUB(Tn, To);
Chris@42 520 Tf = VSUB(Td, Te);
Chris@42 521 Tp = VADD(Tn, To);
Chris@42 522 }
Chris@42 523 {
Chris@42 524 V T1C, T1D, T1Y, T1O, T1P, T1Z;
Chris@42 525 T1C = LD(&(x[WS(vs, 3)]), ms, &(x[WS(vs, 3)]));
Chris@42 526 T1D = LD(&(x[WS(vs, 3) + WS(rs, 4)]), ms, &(x[WS(vs, 3)]));
Chris@42 527 T1Y = VADD(T1C, T1D);
Chris@42 528 T1O = LD(&(x[WS(vs, 3) + WS(rs, 2)]), ms, &(x[WS(vs, 3)]));
Chris@42 529 T1P = LD(&(x[WS(vs, 3) + WS(rs, 6)]), ms, &(x[WS(vs, 3)]));
Chris@42 530 T1Z = VADD(T1O, T1P);
Chris@42 531 T1E = VSUB(T1C, T1D);
Chris@42 532 T25 = VSUB(T1Y, T1Z);
Chris@42 533 T1Q = VSUB(T1O, T1P);
Chris@42 534 T20 = VADD(T1Y, T1Z);
Chris@42 535 }
Chris@42 536 {
Chris@42 537 V T29, T2a, T2v, T2l, T2m, T2w;
Chris@42 538 T29 = LD(&(x[WS(vs, 4)]), ms, &(x[WS(vs, 4)]));
Chris@42 539 T2a = LD(&(x[WS(vs, 4) + WS(rs, 4)]), ms, &(x[WS(vs, 4)]));
Chris@42 540 T2v = VADD(T29, T2a);
Chris@42 541 T2l = LD(&(x[WS(vs, 4) + WS(rs, 2)]), ms, &(x[WS(vs, 4)]));
Chris@42 542 T2m = LD(&(x[WS(vs, 4) + WS(rs, 6)]), ms, &(x[WS(vs, 4)]));
Chris@42 543 T2w = VADD(T2l, T2m);
Chris@42 544 T2b = VSUB(T29, T2a);
Chris@42 545 T2C = VSUB(T2v, T2w);
Chris@42 546 T2n = VSUB(T2l, T2m);
Chris@42 547 T2x = VADD(T2v, T2w);
Chris@42 548 }
Chris@42 549 {
Chris@42 550 V T3K, T3L, T46, T3W, T3X, T47;
Chris@42 551 T3K = LD(&(x[WS(vs, 7)]), ms, &(x[WS(vs, 7)]));
Chris@42 552 T3L = LD(&(x[WS(vs, 7) + WS(rs, 4)]), ms, &(x[WS(vs, 7)]));
Chris@42 553 T46 = VADD(T3K, T3L);
Chris@42 554 T3W = LD(&(x[WS(vs, 7) + WS(rs, 2)]), ms, &(x[WS(vs, 7)]));
Chris@42 555 T3X = LD(&(x[WS(vs, 7) + WS(rs, 6)]), ms, &(x[WS(vs, 7)]));
Chris@42 556 T47 = VADD(T3W, T3X);
Chris@42 557 T3M = VSUB(T3K, T3L);
Chris@42 558 T4d = VSUB(T46, T47);
Chris@42 559 T3Y = VSUB(T3W, T3X);
Chris@42 560 T48 = VADD(T46, T47);
Chris@42 561 }
Chris@42 562 {
Chris@42 563 V Ty, Tz, TU, TK, TL, TV;
Chris@42 564 Ty = LD(&(x[WS(vs, 1)]), ms, &(x[WS(vs, 1)]));
Chris@42 565 Tz = LD(&(x[WS(vs, 1) + WS(rs, 4)]), ms, &(x[WS(vs, 1)]));
Chris@42 566 TU = VADD(Ty, Tz);
Chris@42 567 TK = LD(&(x[WS(vs, 1) + WS(rs, 2)]), ms, &(x[WS(vs, 1)]));
Chris@42 568 TL = LD(&(x[WS(vs, 1) + WS(rs, 6)]), ms, &(x[WS(vs, 1)]));
Chris@42 569 TV = VADD(TK, TL);
Chris@42 570 TA = VSUB(Ty, Tz);
Chris@42 571 T11 = VSUB(TU, TV);
Chris@42 572 TM = VSUB(TK, TL);
Chris@42 573 TW = VADD(TU, TV);
Chris@42 574 }
Chris@42 575 {
Chris@42 576 V T15, T16, T1r, T1h, T1i, T1s;
Chris@42 577 T15 = LD(&(x[WS(vs, 2)]), ms, &(x[WS(vs, 2)]));
Chris@42 578 T16 = LD(&(x[WS(vs, 2) + WS(rs, 4)]), ms, &(x[WS(vs, 2)]));
Chris@42 579 T1r = VADD(T15, T16);
Chris@42 580 T1h = LD(&(x[WS(vs, 2) + WS(rs, 2)]), ms, &(x[WS(vs, 2)]));
Chris@42 581 T1i = LD(&(x[WS(vs, 2) + WS(rs, 6)]), ms, &(x[WS(vs, 2)]));
Chris@42 582 T1s = VADD(T1h, T1i);
