Mercurial > hg > sv-dependency-builds

annotate src/fftw-3.3.5/dft/simd/common/q1bv_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 16:45:30 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 q1bv_8 -include q1b.h -sign 1 */
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 * 121 stack variables, 1 constants, and 128 memory accesses
Chris@42 34 */
Chris@42 35 #include "q1b.h"
Chris@42 36
Chris@42 37 static void q1bv_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 = ii;
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, T45, T44;
Chris@42 46 {
Chris@42 47 V T3, Te, T1E, T1P, Tv, Tp, T26, T20, T2b, T2m, T3M, T2x, T2D, T3X, TA;
Chris@42 48 V TL, T48, T4e, T17, T12, TW, T1i, T2I, T1z, T1t, T2T, T3f, T3q, T34, T3a;
Chris@42 49 V T3H, T3B, Ts, Tw, Tf, Ta, T23, T27, T1Q, T1L, T2A, T2E, T2n, T2i, T4b;
Chris@42 50 V T4f, T3Y, T3T, TZ, T13, TM, TH, T35, T2L, T3j, T1w, T1A, 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 Tv = VADD(Tn, To);
Chris@42 81 Tp = VSUB(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 T26 = VADD(T1Y, T1Z);
Chris@42 85 T20 = VSUB(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 = VSUB(T2v, T2w);
Chris@42 104 T2D = VADD(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 = VSUB(T46, T47);
Chris@42 116 T4e = VADD(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 T12 = VADD(TU, TV);
Chris@42 126 TW = VSUB(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 T1z = VADD(T1r, T1s);
Chris@42 138 T1t = VSUB(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 = VSUB(T32, T33);
Chris@42 158 T3a = VADD(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 T3H = VADD(T3z, T3A);
Chris@42 163 T3B = VSUB(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 = VSUB(Tq, Tr);
Chris@42 179 Tw = VADD(Tq, Tr);
Chris@42 180 Tf = VSUB(T6, T9);
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 = VSUB(T21, T22);
Chris@42 191 T27 = VADD(T21, T22);
Chris@42 192 T1Q = VSUB(T1H, T1K);
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 = VSUB(T2y, T2z);
Chris@42 209 T2E = VADD(T2y, T2z);
Chris@42 210 T2n = VSUB(T2e, T2h);
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 = VSUB(T49, T4a);
Chris@42 221 T4f = VADD(T49, T4a);
Chris@42 222 T3Y = VSUB(T3P, T3S);
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 = VSUB(TX, TY);
Chris@42 236 T13 = VADD(TX, TY);
Chris@42 237 TM = VSUB(TD, TG);
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 = VSUB(T1u, T1v);
Chris@42 248 T1A = VADD(T1u, T1v);
Chris@42 249 T1j = VSUB(T1a, T1d);
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 T3b, T2U, T2P, T3I, T3r, T3m, T11, T25, T39, T4d;
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(Tv, Tw), ms, &(x[0]));
Chris@42 267 ST(&(x[WS(rs, 2)]), VADD(T1z, T1A), ms, &(x[0]));
Chris@42 268 ST(&(x[WS(rs, 7)]), VADD(T4e, T4f), ms, &(x[WS(rs, 1)]));
Chris@42 269 T37 = VSUB(T35, T36);
Chris@42 270 T3b = VADD(T35, T36);
Chris@42 271 T2U = VSUB(T2L, T2O);
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(T2D, T2E), ms, &(x[0]));
Chris@42 276 ST(&(x[WS(rs, 3)]), VADD(T26, T27), ms, &(x[WS(rs, 1)]));
Chris@42 277 ST(&(x[WS(rs, 1)]), VADD(T12, T13), ms, &(x[WS(rs, 1)]));
Chris@42 278 ST(&(x[WS(rs, 5)]), VADD(T3a, T3b), ms, &(x[WS(rs, 1)]));
Chris@42 279 Tt = BYTW(&(W[TWVL * 10]), VFNMSI(Ts, Tp));
Chris@42 280 T4c = BYTW(&(W[TWVL * 10]), VFNMSI(T4b, T48));
Chris@42 281 T3E = VSUB(T3C, T3D);
Chris@42 282 T3I = VADD(T3C, T3D);
Chris@42 283 T3r = VSUB(T3i, T3l);
Chris@42 284 T3m = VADD(T3i, T3l);
Chris@42 285 T2B = BYTW(&(W[TWVL * 10]), VFNMSI(T2A, T2x));
Chris@42 286 T24 = BYTW(&(W[TWVL * 10]), VFNMSI(T23, T20));
Chris@42 287 ST(&(x[WS(vs, 6)]), Tt, ms, &(x[WS(vs, 6)]));
Chris@42 288 ST(&(x[WS(vs, 6) + WS(rs, 7)]), T4c, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 289 }
Chris@42 290 {
Chris@42 291 V T38, T1y, Tu, T10, T1x, T3F, T2C, T3G;
Chris@42 292 T10 = BYTW(&(W[TWVL * 10]), VFNMSI(TZ, TW));
Chris@42 293 ST(&(x[WS(rs, 6)]), VADD(T3H, T3I), ms, &(x[0]));
Chris@42 294 T1x = BYTW(&(W[TWVL * 10]), VFNMSI(T1w, T1t));
Chris@42 295 T3F = BYTW(&(W[TWVL * 10]), VFNMSI(T3E, T3B));
Chris@42 296 ST(&(x[WS(vs, 6) + WS(rs, 4)]), T2B, ms, &(x[WS(vs, 6)]));
Chris@42 297 ST(&(x[WS(vs, 6) + WS(rs, 3)]), T24, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 298 T38 = BYTW(&(W[TWVL * 10]), VFNMSI(T37, T34));
Chris@42 299 T1y = BYTW(&(W[TWVL * 2]), VFMAI(T1w, T1t));
Chris@42 300 ST(&(x[WS(vs, 6) + WS(rs, 1)]), T10, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 301 Tu = BYTW(&(W[TWVL * 2]), VFMAI(Ts, Tp));