Chris@42 583 T17 = VSUB(T15, T16);
Chris@42 584 T1y = VSUB(T1r, T1s);
Chris@42 585 T1j = VSUB(T1h, T1i);
Chris@42 586 T1t = VADD(T1r, T1s);
Chris@42 587 }
Chris@42 588 {
Chris@42 589 V T2G, T2H, T32, T2S, T2T, T33;
Chris@42 590 T2G = LD(&(x[WS(vs, 5)]), ms, &(x[WS(vs, 5)]));
Chris@42 591 T2H = LD(&(x[WS(vs, 5) + WS(rs, 4)]), ms, &(x[WS(vs, 5)]));
Chris@42 592 T32 = VADD(T2G, T2H);
Chris@42 593 T2S = LD(&(x[WS(vs, 5) + WS(rs, 2)]), ms, &(x[WS(vs, 5)]));
Chris@42 594 T2T = LD(&(x[WS(vs, 5) + WS(rs, 6)]), ms, &(x[WS(vs, 5)]));
Chris@42 595 T33 = VADD(T2S, T2T);
Chris@42 596 T2I = VSUB(T2G, T2H);
Chris@42 597 T39 = VSUB(T32, T33);
Chris@42 598 T2U = VSUB(T2S, T2T);
Chris@42 599 T34 = VADD(T32, T33);
Chris@42 600 }
Chris@42 601 {
Chris@42 602 V T3d, T3e, T3z, T3p, T3q, T3A;
Chris@42 603 T3d = LD(&(x[WS(vs, 6)]), ms, &(x[WS(vs, 6)]));
Chris@42 604 T3e = LD(&(x[WS(vs, 6) + WS(rs, 4)]), ms, &(x[WS(vs, 6)]));
Chris@42 605 T3z = VADD(T3d, T3e);
Chris@42 606 T3p = LD(&(x[WS(vs, 6) + WS(rs, 2)]), ms, &(x[WS(vs, 6)]));
Chris@42 607 T3q = LD(&(x[WS(vs, 6) + WS(rs, 6)]), ms, &(x[WS(vs, 6)]));
Chris@42 608 T3A = VADD(T3p, T3q);
Chris@42 609 T3f = VSUB(T3d, T3e);
Chris@42 610 T3G = VSUB(T3z, T3A);
Chris@42 611 T3r = VSUB(T3p, T3q);
Chris@42 612 T3B = VADD(T3z, T3A);
Chris@42 613 }
Chris@42 614 {
Chris@42 615 V T6, Tq, T9, Tr;
Chris@42 616 {
Chris@42 617 V T4, T5, T7, T8;
Chris@42 618 T4 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 619 T5 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@42 620 T6 = VSUB(T4, T5);
Chris@42 621 Tq = VADD(T4, T5);
Chris@42 622 T7 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@42 623 T8 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@42 624 T9 = VSUB(T7, T8);
Chris@42 625 Tr = VADD(T7, T8);
Chris@42 626 }
Chris@42 627 Ta = VMUL(LDK(KP707106781), VADD(T6, T9));
Chris@42 628 Tv = VBYI(VSUB(Tr, Tq));
Chris@42 629 Tc = VMUL(LDK(KP707106781), VSUB(T9, T6));
Chris@42 630 Ts = VADD(Tq, Tr);
Chris@42 631 }
Chris@42 632 {
Chris@42 633 V T1H, T21, T1K, T22;
Chris@42 634 {
Chris@42 635 V T1F, T1G, T1I, T1J;
Chris@42 636 T1F = LD(&(x[WS(vs, 3) + WS(rs, 1)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 637 T1G = LD(&(x[WS(vs, 3) + WS(rs, 5)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 638 T1H = VSUB(T1F, T1G);
Chris@42 639 T21 = VADD(T1F, T1G);
Chris@42 640 T1I = LD(&(x[WS(vs, 3) + WS(rs, 7)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 641 T1J = LD(&(x[WS(vs, 3) + WS(rs, 3)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 642 T1K = VSUB(T1I, T1J);
Chris@42 643 T22 = VADD(T1I, T1J);
Chris@42 644 }
Chris@42 645 T1L = VMUL(LDK(KP707106781), VADD(T1H, T1K));
Chris@42 646 T26 = VBYI(VSUB(T22, T21));
Chris@42 647 T1N = VMUL(LDK(KP707106781), VSUB(T1K, T1H));