Chris@42 302 ST(&(x[WS(vs, 6) + WS(rs, 2)]), T1x, ms, &(x[WS(vs, 6)]));
Chris@42 303 ST(&(x[WS(vs, 6) + WS(rs, 6)]), T3F, ms, &(x[WS(vs, 6)]));
Chris@42 304 T2C = BYTW(&(W[TWVL * 2]), VFMAI(T2A, T2x));
Chris@42 305 T3G = BYTW(&(W[TWVL * 2]), VFMAI(T3E, T3B));
Chris@42 306 ST(&(x[WS(vs, 6) + WS(rs, 5)]), T38, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 307 ST(&(x[WS(vs, 2) + WS(rs, 2)]), T1y, ms, &(x[WS(vs, 2)]));
Chris@42 308 T11 = BYTW(&(W[TWVL * 2]), VFMAI(TZ, TW));
Chris@42 309 ST(&(x[WS(vs, 2)]), Tu, ms, &(x[WS(vs, 2)]));
Chris@42 310 T25 = BYTW(&(W[TWVL * 2]), VFMAI(T23, T20));
Chris@42 311 T39 = BYTW(&(W[TWVL * 2]), VFMAI(T37, T34));
Chris@42 312 ST(&(x[WS(vs, 2) + WS(rs, 4)]), T2C, ms, &(x[WS(vs, 2)]));
Chris@42 313 ST(&(x[WS(vs, 2) + WS(rs, 6)]), T3G, ms, &(x[WS(vs, 2)]));
Chris@42 314 T4d = BYTW(&(W[TWVL * 2]), VFMAI(T4b, T48));
Chris@42 315 }
Chris@42 316 }
Chris@42 317 {
Chris@42 318 V Tj, Tk, T2r, T2j, T2o, T2s, Ti, Th, T1M, T1R, T41, T40;
Chris@42 319 {
Chris@42 320 V T3c, T4g, T3J, T2F, Tx, T1B;
Chris@42 321 Tx = BYTW(&(W[TWVL * 6]), VSUB(Tv, Tw));
Chris@42 322 ST(&(x[WS(vs, 2) + WS(rs, 1)]), T11, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 323 T1B = BYTW(&(W[TWVL * 6]), VSUB(T1z, T1A));
Chris@42 324 ST(&(x[WS(vs, 2) + WS(rs, 3)]), T25, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 325 ST(&(x[WS(vs, 2) + WS(rs, 5)]), T39, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 326 T3c = BYTW(&(W[TWVL * 6]), VSUB(T3a, T3b));
Chris@42 327 T4g = BYTW(&(W[TWVL * 6]), VSUB(T4e, T4f));
Chris@42 328 ST(&(x[WS(vs, 2) + WS(rs, 7)]), T4d, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 329 ST(&(x[WS(vs, 4)]), Tx, ms, &(x[WS(vs, 4)]));
Chris@42 330 T3J = BYTW(&(W[TWVL * 6]), VSUB(T3H, T3I));
Chris@42 331 ST(&(x[WS(vs, 4) + WS(rs, 2)]), T1B, ms, &(x[WS(vs, 4)]));
Chris@42 332 T2F = BYTW(&(W[TWVL * 6]), VSUB(T2D, T2E));
Chris@42 333 {
Chris@42 334 V T14, Tb, Tg, T28, T3U, T3Z;
Chris@42 335 T28 = BYTW(&(W[TWVL * 6]), VSUB(T26, T27));
Chris@42 336 ST(&(x[WS(vs, 4) + WS(rs, 5)]), T3c, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 337 ST(&(x[WS(vs, 4) + WS(rs, 7)]), T4g, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 338 T14 = BYTW(&(W[TWVL * 6]), VSUB(T12, T13));
Chris@42 339 Tj = VFMA(LDK(KP707106781), Ta, T3);
Chris@42 340 Tb = VFNMS(LDK(KP707106781), Ta, T3);
Chris@42 341 ST(&(x[WS(vs, 4) + WS(rs, 6)]), T3J, ms, &(x[WS(vs, 4)]));
Chris@42 342 Tk = VFMA(LDK(KP707106781), Tf, Te);
Chris@42 343 Tg = VFNMS(LDK(KP707106781), Tf, Te);
Chris@42 344 ST(&(x[WS(vs, 4) + WS(rs, 4)]), T2F, ms, &(x[WS(vs, 4)]));
Chris@42 345 ST(&(x[WS(vs, 4) + WS(rs, 3)]), T28, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 346 T3U = VFNMS(LDK(KP707106781), T3T, T3M);
Chris@42 347 T42 = VFMA(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, 4) + WS(rs, 1)]), T14, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 351 T2r = VFMA(LDK(KP707106781), T2i, T2b);
Chris@42 352 T2j = VFNMS(LDK(KP707106781), T2i, T2b);
Chris@42 353 T2o = VFNMS(LDK(KP707106781), T2n, T2m);
Chris@42 354 T2s = VFMA(LDK(KP707106781), T2n, T2m);
Chris@42 355 Ti = BYTW(&(W[TWVL * 8]), VFMAI(Tg, Tb));
Chris@42 356 Th = BYTW(&(W[TWVL * 4]), VFNMSI(Tg, Tb));
Chris@42 357 T1U = VFMA(LDK(KP707106781), T1L, T1E);
Chris@42 358 T1M = VFNMS(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 = BYTW(&(W[TWVL * 8]), VFMAI(T3Z, T3U));
Chris@42 362 T40 = BYTW(&(W[TWVL * 4]), 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 TI, TN, T1f, T1k, T3n, T3s;
Chris@42 369 {
Chris@42 370 V T1T, T1S, T2q, T2p;
Chris@42 371 TQ = VFMA(LDK(KP707106781), TH, TA);
Chris@42 372 TI = VFNMS(LDK(KP707106781), TH, TA);
Chris@42 373 T2q = BYTW(&(W[TWVL * 8]), VFMAI(T2o, T2j));
Chris@42 374 T2p = BYTW(&(W[TWVL * 4]), VFNMSI(T2o, T2j));
Chris@42 375 ST(&(x[WS(vs, 5)]), Ti, ms, &(x[WS(vs, 5)]));
Chris@42 376 ST(&(x[WS(vs, 3)]), Th, ms, &(x[WS(vs, 3)]));
Chris@42 377 T1T = BYTW(&(W[TWVL * 8]), VFMAI(T1R, T1M));
Chris@42 378 T1S = BYTW(&(W[TWVL * 4]), VFNMSI(T1R, T1M));
Chris@42 379 ST(&(x[WS(vs, 5) + WS(rs, 7)]), T41, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 380 ST(&(x[WS(vs, 3) + WS(rs, 7)]), T40, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 381 ST(&(x[WS(vs, 5) + WS(rs, 4)]), T2q, ms, &(x[WS(vs, 5)]));
Chris@42 382 ST(&(x[WS(vs, 3) + WS(rs, 4)]), T2p, ms, &(x[WS(vs, 3)]));
Chris@42 383 TN = VFNMS(LDK(KP707106781), TM, TL);
Chris@42 384 TR = VFMA(LDK(KP707106781), TM, TL);
Chris@42 385 T1n = VFMA(LDK(KP707106781), T1e, T17);
Chris@42 386 T1f = VFNMS(LDK(KP707106781), T1e, T17);