Chris@42 648 T23 = VADD(T21, T22);
Chris@42 649 }
Chris@42 650 {
Chris@42 651 V T2e, T2y, T2h, T2z;
Chris@42 652 {
Chris@42 653 V T2c, T2d, T2f, T2g;
Chris@42 654 T2c = LD(&(x[WS(vs, 4) + WS(rs, 1)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 655 T2d = LD(&(x[WS(vs, 4) + WS(rs, 5)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 656 T2e = VSUB(T2c, T2d);
Chris@42 657 T2y = VADD(T2c, T2d);
Chris@42 658 T2f = LD(&(x[WS(vs, 4) + WS(rs, 7)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 659 T2g = LD(&(x[WS(vs, 4) + WS(rs, 3)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 660 T2h = VSUB(T2f, T2g);
Chris@42 661 T2z = VADD(T2f, T2g);
Chris@42 662 }
Chris@42 663 T2i = VMUL(LDK(KP707106781), VADD(T2e, T2h));
Chris@42 664 T2D = VBYI(VSUB(T2z, T2y));
Chris@42 665 T2k = VMUL(LDK(KP707106781), VSUB(T2h, T2e));
Chris@42 666 T2A = VADD(T2y, T2z);
Chris@42 667 }
Chris@42 668 {
Chris@42 669 V T3P, T49, T3S, T4a;
Chris@42 670 {
Chris@42 671 V T3N, T3O, T3Q, T3R;
Chris@42 672 T3N = LD(&(x[WS(vs, 7) + WS(rs, 1)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 673 T3O = LD(&(x[WS(vs, 7) + WS(rs, 5)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 674 T3P = VSUB(T3N, T3O);
Chris@42 675 T49 = VADD(T3N, T3O);
Chris@42 676 T3Q = LD(&(x[WS(vs, 7) + WS(rs, 7)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 677 T3R = LD(&(x[WS(vs, 7) + WS(rs, 3)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 678 T3S = VSUB(T3Q, T3R);
Chris@42 679 T4a = VADD(T3Q, T3R);
Chris@42 680 }
Chris@42 681 T3T = VMUL(LDK(KP707106781), VADD(T3P, T3S));
Chris@42 682 T4e = VBYI(VSUB(T4a, T49));
Chris@42 683 T3V = VMUL(LDK(KP707106781), VSUB(T3S, T3P));
Chris@42 684 T4b = VADD(T49, T4a);
Chris@42 685 }
Chris@42 686 {
Chris@42 687 V TD, TX, TG, TY;
Chris@42 688 {
Chris@42 689 V TB, TC, TE, TF;
Chris@42 690 TB = LD(&(x[WS(vs, 1) + WS(rs, 1)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 691 TC = LD(&(x[WS(vs, 1) + WS(rs, 5)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 692 TD = VSUB(TB, TC);
Chris@42 693 TX = VADD(TB, TC);
Chris@42 694 TE = LD(&(x[WS(vs, 1) + WS(rs, 7)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 695 TF = LD(&(x[WS(vs, 1) + WS(rs, 3)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 696 TG = VSUB(TE, TF);
Chris@42 697 TY = VADD(TE, TF);
Chris@42 698 }
Chris@42 699 TH = VMUL(LDK(KP707106781), VADD(TD, TG));
Chris@42 700 T12 = VBYI(VSUB(TY, TX));
Chris@42 701 TJ = VMUL(LDK(KP707106781), VSUB(TG, TD));
Chris@42 702 TZ = VADD(TX, TY);
Chris@42 703 }
Chris@42 704 {
Chris@42 705 V T1a, T1u, T1d, T1v;
Chris@42 706 {
Chris@42 707 V T18, T19, T1b, T1c;
Chris@42 708 T18 = LD(&(x[WS(vs, 2) + WS(rs, 1)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 709 T19 = LD(&(x[WS(vs, 2) + WS(rs, 5)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 710 T1a = VSUB(T18, T19);