Chris@42 387 ST(&(x[WS(vs, 5) + WS(rs, 3)]), T1T, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 388 ST(&(x[WS(vs, 3) + WS(rs, 3)]), T1S, ms, &(x[WS(vs, 3) + 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 = VFMA(LDK(KP707106781), T3m, T3f);
Chris@42 392 T3n = VFNMS(LDK(KP707106781), T3m, T3f);
Chris@42 393 T3s = VFNMS(LDK(KP707106781), T3r, T3q);
Chris@42 394 T3w = VFMA(LDK(KP707106781), T3r, T3q);
Chris@42 395 }
Chris@42 396 {
Chris@42 397 V T2Q, TP, TO, T2V, T2X, T2W;
Chris@42 398 T2Y = VFMA(LDK(KP707106781), T2P, T2I);
Chris@42 399 T2Q = VFNMS(LDK(KP707106781), T2P, T2I);
Chris@42 400 TP = BYTW(&(W[TWVL * 8]), VFMAI(TN, TI));
Chris@42 401 TO = BYTW(&(W[TWVL * 4]), VFNMSI(TN, TI));
Chris@42 402 T2V = VFNMS(LDK(KP707106781), T2U, T2T);
Chris@42 403 T2Z = VFMA(LDK(KP707106781), T2U, T2T);
Chris@42 404 {
Chris@42 405 V T1m, T1l, T3u, T3t;
Chris@42 406 T1m = BYTW(&(W[TWVL * 8]), VFMAI(T1k, T1f));
Chris@42 407 T1l = BYTW(&(W[TWVL * 4]), VFNMSI(T1k, T1f));
Chris@42 408 T3u = BYTW(&(W[TWVL * 8]), VFMAI(T3s, T3n));
Chris@42 409 T3t = BYTW(&(W[TWVL * 4]), VFNMSI(T3s, T3n));
Chris@42 410 ST(&(x[WS(vs, 5) + WS(rs, 1)]), TP, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 411 ST(&(x[WS(vs, 3) + WS(rs, 1)]), TO, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 412 T2X = BYTW(&(W[TWVL * 8]), VFMAI(T2V, T2Q));
Chris@42 413 T2W = BYTW(&(W[TWVL * 4]), VFNMSI(T2V, T2Q));
Chris@42 414 ST(&(x[WS(vs, 5) + WS(rs, 2)]), T1m, ms, &(x[WS(vs, 5)]));
Chris@42 415 ST(&(x[WS(vs, 3) + WS(rs, 2)]), T1l, ms, &(x[WS(vs, 3)]));
Chris@42 416 ST(&(x[WS(vs, 5) + WS(rs, 6)]), T3u, ms, &(x[WS(vs, 5)]));
Chris@42 417 ST(&(x[WS(vs, 3) + WS(rs, 6)]), T3t, ms, &(x[WS(vs, 3)]));
Chris@42 418 }
Chris@42 419 ST(&(x[WS(vs, 5) + WS(rs, 5)]), T2X, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 420 ST(&(x[WS(vs, 3) + WS(rs, 5)]), T2W, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 421 }
Chris@42 422 }
Chris@42 423 {
Chris@42 424 V T3y, T3x, T1q, T1p;
Chris@42 425 T1q = BYTW(&(W[TWVL * 12]), VFNMSI(T1o, T1n));
Chris@42 426 T1p = BYTW(&(W[0]), VFMAI(T1o, T1n));
Chris@42 427 {
Chris@42 428 V Tm, Tl, T2u, T2t;
Chris@42 429 Tm = BYTW(&(W[TWVL * 12]), VFNMSI(Tk, Tj));
Chris@42 430 Tl = BYTW(&(W[0]), VFMAI(Tk, Tj));
Chris@42 431 T2u = BYTW(&(W[TWVL * 12]), VFNMSI(T2s, T2r));
Chris@42 432 T2t = BYTW(&(W[0]), VFMAI(T2s, T2r));
Chris@42 433 ST(&(x[WS(vs, 7) + WS(rs, 2)]), T1q, ms, &(x[WS(vs, 7)]));
Chris@42 434 ST(&(x[WS(vs, 1) + WS(rs, 2)]), T1p, ms, &(x[WS(vs, 1)]));
Chris@42 435 T3y = BYTW(&(W[TWVL * 12]), VFNMSI(T3w, T3v));
Chris@42 436 T3x = BYTW(&(W[0]), VFMAI(T3w, T3v));
Chris@42 437 ST(&(x[WS(vs, 7)]), Tm, ms, &(x[WS(vs, 7)]));
Chris@42 438 ST(&(x[WS(vs, 1)]), Tl, ms, &(x[WS(vs, 1)]));
Chris@42 439 ST(&(x[WS(vs, 7) + WS(rs, 4)]), T2u, ms, &(x[WS(vs, 7)]));
Chris@42 440 ST(&(x[WS(vs, 1) + WS(rs, 4)]), T2t, ms, &(x[WS(vs, 1)]));
Chris@42 441 }
Chris@42 442 ST(&(x[WS(vs, 7) + WS(rs, 6)]), T3y, ms, &(x[WS(vs, 7)]));
Chris@42 443 ST(&(x[WS(vs, 1) + WS(rs, 6)]), T3x, ms, &(x[WS(vs, 1)]));
Chris@42 444 TT = BYTW(&(W[TWVL * 12]), VFNMSI(TR, TQ));
Chris@42 445 TS = BYTW(&(W[0]), VFMAI(TR, TQ));
Chris@42 446 }
Chris@42 447 }
Chris@42 448 }
Chris@42 449 }
Chris@42 450 }
Chris@42 451 {
Chris@42 452 V T1X, T1W, T31, T30;
Chris@42 453 T1X = BYTW(&(W[TWVL * 12]), VFNMSI(T1V, T1U));
Chris@42 454 T1W = BYTW(&(W[0]), VFMAI(T1V, T1U));
Chris@42 455 T31 = BYTW(&(W[TWVL * 12]), VFNMSI(T2Z, T2Y));
Chris@42 456 T30 = BYTW(&(W[0]), VFMAI(T2Z, T2Y));
Chris@42 457 ST(&(x[WS(vs, 7) + WS(rs, 1)]), TT, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 458 ST(&(x[WS(vs, 1) + WS(rs, 1)]), TS, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 459 T45 = BYTW(&(W[TWVL * 12]), VFNMSI(T43, T42));
Chris@42 460 T44 = BYTW(&(W[0]), VFMAI(T43, T42));
Chris@42 461 ST(&(x[WS(vs, 7) + WS(rs, 3)]), T1X, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 462 ST(&(x[WS(vs, 1) + WS(rs, 3)]), T1W, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 463 ST(&(x[WS(vs, 7) + WS(rs, 5)]), T31, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 464 ST(&(x[WS(vs, 1) + WS(rs, 5)]), T30, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 465 }
Chris@42 466 ST(&(x[WS(vs, 7) + WS(rs, 7)]), T45, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 467 ST(&(x[WS(vs, 1) + WS(rs, 7)]), T44, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 468 }
Chris@42 469 }
Chris@42 470 VLEAVE();
Chris@42 471 }
Chris@42 472
Chris@42 473 static const tw_instr twinstr[] = {
Chris@42 474 VTW(0, 1),
Chris@42 475 VTW(0, 2),
Chris@42 476 VTW(0, 3),
Chris@42 477 VTW(0, 4),
Chris@42 478 VTW(0, 5),
Chris@42 479 VTW(0, 6),
Chris@42 480 VTW(0, 7),
Chris@42 481 {TW_NEXT, VL, 0}
Chris@42 482 };
Chris@42 483
Chris@42 484 static const ct_desc desc = { 8, XSIMD_STRING("q1bv_8"), twinstr, &GENUS, {184, 112, 80, 0}, 0, 0, 0 };
Chris@42 485
Chris@42 486 void XSIMD(codelet_q1bv_8) (planner *p) {
Chris@42 487 X(kdft_difsq_register) (p, q1bv_8, &desc);
Chris@42 488 }
Chris@42 489 #else /* HAVE_FMA */
Chris@42 490
Chris@42 491 /* Generated by: ../../../genfft/gen_twidsq_c.native -simd -compact -variables 4 -pipeline-latency 8 -n 8 -dif -name q1bv_8 -include q1b.h -sign 1 */
Chris@42 492
Chris@42 493 /*
Chris@42 494 * This function contains 264 FP additions, 128 FP multiplications,
Chris@42 495 * (or, 264 additions, 128 multiplications, 0 fused multiply/add),
Chris@42 496 * 77 stack variables, 1 constants, and 128 memory accesses
Chris@42 497 */
Chris@42 498 #include "q1b.h"
Chris@42 499
Chris@42 500 static void q1bv_8(R *ri, R *ii, const R *W, stride rs, stride vs, INT mb, INT me, INT ms)
Chris@42 501 {
Chris@42 502 DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
Chris@42 503 {
Chris@42 504 INT m;
Chris@42 505 R *x;
Chris@42 506 x = ii;
Chris@42 507 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@42 508 V Ta, Tv, Te, Tp, T1L, T26, T1P, T20, T2i, T2D, T2m, T2x, T3T, T4e, T3X;
Chris@42 509 V T48, TH, T12, TL, TW, T1e, T1z, T1i, T1t, T2P, T3a, T2T, T34, T3m, T3H;
Chris@42 510 V T3q, T3B, T7, Tw, Tf, Ts, T1I, T27, T1Q, T23, T2f, T2E, T2n, T2A, T3Q;
Chris@42 511 V T4f, T3Y, T4b, TE, T13, TM, TZ, T1b, T1A, T1j, T1w, T2M, T3b, T2U, T37;
Chris@42 512 V T3j, T3I, T3r, T3E, T28, T14;
Chris@42 513 {
Chris@42 514 V T8, T9, To, Tc, Td, Tn;
Chris@42 515 T8 = LD(&(x[WS(rs, 2)]), ms, &(x[0]));
Chris@42 516 T9 = LD(&(x[WS(rs, 6)]), ms, &(x[0]));
Chris@42 517 To = VADD(T8, T9);
Chris@42 518 Tc = LD(&(x[0]), ms, &(x[0]));
Chris@42 519 Td = LD(&(x[WS(rs, 4)]), ms, &(x[0]));
Chris@42 520 Tn = VADD(Tc, Td);
Chris@42 521 Ta = VSUB(T8, T9);
Chris@42 522 Tv = VADD(Tn, To);
Chris@42 523 Te = VSUB(Tc, Td);
Chris@42 524 Tp = VSUB(Tn, To);
Chris@42 525 }
Chris@42 526 {
Chris@42 527 V T1J, T1K, T1Z, T1N, T1O, T1Y;
Chris@42 528 T1J = LD(&(x[WS(vs, 3) + WS(rs, 2)]), ms, &(x[WS(vs, 3)]));
Chris@42 529 T1K = LD(&(x[WS(vs, 3) + WS(rs, 6)]), ms, &(x[WS(vs, 3)]));
Chris@42 530 T1Z = VADD(T1J, T1K);
Chris@42 531 T1N = LD(&(x[WS(vs, 3)]), ms, &(x[WS(vs, 3)]));
Chris@42 532 T1O = LD(&(x[WS(vs, 3) + WS(rs, 4)]), ms, &(x[WS(vs, 3)]));
Chris@42 533 T1Y = VADD(T1N, T1O);
Chris@42 534 T1L = VSUB(T1J, T1K);
Chris@42 535 T26 = VADD(T1Y, T1Z);
Chris@42 536 T1P = VSUB(T1N, T1O);
Chris@42 537 T20 = VSUB(T1Y, T1Z);
Chris@42 538 }
Chris@42 539 {
Chris@42 540 V T2g, T2h, T2w, T2k, T2l, T2v;
Chris@42 541 T2g = LD(&(x[WS(vs, 4) + WS(rs, 2)]), ms, &(x[WS(vs, 4)]));
Chris@42 542 T2h = LD(&(x[WS(vs, 4) + WS(rs, 6)]), ms, &(x[WS(vs, 4)]));
Chris@42 543 T2w = VADD(T2g, T2h);
Chris@42 544 T2k = LD(&(x[WS(vs, 4)]), ms, &(x[WS(vs, 4)]));
Chris@42 545 T2l = LD(&(x[WS(vs, 4) + WS(rs, 4)]), ms, &(x[WS(vs, 4)]));
Chris@42 546 T2v = VADD(T2k, T2l);
Chris@42 547 T2i = VSUB(T2g, T2h);
Chris@42 548 T2D = VADD(T2v, T2w);
Chris@42 549 T2m = VSUB(T2k, T2l);
Chris@42 550 T2x = VSUB(T2v, T2w);
Chris@42 551 }
Chris@42 552 {
Chris@42 553 V T3R, T3S, T47, T3V, T3W, T46;
Chris@42 554 T3R = LD(&(x[WS(vs, 7) + WS(rs, 2)]), ms, &(x[WS(vs, 7)]));
Chris@42 555 T3S = LD(&(x[WS(vs, 7) + WS(rs, 6)]), ms, &(x[WS(vs, 7)]));
Chris@42 556 T47 = VADD(T3R, T3S);
Chris@42 557 T3V = LD(&(x[WS(vs, 7)]), ms, &(x[WS(vs, 7)]));
Chris@42 558 T3W = LD(&(x[WS(vs, 7) + WS(rs, 4)]), ms, &(x[WS(vs, 7)]));
Chris@42 559 T46 = VADD(T3V, T3W);
Chris@42 560 T3T = VSUB(T3R, T3S);
Chris@42 561 T4e = VADD(T46, T47);
Chris@42 562 T3X = VSUB(T3V, T3W);
Chris@42 563 T48 = VSUB(T46, T47);
Chris@42 564 }
Chris@42 565 {
Chris@42 566 V TF, TG, TV, TJ, TK, TU;
Chris@42 567 TF = LD(&(x[WS(vs, 1) + WS(rs, 2)]), ms, &(x[WS(vs, 1)]));
Chris@42 568 TG = LD(&(x[WS(vs, 1) + WS(rs, 6)]), ms, &(x[WS(vs, 1)]));
Chris@42 569 TV = VADD(TF, TG);
Chris@42 570 TJ = LD(&(x[WS(vs, 1)]), ms, &(x[WS(vs, 1)]));
Chris@42 571 TK = LD(&(x[WS(vs, 1) + WS(rs, 4)]), ms, &(x[WS(vs, 1)]));
Chris@42 572 TU = VADD(TJ, TK);
Chris@42 573 TH = VSUB(TF, TG);
Chris@42 574 T12 = VADD(TU, TV);
Chris@42 575 TL = VSUB(TJ, TK);
Chris@42 576 TW = VSUB(TU, TV);
Chris@42 577 }
Chris@42 578 {
Chris@42 579 V T1c, T1d, T1s, T1g, T1h, T1r;
Chris@42 580 T1c = LD(&(x[WS(vs, 2) + WS(rs, 2)]), ms, &(x[WS(vs, 2)]));