Chris@42 711 T1u = VADD(T18, T19);
Chris@42 712 T1b = LD(&(x[WS(vs, 2) + WS(rs, 7)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 713 T1c = LD(&(x[WS(vs, 2) + WS(rs, 3)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 714 T1d = VSUB(T1b, T1c);
Chris@42 715 T1v = VADD(T1b, T1c);
Chris@42 716 }
Chris@42 717 T1e = VMUL(LDK(KP707106781), VADD(T1a, T1d));
Chris@42 718 T1z = VBYI(VSUB(T1v, T1u));
Chris@42 719 T1g = VMUL(LDK(KP707106781), VSUB(T1d, T1a));
Chris@42 720 T1w = VADD(T1u, T1v);
Chris@42 721 }
Chris@42 722 {
Chris@42 723 V T2L, T35, T2O, T36;
Chris@42 724 {
Chris@42 725 V T2J, T2K, T2M, T2N;
Chris@42 726 T2J = LD(&(x[WS(vs, 5) + WS(rs, 1)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 727 T2K = LD(&(x[WS(vs, 5) + WS(rs, 5)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 728 T2L = VSUB(T2J, T2K);
Chris@42 729 T35 = VADD(T2J, T2K);
Chris@42 730 T2M = LD(&(x[WS(vs, 5) + WS(rs, 7)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 731 T2N = LD(&(x[WS(vs, 5) + WS(rs, 3)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 732 T2O = VSUB(T2M, T2N);
Chris@42 733 T36 = VADD(T2M, T2N);
Chris@42 734 }
Chris@42 735 T2P = VMUL(LDK(KP707106781), VADD(T2L, T2O));
Chris@42 736 T3a = VBYI(VSUB(T36, T35));
Chris@42 737 T2R = VMUL(LDK(KP707106781), VSUB(T2O, T2L));
Chris@42 738 T37 = VADD(T35, T36);
Chris@42 739 }
Chris@42 740 {
Chris@42 741 V T3i, T3C, T3l, T3D;
Chris@42 742 {
Chris@42 743 V T3g, T3h, T3j, T3k;
Chris@42 744 T3g = LD(&(x[WS(vs, 6) + WS(rs, 1)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 745 T3h = LD(&(x[WS(vs, 6) + WS(rs, 5)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 746 T3i = VSUB(T3g, T3h);
Chris@42 747 T3C = VADD(T3g, T3h);
Chris@42 748 T3j = LD(&(x[WS(vs, 6) + WS(rs, 7)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 749 T3k = LD(&(x[WS(vs, 6) + WS(rs, 3)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 750 T3l = VSUB(T3j, T3k);
Chris@42 751 T3D = VADD(T3j, T3k);
Chris@42 752 }
Chris@42 753 T3m = VMUL(LDK(KP707106781), VADD(T3i, T3l));
Chris@42 754 T3H = VBYI(VSUB(T3D, T3C));
Chris@42 755 T3o = VMUL(LDK(KP707106781), VSUB(T3l, T3i));
Chris@42 756 T3E = VADD(T3C, T3D);
Chris@42 757 }
Chris@42 758 ST(&(x[0]), VADD(Tp, Ts), ms, &(x[0]));
Chris@42 759 ST(&(x[WS(rs, 2)]), VADD(T1t, T1w), ms, &(x[0]));
Chris@42 760 ST(&(x[WS(rs, 5)]), VADD(T34, T37), ms, &(x[WS(rs, 1)]));
Chris@42 761 ST(&(x[WS(rs, 7)]), VADD(T48, T4b), ms, &(x[WS(rs, 1)]));
Chris@42 762 ST(&(x[WS(rs, 6)]), VADD(T3B, T3E), ms, &(x[0]));
Chris@42 763 ST(&(x[WS(rs, 4)]), VADD(T2x, T2A), ms, &(x[0]));
Chris@42 764 {
Chris@42 765 V Tt, T4c, T2B, T24;
Chris@42 766 ST(&(x[WS(rs, 3)]), VADD(T20, T23), ms, &(x[WS(rs, 1)]));
Chris@42 767 ST(&(x[WS(rs, 1)]), VADD(TW, TZ), ms, &(x[WS(rs, 1)]));
Chris@42 768 Tt = BYTWJ(&(W[TWVL * 6]), VSUB(Tp, Ts));
Chris@42 769 ST(&(x[WS(vs, 4)]), Tt, ms, &(x[WS(vs, 4)]));
Chris@42 770 T4c = BYTWJ(&(W[TWVL * 6]), VSUB(T48, T4b));
Chris@42 771 ST(&(x[WS(vs, 4) + WS(rs, 7)]), T4c, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 772 T2B = BYTWJ(&(W[TWVL * 6]), VSUB(T2x, T2A));
Chris@42 773 ST(&(x[WS(vs, 4) + WS(rs, 4)]), T2B, ms, &(x[WS(vs, 4)]));
Chris@42 774 T24 = BYTWJ(&(W[TWVL * 6]), VSUB(T20, T23));
Chris@42 775 ST(&(x[WS(vs, 4) + WS(rs, 3)]), T24, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 776 }
Chris@42 777 {
Chris@42 778 V T10, T1x, T3F, T38, T1A, Tw;
Chris@42 779 T10 = BYTWJ(&(W[TWVL * 6]), VSUB(TW, TZ));
Chris@42 780 ST(&(x[WS(vs, 4) + WS(rs, 1)]), T10, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 781 T1x = BYTWJ(&(W[TWVL * 6]), VSUB(T1t, T1w));
Chris@42 782 ST(&(x[WS(vs, 4) + WS(rs, 2)]), T1x, ms, &(x[WS(vs, 4)]));
Chris@42 783 T3F = BYTWJ(&(W[TWVL * 6]), VSUB(T3B, T3E));
Chris@42 784 ST(&(x[WS(vs, 4) + WS(rs, 6)]), T3F, ms, &(x[WS(vs, 4)]));
Chris@42 785 T38 = BYTWJ(&(W[TWVL * 6]), VSUB(T34, T37));
Chris@42 786 ST(&(x[WS(vs, 4) + WS(rs, 5)]), T38, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 787 T1A = BYTWJ(&(W[TWVL * 10]), VSUB(T1y, T1z));
Chris@42 788 ST(&(x[WS(vs, 6) + WS(rs, 2)]), T1A, ms, &(x[WS(vs, 6)]));
Chris@42 789 Tw = BYTWJ(&(W[TWVL * 10]), VSUB(Tu, Tv));
Chris@42 790 ST(&(x[WS(vs, 6)]), Tw, ms, &(x[WS(vs, 6)]));
Chris@42 791 }
Chris@42 792 {
Chris@42 793 V T2E, T3I, T13, T27, T3b, T4f;
Chris@42 794 T2E = BYTWJ(&(W[TWVL * 10]), VSUB(T2C, T2D));
Chris@42 795 ST(&(x[WS(vs, 6) + WS(rs, 4)]), T2E, ms, &(x[WS(vs, 6)]));
Chris@42 796 T3I = BYTWJ(&(W[TWVL * 10]), VSUB(T3G, T3H));
Chris@42 797 ST(&(x[WS(vs, 6) + WS(rs, 6)]), T3I, ms, &(x[WS(vs, 6)]));
Chris@42 798 T13 = BYTWJ(&(W[TWVL * 10]), VSUB(T11, T12));
Chris@42 799 ST(&(x[WS(vs, 6) + WS(rs, 1)]), T13, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 800 T27 = BYTWJ(&(W[TWVL * 10]), VSUB(T25, T26));
Chris@42 801 ST(&(x[WS(vs, 6) + WS(rs, 3)]), T27, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 802 T3b = BYTWJ(&(W[TWVL * 10]), VSUB(T39, T3a));
Chris@42 803 ST(&(x[WS(vs, 6) + WS(rs, 5)]), T3b, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 804 T4f = BYTWJ(&(W[TWVL * 10]), VSUB(T4d, T4e));
Chris@42 805 ST(&(x[WS(vs, 6) + WS(rs, 7)]), T4f, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 806 }
Chris@42 807 {
Chris@42 808 V Tx, T1B, T3c, T4g, T3J, T2F;
Chris@42 809 Tx = BYTWJ(&(W[TWVL * 2]), VADD(Tu, Tv));
Chris@42 810 ST(&(x[WS(vs, 2)]), Tx, ms, &(x[WS(vs, 2)]));
Chris@42 811 T1B = BYTWJ(&(W[TWVL * 2]), VADD(T1y, T1z));