Chris@42 581 T1d = LD(&(x[WS(vs, 2) + WS(rs, 6)]), ms, &(x[WS(vs, 2)]));
Chris@42 582 T1s = VADD(T1c, T1d);
Chris@42 583 T1g = LD(&(x[WS(vs, 2)]), ms, &(x[WS(vs, 2)]));
Chris@42 584 T1h = LD(&(x[WS(vs, 2) + WS(rs, 4)]), ms, &(x[WS(vs, 2)]));
Chris@42 585 T1r = VADD(T1g, T1h);
Chris@42 586 T1e = VSUB(T1c, T1d);
Chris@42 587 T1z = VADD(T1r, T1s);
Chris@42 588 T1i = VSUB(T1g, T1h);
Chris@42 589 T1t = VSUB(T1r, T1s);
Chris@42 590 }
Chris@42 591 {
Chris@42 592 V T2N, T2O, T33, T2R, T2S, T32;
Chris@42 593 T2N = LD(&(x[WS(vs, 5) + WS(rs, 2)]), ms, &(x[WS(vs, 5)]));
Chris@42 594 T2O = LD(&(x[WS(vs, 5) + WS(rs, 6)]), ms, &(x[WS(vs, 5)]));
Chris@42 595 T33 = VADD(T2N, T2O);
Chris@42 596 T2R = LD(&(x[WS(vs, 5)]), ms, &(x[WS(vs, 5)]));
Chris@42 597 T2S = LD(&(x[WS(vs, 5) + WS(rs, 4)]), ms, &(x[WS(vs, 5)]));
Chris@42 598 T32 = VADD(T2R, T2S);
Chris@42 599 T2P = VSUB(T2N, T2O);
Chris@42 600 T3a = VADD(T32, T33);
Chris@42 601 T2T = VSUB(T2R, T2S);
Chris@42 602 T34 = VSUB(T32, T33);
Chris@42 603 }
Chris@42 604 {
Chris@42 605 V T3k, T3l, T3A, T3o, T3p, T3z;
Chris@42 606 T3k = LD(&(x[WS(vs, 6) + WS(rs, 2)]), ms, &(x[WS(vs, 6)]));
Chris@42 607 T3l = LD(&(x[WS(vs, 6) + WS(rs, 6)]), ms, &(x[WS(vs, 6)]));
Chris@42 608 T3A = VADD(T3k, T3l);
Chris@42 609 T3o = LD(&(x[WS(vs, 6)]), ms, &(x[WS(vs, 6)]));
Chris@42 610 T3p = LD(&(x[WS(vs, 6) + WS(rs, 4)]), ms, &(x[WS(vs, 6)]));
Chris@42 611 T3z = VADD(T3o, T3p);
Chris@42 612 T3m = VSUB(T3k, T3l);
Chris@42 613 T3H = VADD(T3z, T3A);
Chris@42 614 T3q = VSUB(T3o, T3p);
Chris@42 615 T3B = VSUB(T3z, T3A);
Chris@42 616 }
Chris@42 617 {
Chris@42 618 V T3, Tq, T6, Tr;
Chris@42 619 {
Chris@42 620 V T1, T2, T4, T5;
Chris@42 621 T1 = LD(&(x[WS(rs, 1)]), ms, &(x[WS(rs, 1)]));
Chris@42 622 T2 = LD(&(x[WS(rs, 5)]), ms, &(x[WS(rs, 1)]));
Chris@42 623 T3 = VSUB(T1, T2);
Chris@42 624 Tq = VADD(T1, T2);
Chris@42 625 T4 = LD(&(x[WS(rs, 7)]), ms, &(x[WS(rs, 1)]));
Chris@42 626 T5 = LD(&(x[WS(rs, 3)]), ms, &(x[WS(rs, 1)]));
Chris@42 627 T6 = VSUB(T4, T5);
Chris@42 628 Tr = VADD(T4, T5);
Chris@42 629 }
Chris@42 630 T7 = VMUL(LDK(KP707106781), VSUB(T3, T6));
Chris@42 631 Tw = VADD(Tq, Tr);
Chris@42 632 Tf = VMUL(LDK(KP707106781), VADD(T3, T6));
Chris@42 633 Ts = VBYI(VSUB(Tq, Tr));
Chris@42 634 }
Chris@42 635 {
Chris@42 636 V T1E, T21, T1H, T22;
Chris@42 637 {
Chris@42 638 V T1C, T1D, T1F, T1G;
Chris@42 639 T1C = LD(&(x[WS(vs, 3) + WS(rs, 1)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 640 T1D = LD(&(x[WS(vs, 3) + WS(rs, 5)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 641 T1E = VSUB(T1C, T1D);
Chris@42 642 T21 = VADD(T1C, T1D);
Chris@42 643 T1F = LD(&(x[WS(vs, 3) + WS(rs, 7)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 644 T1G = LD(&(x[WS(vs, 3) + WS(rs, 3)]), ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 645 T1H = VSUB(T1F, T1G);
Chris@42 646 T22 = VADD(T1F, T1G);
Chris@42 647 }
Chris@42 648 T1I = VMUL(LDK(KP707106781), VSUB(T1E, T1H));
Chris@42 649 T27 = VADD(T21, T22);
Chris@42 650 T1Q = VMUL(LDK(KP707106781), VADD(T1E, T1H));
Chris@42 651 T23 = VBYI(VSUB(T21, T22));
Chris@42 652 }
Chris@42 653 {
Chris@42 654 V T2b, T2y, T2e, T2z;
Chris@42 655 {
Chris@42 656 V T29, T2a, T2c, T2d;
Chris@42 657 T29 = LD(&(x[WS(vs, 4) + WS(rs, 1)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 658 T2a = LD(&(x[WS(vs, 4) + WS(rs, 5)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 659 T2b = VSUB(T29, T2a);
Chris@42 660 T2y = VADD(T29, T2a);
Chris@42 661 T2c = LD(&(x[WS(vs, 4) + WS(rs, 7)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 662 T2d = LD(&(x[WS(vs, 4) + WS(rs, 3)]), ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 663 T2e = VSUB(T2c, T2d);
Chris@42 664 T2z = VADD(T2c, T2d);
Chris@42 665 }
Chris@42 666 T2f = VMUL(LDK(KP707106781), VSUB(T2b, T2e));
Chris@42 667 T2E = VADD(T2y, T2z);
Chris@42 668 T2n = VMUL(LDK(KP707106781), VADD(T2b, T2e));
Chris@42 669 T2A = VBYI(VSUB(T2y, T2z));
Chris@42 670 }
Chris@42 671 {
Chris@42 672 V T3M, T49, T3P, T4a;
Chris@42 673 {
Chris@42 674 V T3K, T3L, T3N, T3O;
Chris@42 675 T3K = LD(&(x[WS(vs, 7) + WS(rs, 1)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 676 T3L = LD(&(x[WS(vs, 7) + WS(rs, 5)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 677 T3M = VSUB(T3K, T3L);
Chris@42 678 T49 = VADD(T3K, T3L);
Chris@42 679 T3N = LD(&(x[WS(vs, 7) + WS(rs, 7)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 680 T3O = LD(&(x[WS(vs, 7) + WS(rs, 3)]), ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 681 T3P = VSUB(T3N, T3O);
Chris@42 682 T4a = VADD(T3N, T3O);
Chris@42 683 }