Chris@42 812 ST(&(x[WS(vs, 2) + WS(rs, 2)]), T1B, ms, &(x[WS(vs, 2)]));
Chris@42 813 T3c = BYTWJ(&(W[TWVL * 2]), VADD(T39, T3a));
Chris@42 814 ST(&(x[WS(vs, 2) + WS(rs, 5)]), T3c, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 815 T4g = BYTWJ(&(W[TWVL * 2]), VADD(T4d, T4e));
Chris@42 816 ST(&(x[WS(vs, 2) + WS(rs, 7)]), T4g, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 817 T3J = BYTWJ(&(W[TWVL * 2]), VADD(T3G, T3H));
Chris@42 818 ST(&(x[WS(vs, 2) + WS(rs, 6)]), T3J, ms, &(x[WS(vs, 2)]));
Chris@42 819 T2F = BYTWJ(&(W[TWVL * 2]), VADD(T2C, T2D));
Chris@42 820 ST(&(x[WS(vs, 2) + WS(rs, 4)]), T2F, ms, &(x[WS(vs, 2)]));
Chris@42 821 }
Chris@42 822 T28 = BYTWJ(&(W[TWVL * 2]), VADD(T25, T26));
Chris@42 823 ST(&(x[WS(vs, 2) + WS(rs, 3)]), T28, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 824 T14 = BYTWJ(&(W[TWVL * 2]), VADD(T11, T12));
Chris@42 825 ST(&(x[WS(vs, 2) + WS(rs, 1)]), T14, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 826 {
Chris@42 827 V Th, Ti, Tb, Tg;
Chris@42 828 Tb = VADD(T3, Ta);
Chris@42 829 Tg = VBYI(VSUB(Tc, Tf));
Chris@42 830 Th = BYTWJ(&(W[TWVL * 12]), VSUB(Tb, Tg));
Chris@42 831 Ti = BYTWJ(&(W[0]), VADD(Tb, Tg));
Chris@42 832 ST(&(x[WS(vs, 7)]), Th, ms, &(x[WS(vs, 7)]));
Chris@42 833 ST(&(x[WS(vs, 1)]), Ti, ms, &(x[WS(vs, 1)]));
Chris@42 834 }
Chris@42 835 {
Chris@42 836 V T40, T41, T3U, T3Z;
Chris@42 837 T3U = VADD(T3M, T3T);
Chris@42 838 T3Z = VBYI(VSUB(T3V, T3Y));
Chris@42 839 T40 = BYTWJ(&(W[TWVL * 12]), VSUB(T3U, T3Z));
Chris@42 840 T41 = BYTWJ(&(W[0]), VADD(T3U, T3Z));
Chris@42 841 ST(&(x[WS(vs, 7) + WS(rs, 7)]), T40, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 842 ST(&(x[WS(vs, 1) + WS(rs, 7)]), T41, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 843 }
Chris@42 844 {
Chris@42 845 V T2p, T2q, T2j, T2o;
Chris@42 846 T2j = VADD(T2b, T2i);
Chris@42 847 T2o = VBYI(VSUB(T2k, T2n));
Chris@42 848 T2p = BYTWJ(&(W[TWVL * 12]), VSUB(T2j, T2o));
Chris@42 849 T2q = BYTWJ(&(W[0]), VADD(T2j, T2o));
Chris@42 850 ST(&(x[WS(vs, 7) + WS(rs, 4)]), T2p, ms, &(x[WS(vs, 7)]));
Chris@42 851 ST(&(x[WS(vs, 1) + WS(rs, 4)]), T2q, ms, &(x[WS(vs, 1)]));
Chris@42 852 }
Chris@42 853 {
Chris@42 854 V T1S, T1T, T1M, T1R;
Chris@42 855 T1M = VADD(T1E, T1L);
Chris@42 856 T1R = VBYI(VSUB(T1N, T1Q));
Chris@42 857 T1S = BYTWJ(&(W[TWVL * 12]), VSUB(T1M, T1R));
Chris@42 858 T1T = BYTWJ(&(W[0]), VADD(T1M, T1R));
Chris@42 859 ST(&(x[WS(vs, 7) + WS(rs, 3)]), T1S, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 860 ST(&(x[WS(vs, 1) + WS(rs, 3)]), T1T, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 861 }
Chris@42 862 {
Chris@42 863 V TO, TP, TI, TN;
Chris@42 864 TI = VADD(TA, TH);
Chris@42 865 TN = VBYI(VSUB(TJ, TM));
Chris@42 866 TO = BYTWJ(&(W[TWVL * 12]), VSUB(TI, TN));