Chris@42 684 T3Q = VMUL(LDK(KP707106781), VSUB(T3M, T3P));
Chris@42 685 T4f = VADD(T49, T4a);
Chris@42 686 T3Y = VMUL(LDK(KP707106781), VADD(T3M, T3P));
Chris@42 687 T4b = VBYI(VSUB(T49, T4a));
Chris@42 688 }
Chris@42 689 {
Chris@42 690 V TA, TX, TD, TY;
Chris@42 691 {
Chris@42 692 V Ty, Tz, TB, TC;
Chris@42 693 Ty = LD(&(x[WS(vs, 1) + WS(rs, 1)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 694 Tz = LD(&(x[WS(vs, 1) + WS(rs, 5)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 695 TA = VSUB(Ty, Tz);
Chris@42 696 TX = VADD(Ty, Tz);
Chris@42 697 TB = LD(&(x[WS(vs, 1) + WS(rs, 7)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 698 TC = LD(&(x[WS(vs, 1) + WS(rs, 3)]), ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 699 TD = VSUB(TB, TC);
Chris@42 700 TY = VADD(TB, TC);
Chris@42 701 }
Chris@42 702 TE = VMUL(LDK(KP707106781), VSUB(TA, TD));
Chris@42 703 T13 = VADD(TX, TY);
Chris@42 704 TM = VMUL(LDK(KP707106781), VADD(TA, TD));
Chris@42 705 TZ = VBYI(VSUB(TX, TY));
Chris@42 706 }
Chris@42 707 {
Chris@42 708 V T17, T1u, T1a, T1v;
Chris@42 709 {
Chris@42 710 V T15, T16, T18, T19;
Chris@42 711 T15 = LD(&(x[WS(vs, 2) + WS(rs, 1)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 712 T16 = LD(&(x[WS(vs, 2) + WS(rs, 5)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 713 T17 = VSUB(T15, T16);
Chris@42 714 T1u = VADD(T15, T16);
Chris@42 715 T18 = LD(&(x[WS(vs, 2) + WS(rs, 7)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 716 T19 = LD(&(x[WS(vs, 2) + WS(rs, 3)]), ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 717 T1a = VSUB(T18, T19);
Chris@42 718 T1v = VADD(T18, T19);
Chris@42 719 }
Chris@42 720 T1b = VMUL(LDK(KP707106781), VSUB(T17, T1a));
Chris@42 721 T1A = VADD(T1u, T1v);
Chris@42 722 T1j = VMUL(LDK(KP707106781), VADD(T17, T1a));
Chris@42 723 T1w = VBYI(VSUB(T1u, T1v));
Chris@42 724 }
Chris@42 725 {
Chris@42 726 V T2I, T35, T2L, T36;
Chris@42 727 {
Chris@42 728 V T2G, T2H, T2J, T2K;
Chris@42 729 T2G = LD(&(x[WS(vs, 5) + WS(rs, 1)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 730 T2H = LD(&(x[WS(vs, 5) + WS(rs, 5)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 731 T2I = VSUB(T2G, T2H);
Chris@42 732 T35 = VADD(T2G, T2H);
Chris@42 733 T2J = LD(&(x[WS(vs, 5) + WS(rs, 7)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 734 T2K = LD(&(x[WS(vs, 5) + WS(rs, 3)]), ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 735 T2L = VSUB(T2J, T2K);
Chris@42 736 T36 = VADD(T2J, T2K);
Chris@42 737 }
Chris@42 738 T2M = VMUL(LDK(KP707106781), VSUB(T2I, T2L));
Chris@42 739 T3b = VADD(T35, T36);
Chris@42 740 T2U = VMUL(LDK(KP707106781), VADD(T2I, T2L));
Chris@42 741 T37 = VBYI(VSUB(T35, T36));
Chris@42 742 }
Chris@42 743 {
Chris@42 744 V T3f, T3C, T3i, T3D;
Chris@42 745 {
Chris@42 746 V T3d, T3e, T3g, T3h;
Chris@42 747 T3d = LD(&(x[WS(vs, 6) + WS(rs, 1)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 748 T3e = LD(&(x[WS(vs, 6) + WS(rs, 5)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 749 T3f = VSUB(T3d, T3e);
Chris@42 750 T3C = VADD(T3d, T3e);
Chris@42 751 T3g = LD(&(x[WS(vs, 6) + WS(rs, 7)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 752 T3h = LD(&(x[WS(vs, 6) + WS(rs, 3)]), ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 753 T3i = VSUB(T3g, T3h);
Chris@42 754 T3D = VADD(T3g, T3h);
Chris@42 755 }
Chris@42 756 T3j = VMUL(LDK(KP707106781), VSUB(T3f, T3i));
Chris@42 757 T3I = VADD(T3C, T3D);
Chris@42 758 T3r = VMUL(LDK(KP707106781), VADD(T3f, T3i));
Chris@42 759 T3E = VBYI(VSUB(T3C, T3D));
Chris@42 760 }
Chris@42 761 ST(&(x[0]), VADD(Tv, Tw), ms, &(x[0]));
Chris@42 762 ST(&(x[WS(rs, 2)]), VADD(T1z, T1A), ms, &(x[0]));
Chris@42 763 ST(&(x[WS(rs, 5)]), VADD(T3a, T3b), ms, &(x[WS(rs, 1)]));
Chris@42 764 ST(&(x[WS(rs, 7)]), VADD(T4e, T4f), ms, &(x[WS(rs, 1)]));
Chris@42 765 ST(&(x[WS(rs, 6)]), VADD(T3H, T3I), ms, &(x[0]));
Chris@42 766 ST(&(x[WS(rs, 4)]), VADD(T2D, T2E), ms, &(x[0]));
Chris@42 767 {
Chris@42 768 V Tt, T4c, T2B, T24;
Chris@42 769 ST(&(x[WS(rs, 3)]), VADD(T26, T27), ms, &(x[WS(rs, 1)]));
Chris@42 770 ST(&(x[WS(rs, 1)]), VADD(T12, T13), ms, &(x[WS(rs, 1)]));
Chris@42 771 Tt = BYTW(&(W[TWVL * 10]), VSUB(Tp, Ts));
Chris@42 772 ST(&(x[WS(vs, 6)]), Tt, ms, &(x[WS(vs, 6)]));
Chris@42 773 T4c = BYTW(&(W[TWVL * 10]), VSUB(T48, T4b));
Chris@42 774 ST(&(x[WS(vs, 6) + WS(rs, 7)]), T4c, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 775 T2B = BYTW(&(W[TWVL * 10]), VSUB(T2x, T2A));
Chris@42 776 ST(&(x[WS(vs, 6) + WS(rs, 4)]), T2B, ms, &(x[WS(vs, 6)]));
Chris@42 777 T24 = BYTW(&(W[TWVL * 10]), VSUB(T20, T23));
Chris@42 778 ST(&(x[WS(vs, 6) + WS(rs, 3)]), T24, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 779 }
Chris@42 780 {
Chris@42 781 V T10, T1x, T3F, T38, T1y, Tu;
Chris@42 782 T10 = BYTW(&(W[TWVL * 10]), VSUB(TW, TZ));
Chris@42 783 ST(&(x[WS(vs, 6) + WS(rs, 1)]), T10, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 784 T1x = BYTW(&(W[TWVL * 10]), VSUB(T1t, T1w));
Chris@42 785 ST(&(x[WS(vs, 6) + WS(rs, 2)]), T1x, ms, &(x[WS(vs, 6)]));
Chris@42 786 T3F = BYTW(&(W[TWVL * 10]), VSUB(T3B, T3E));
Chris@42 787 ST(&(x[WS(vs, 6) + WS(rs, 6)]), T3F, ms, &(x[WS(vs, 6)]));
Chris@42 788 T38 = BYTW(&(W[TWVL * 10]), VSUB(T34, T37));
Chris@42 789 ST(&(x[WS(vs, 6) + WS(rs, 5)]), T38, ms, &(x[WS(vs, 6) + WS(rs, 1)]));
Chris@42 790 T1y = BYTW(&(W[TWVL * 2]), VADD(T1t, T1w));
Chris@42 791 ST(&(x[WS(vs, 2) + WS(rs, 2)]), T1y, ms, &(x[WS(vs, 2)]));
Chris@42 792 Tu = BYTW(&(W[TWVL * 2]), VADD(Tp, Ts));
Chris@42 793 ST(&(x[WS(vs, 2)]), Tu, ms, &(x[WS(vs, 2)]));
Chris@42 794 }
Chris@42 795 {
Chris@42 796 V T2C, T3G, T11, T25, T39, T4d;
Chris@42 797 T2C = BYTW(&(W[TWVL * 2]), VADD(T2x, T2A));
Chris@42 798 ST(&(x[WS(vs, 2) + WS(rs, 4)]), T2C, ms, &(x[WS(vs, 2)]));
Chris@42 799 T3G = BYTW(&(W[TWVL * 2]), VADD(T3B, T3E));
Chris@42 800 ST(&(x[WS(vs, 2) + WS(rs, 6)]), T3G, ms, &(x[WS(vs, 2)]));
Chris@42 801 T11 = BYTW(&(W[TWVL * 2]), VADD(TW, TZ));
Chris@42 802 ST(&(x[WS(vs, 2) + WS(rs, 1)]), T11, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 803 T25 = BYTW(&(W[TWVL * 2]), VADD(T20, T23));
Chris@42 804 ST(&(x[WS(vs, 2) + WS(rs, 3)]), T25, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 805 T39 = BYTW(&(W[TWVL * 2]), VADD(T34, T37));
Chris@42 806 ST(&(x[WS(vs, 2) + WS(rs, 5)]), T39, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 807 T4d = BYTW(&(W[TWVL * 2]), VADD(T48, T4b));
Chris@42 808 ST(&(x[WS(vs, 2) + WS(rs, 7)]), T4d, ms, &(x[WS(vs, 2) + WS(rs, 1)]));
Chris@42 809 }
Chris@42 810 {
Chris@42 811 V Tx, T1B, T3c, T4g, T3J, T2F;
Chris@42 812 Tx = BYTW(&(W[TWVL * 6]), VSUB(Tv, Tw));
Chris@42 813 ST(&(x[WS(vs, 4)]), Tx, ms, &(x[WS(vs, 4)]));
Chris@42 814 T1B = BYTW(&(W[TWVL * 6]), VSUB(T1z, T1A));
Chris@42 815 ST(&(x[WS(vs, 4) + WS(rs, 2)]), T1B, ms, &(x[WS(vs, 4)]));
Chris@42 816 T3c = BYTW(&(W[TWVL * 6]), VSUB(T3a, T3b));
Chris@42 817 ST(&(x[WS(vs, 4) + WS(rs, 5)]), T3c, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 818 T4g = BYTW(&(W[TWVL * 6]), VSUB(T4e, T4f));
Chris@42 819 ST(&(x[WS(vs, 4) + WS(rs, 7)]), T4g, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 820 T3J = BYTW(&(W[TWVL * 6]), VSUB(T3H, T3I));
Chris@42 821 ST(&(x[WS(vs, 4) + WS(rs, 6)]), T3J, ms, &(x[WS(vs, 4)]));
Chris@42 822 T2F = BYTW(&(W[TWVL * 6]), VSUB(T2D, T2E));
Chris@42 823 ST(&(x[WS(vs, 4) + WS(rs, 4)]), T2F, ms, &(x[WS(vs, 4)]));
Chris@42 824 }
Chris@42 825 T28 = BYTW(&(W[TWVL * 6]), VSUB(T26, T27));
Chris@42 826 ST(&(x[WS(vs, 4) + WS(rs, 3)]), T28, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 827 T14 = BYTW(&(W[TWVL * 6]), VSUB(T12, T13));
Chris@42 828 ST(&(x[WS(vs, 4) + WS(rs, 1)]), T14, ms, &(x[WS(vs, 4) + WS(rs, 1)]));
Chris@42 829 {
Chris@42 830 V Th, Ti, Tb, Tg;
Chris@42 831 Tb = VBYI(VSUB(T7, Ta));
Chris@42 832 Tg = VSUB(Te, Tf);
Chris@42 833 Th = BYTW(&(W[TWVL * 4]), VADD(Tb, Tg));
Chris@42 834 Ti = BYTW(&(W[TWVL * 8]), VSUB(Tg, Tb));
Chris@42 835 ST(&(x[WS(vs, 3)]), Th, ms, &(x[WS(vs, 3)]));
Chris@42 836 ST(&(x[WS(vs, 5)]), Ti, ms, &(x[WS(vs, 5)]));
Chris@42 837 }
Chris@42 838 {
Chris@42 839 V T40, T41, T3U, T3Z;
Chris@42 840 T3U = VBYI(VSUB(T3Q, T3T));
Chris@42 841 T3Z = VSUB(T3X, T3Y);
Chris@42 842 T40 = BYTW(&(W[TWVL * 4]), VADD(T3U, T3Z));
Chris@42 843 T41 = BYTW(&(W[TWVL * 8]), VSUB(T3Z, T3U));
Chris@42 844 ST(&(x[WS(vs, 3) + WS(rs, 7)]), T40, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 845 ST(&(x[WS(vs, 5) + WS(rs, 7)]), T41, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 846 }
Chris@42 847 {
Chris@42 848 V T2p, T2q, T2j, T2o;
Chris@42 849 T2j = VBYI(VSUB(T2f, T2i));
Chris@42 850 T2o = VSUB(T2m, T2n);
Chris@42 851 T2p = BYTW(&(W[TWVL * 4]), VADD(T2j, T2o));
Chris@42 852 T2q = BYTW(&(W[TWVL * 8]), VSUB(T2o, T2j));
Chris@42 853 ST(&(x[WS(vs, 3) + WS(rs, 4)]), T2p, ms, &(x[WS(vs, 3)]));
Chris@42 854 ST(&(x[WS(vs, 5) + WS(rs, 4)]), T2q, ms, &(x[WS(vs, 5)]));
Chris@42 855 }
Chris@42 856 {
Chris@42 857 V T1S, T1T, T1M, T1R;
Chris@42 858 T1M = VBYI(VSUB(T1I, T1L));
Chris@42 859 T1R = VSUB(T1P, T1Q);
Chris@42 860 T1S = BYTW(&(W[TWVL * 4]), VADD(T1M, T1R));
Chris@42 861 T1T = BYTW(&(W[TWVL * 8]), VSUB(T1R, T1M));