Chris@42 867 TP = BYTWJ(&(W[0]), VADD(TI, TN));
Chris@42 868 ST(&(x[WS(vs, 7) + WS(rs, 1)]), TO, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 869 ST(&(x[WS(vs, 1) + WS(rs, 1)]), TP, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 870 }
Chris@42 871 {
Chris@42 872 V T1l, T1m, T1f, T1k;
Chris@42 873 T1f = VADD(T17, T1e);
Chris@42 874 T1k = VBYI(VSUB(T1g, T1j));
Chris@42 875 T1l = BYTWJ(&(W[TWVL * 12]), VSUB(T1f, T1k));
Chris@42 876 T1m = BYTWJ(&(W[0]), VADD(T1f, T1k));
Chris@42 877 ST(&(x[WS(vs, 7) + WS(rs, 2)]), T1l, ms, &(x[WS(vs, 7)]));
Chris@42 878 ST(&(x[WS(vs, 1) + WS(rs, 2)]), T1m, ms, &(x[WS(vs, 1)]));
Chris@42 879 }
Chris@42 880 {
Chris@42 881 V T3t, T3u, T3n, T3s;
Chris@42 882 T3n = VADD(T3f, T3m);
Chris@42 883 T3s = VBYI(VSUB(T3o, T3r));
Chris@42 884 T3t = BYTWJ(&(W[TWVL * 12]), VSUB(T3n, T3s));
Chris@42 885 T3u = BYTWJ(&(W[0]), VADD(T3n, T3s));
Chris@42 886 ST(&(x[WS(vs, 7) + WS(rs, 6)]), T3t, ms, &(x[WS(vs, 7)]));
Chris@42 887 ST(&(x[WS(vs, 1) + WS(rs, 6)]), T3u, ms, &(x[WS(vs, 1)]));
Chris@42 888 }
Chris@42 889 {
Chris@42 890 V T2W, T2X, T2Q, T2V;
Chris@42 891 T2Q = VADD(T2I, T2P);
Chris@42 892 T2V = VBYI(VSUB(T2R, T2U));
Chris@42 893 T2W = BYTWJ(&(W[TWVL * 12]), VSUB(T2Q, T2V));
Chris@42 894 T2X = BYTWJ(&(W[0]), VADD(T2Q, T2V));
Chris@42 895 ST(&(x[WS(vs, 7) + WS(rs, 5)]), T2W, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 896 ST(&(x[WS(vs, 1) + WS(rs, 5)]), T2X, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 897 }
Chris@42 898 {
Chris@42 899 V T1p, T1q, T1n, T1o;
Chris@42 900 T1n = VSUB(T17, T1e);
Chris@42 901 T1o = VBYI(VADD(T1j, T1g));
Chris@42 902 T1p = BYTWJ(&(W[TWVL * 8]), VSUB(T1n, T1o));
Chris@42 903 T1q = BYTWJ(&(W[TWVL * 4]), VADD(T1n, T1o));
Chris@42 904 ST(&(x[WS(vs, 5) + WS(rs, 2)]), T1p, ms, &(x[WS(vs, 5)]));
Chris@42 905 ST(&(x[WS(vs, 3) + WS(rs, 2)]), T1q, ms, &(x[WS(vs, 3)]));
Chris@42 906 }
Chris@42 907 {
Chris@42 908 V Tl, Tm, Tj, Tk;
Chris@42 909 Tj = VSUB(T3, Ta);
Chris@42 910 Tk = VBYI(VADD(Tf, Tc));
Chris@42 911 Tl = BYTWJ(&(W[TWVL * 8]), VSUB(Tj, Tk));
Chris@42 912 Tm = BYTWJ(&(W[TWVL * 4]), VADD(Tj, Tk));
Chris@42 913 ST(&(x[WS(vs, 5)]), Tl, ms, &(x[WS(vs, 5)]));
Chris@42 914 ST(&(x[WS(vs, 3)]), Tm, ms, &(x[WS(vs, 3)]));
Chris@42 915 }
Chris@42 916 {
Chris@42 917 V T2t, T2u, T2r, T2s;
Chris@42 918 T2r = VSUB(T2b, T2i);
Chris@42 919 T2s = VBYI(VADD(T2n, T2k));
Chris@42 920 T2t = BYTWJ(&(W[TWVL * 8]), VSUB(T2r, T2s));
Chris@42 921 T2u = BYTWJ(&(W[TWVL * 4]), VADD(T2r, T2s));
Chris@42 922 ST(&(x[WS(vs, 5) + WS(rs, 4)]), T2t, ms, &(x[WS(vs, 5)]));
Chris@42 923 ST(&(x[WS(vs, 3) + WS(rs, 4)]), T2u, ms, &(x[WS(vs, 3)]));
Chris@42 924 }
Chris@42 925 {
Chris@42 926 V T3x, T3y, T3v, T3w;
Chris@42 927 T3v = VSUB(T3f, T3m);
Chris@42 928 T3w = VBYI(VADD(T3r, T3o));
Chris@42 929 T3x = BYTWJ(&(W[TWVL * 8]), VSUB(T3v, T3w));
Chris@42 930 T3y = BYTWJ(&(W[TWVL * 4]), VADD(T3v, T3w));
Chris@42 931 ST(&(x[WS(vs, 5) + WS(rs, 6)]), T3x, ms, &(x[WS(vs, 5)]));
Chris@42 932 ST(&(x[WS(vs, 3) + WS(rs, 6)]), T3y, ms, &(x[WS(vs, 3)]));
Chris@42 933 }
Chris@42 934 {
Chris@42 935 V TS, TT, TQ, TR;
Chris@42 936 TQ = VSUB(TA, TH);
Chris@42 937 TR = VBYI(VADD(TM, TJ));
Chris@42 938 TS = BYTWJ(&(W[TWVL * 8]), VSUB(TQ, TR));
Chris@42 939 TT = BYTWJ(&(W[TWVL * 4]), VADD(TQ, TR));
Chris@42 940 ST(&(x[WS(vs, 5) + WS(rs, 1)]), TS, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 941 ST(&(x[WS(vs, 3) + WS(rs, 1)]), TT, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 942 }
Chris@42 943 {
Chris@42 944 V T1W, T1X, T1U, T1V;
Chris@42 945 T1U = VSUB(T1E, T1L);
Chris@42 946 T1V = VBYI(VADD(T1Q, T1N));
Chris@42 947 T1W = BYTWJ(&(W[TWVL * 8]), VSUB(T1U, T1V));
Chris@42 948 T1X = BYTWJ(&(W[TWVL * 4]), VADD(T1U, T1V));
Chris@42 949 ST(&(x[WS(vs, 5) + WS(rs, 3)]), T1W, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 950 ST(&(x[WS(vs, 3) + WS(rs, 3)]), T1X, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 951 }
Chris@42 952 {
Chris@42 953 V T30, T31, T2Y, T2Z;
Chris@42 954 T2Y = VSUB(T2I, T2P);
Chris@42 955 T2Z = VBYI(VADD(T2U, T2R));
Chris@42 956 T30 = BYTWJ(&(W[TWVL * 8]), VSUB(T2Y, T2Z));
Chris@42 957 T31 = BYTWJ(&(W[TWVL * 4]), VADD(T2Y, T2Z));
Chris@42 958 ST(&(x[WS(vs, 5) + WS(rs, 5)]), T30, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 959 ST(&(x[WS(vs, 3) + WS(rs, 5)]), T31, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 960 }
Chris@42 961 {
Chris@42 962 V T44, T45, T42, T43;
Chris@42 963 T42 = VSUB(T3M, T3T);
Chris@42 964 T43 = VBYI(VADD(T3Y, T3V));
Chris@42 965 T44 = BYTWJ(&(W[TWVL * 8]), VSUB(T42, T43));
Chris@42 966 T45 = BYTWJ(&(W[TWVL * 4]), VADD(T42, T43));
Chris@42 967 ST(&(x[WS(vs, 5) + WS(rs, 7)]), T44, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 968 ST(&(x[WS(vs, 3) + WS(rs, 7)]), T45, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 969 }
Chris@42 970 }
Chris@42 971 }
Chris@42 972 VLEAVE();
Chris@42 973 }
Chris@42 974
Chris@42 975 static const tw_instr twinstr[] = {
Chris@42 976 VTW(0, 1),
Chris@42 977 VTW(0, 2),
Chris@42 978 VTW(0, 3),
Chris@42 979 VTW(0, 4),
Chris@42 980 VTW(0, 5),
Chris@42 981 VTW(0, 6),
Chris@42 982 VTW(0, 7),
Chris@42 983 {TW_NEXT, VL, 0}
Chris@42 984 };
Chris@42 985
Chris@42 986 static const ct_desc desc = { 8, XSIMD_STRING("q1fv_8"), twinstr, &GENUS, {264, 128, 0, 0}, 0, 0, 0 };
Chris@42 987
Chris@42 988 void XSIMD(codelet_q1fv_8) (planner *p) {
Chris@42 989 X(kdft_difsq_register) (p, q1fv_8, &desc);
Chris@42 990 }
Chris@42 991 #endif /* HAVE_FMA */