Chris@42 862 ST(&(x[WS(vs, 3) + WS(rs, 3)]), T1S, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 863 ST(&(x[WS(vs, 5) + WS(rs, 3)]), T1T, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 864 }
Chris@42 865 {
Chris@42 866 V TO, TP, TI, TN;
Chris@42 867 TI = VBYI(VSUB(TE, TH));
Chris@42 868 TN = VSUB(TL, TM);
Chris@42 869 TO = BYTW(&(W[TWVL * 4]), VADD(TI, TN));
Chris@42 870 TP = BYTW(&(W[TWVL * 8]), VSUB(TN, TI));
Chris@42 871 ST(&(x[WS(vs, 3) + WS(rs, 1)]), TO, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 872 ST(&(x[WS(vs, 5) + WS(rs, 1)]), TP, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 873 }
Chris@42 874 {
Chris@42 875 V T1l, T1m, T1f, T1k;
Chris@42 876 T1f = VBYI(VSUB(T1b, T1e));
Chris@42 877 T1k = VSUB(T1i, T1j);
Chris@42 878 T1l = BYTW(&(W[TWVL * 4]), VADD(T1f, T1k));
Chris@42 879 T1m = BYTW(&(W[TWVL * 8]), VSUB(T1k, T1f));
Chris@42 880 ST(&(x[WS(vs, 3) + WS(rs, 2)]), T1l, ms, &(x[WS(vs, 3)]));
Chris@42 881 ST(&(x[WS(vs, 5) + WS(rs, 2)]), T1m, ms, &(x[WS(vs, 5)]));
Chris@42 882 }
Chris@42 883 {
Chris@42 884 V T3t, T3u, T3n, T3s;
Chris@42 885 T3n = VBYI(VSUB(T3j, T3m));
Chris@42 886 T3s = VSUB(T3q, T3r);
Chris@42 887 T3t = BYTW(&(W[TWVL * 4]), VADD(T3n, T3s));
Chris@42 888 T3u = BYTW(&(W[TWVL * 8]), VSUB(T3s, T3n));
Chris@42 889 ST(&(x[WS(vs, 3) + WS(rs, 6)]), T3t, ms, &(x[WS(vs, 3)]));
Chris@42 890 ST(&(x[WS(vs, 5) + WS(rs, 6)]), T3u, ms, &(x[WS(vs, 5)]));
Chris@42 891 }
Chris@42 892 {
Chris@42 893 V T2W, T2X, T2Q, T2V;
Chris@42 894 T2Q = VBYI(VSUB(T2M, T2P));
Chris@42 895 T2V = VSUB(T2T, T2U);
Chris@42 896 T2W = BYTW(&(W[TWVL * 4]), VADD(T2Q, T2V));
Chris@42 897 T2X = BYTW(&(W[TWVL * 8]), VSUB(T2V, T2Q));
Chris@42 898 ST(&(x[WS(vs, 3) + WS(rs, 5)]), T2W, ms, &(x[WS(vs, 3) + WS(rs, 1)]));
Chris@42 899 ST(&(x[WS(vs, 5) + WS(rs, 5)]), T2X, ms, &(x[WS(vs, 5) + WS(rs, 1)]));
Chris@42 900 }
Chris@42 901 {
Chris@42 902 V T1p, T1q, T1n, T1o;
Chris@42 903 T1n = VBYI(VADD(T1e, T1b));
Chris@42 904 T1o = VADD(T1i, T1j);
Chris@42 905 T1p = BYTW(&(W[0]), VADD(T1n, T1o));
Chris@42 906 T1q = BYTW(&(W[TWVL * 12]), VSUB(T1o, T1n));
Chris@42 907 ST(&(x[WS(vs, 1) + WS(rs, 2)]), T1p, ms, &(x[WS(vs, 1)]));
Chris@42 908 ST(&(x[WS(vs, 7) + WS(rs, 2)]), T1q, ms, &(x[WS(vs, 7)]));
Chris@42 909 }
Chris@42 910 {
Chris@42 911 V Tl, Tm, Tj, Tk;
Chris@42 912 Tj = VBYI(VADD(Ta, T7));
Chris@42 913 Tk = VADD(Te, Tf);
Chris@42 914 Tl = BYTW(&(W[0]), VADD(Tj, Tk));
Chris@42 915 Tm = BYTW(&(W[TWVL * 12]), VSUB(Tk, Tj));
Chris@42 916 ST(&(x[WS(vs, 1)]), Tl, ms, &(x[WS(vs, 1)]));
Chris@42 917 ST(&(x[WS(vs, 7)]), Tm, ms, &(x[WS(vs, 7)]));
Chris@42 918 }
Chris@42 919 {
Chris@42 920 V T2t, T2u, T2r, T2s;
Chris@42 921 T2r = VBYI(VADD(T2i, T2f));
Chris@42 922 T2s = VADD(T2m, T2n);
Chris@42 923 T2t = BYTW(&(W[0]), VADD(T2r, T2s));
Chris@42 924 T2u = BYTW(&(W[TWVL * 12]), VSUB(T2s, T2r));
Chris@42 925 ST(&(x[WS(vs, 1) + WS(rs, 4)]), T2t, ms, &(x[WS(vs, 1)]));
Chris@42 926 ST(&(x[WS(vs, 7) + WS(rs, 4)]), T2u, ms, &(x[WS(vs, 7)]));
Chris@42 927 }
Chris@42 928 {
Chris@42 929 V T3x, T3y, T3v, T3w;
Chris@42 930 T3v = VBYI(VADD(T3m, T3j));
Chris@42 931 T3w = VADD(T3q, T3r);
Chris@42 932 T3x = BYTW(&(W[0]), VADD(T3v, T3w));
Chris@42 933 T3y = BYTW(&(W[TWVL * 12]), VSUB(T3w, T3v));
Chris@42 934 ST(&(x[WS(vs, 1) + WS(rs, 6)]), T3x, ms, &(x[WS(vs, 1)]));
Chris@42 935 ST(&(x[WS(vs, 7) + WS(rs, 6)]), T3y, ms, &(x[WS(vs, 7)]));
Chris@42 936 }
Chris@42 937 {
Chris@42 938 V TS, TT, TQ, TR;
Chris@42 939 TQ = VBYI(VADD(TH, TE));
Chris@42 940 TR = VADD(TL, TM);
Chris@42 941 TS = BYTW(&(W[0]), VADD(TQ, TR));
Chris@42 942 TT = BYTW(&(W[TWVL * 12]), VSUB(TR, TQ));
Chris@42 943 ST(&(x[WS(vs, 1) + WS(rs, 1)]), TS, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 944 ST(&(x[WS(vs, 7) + WS(rs, 1)]), TT, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 945 }
Chris@42 946 {
Chris@42 947 V T1W, T1X, T1U, T1V;
Chris@42 948 T1U = VBYI(VADD(T1L, T1I));
Chris@42 949 T1V = VADD(T1P, T1Q);
Chris@42 950 T1W = BYTW(&(W[0]), VADD(T1U, T1V));
Chris@42 951 T1X = BYTW(&(W[TWVL * 12]), VSUB(T1V, T1U));
Chris@42 952 ST(&(x[WS(vs, 1) + WS(rs, 3)]), T1W, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 953 ST(&(x[WS(vs, 7) + WS(rs, 3)]), T1X, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 954 }
Chris@42 955 {
Chris@42 956 V T30, T31, T2Y, T2Z;
Chris@42 957 T2Y = VBYI(VADD(T2P, T2M));
Chris@42 958 T2Z = VADD(T2T, T2U);
Chris@42 959 T30 = BYTW(&(W[0]), VADD(T2Y, T2Z));
Chris@42 960 T31 = BYTW(&(W[TWVL * 12]), VSUB(T2Z, T2Y));
Chris@42 961 ST(&(x[WS(vs, 1) + WS(rs, 5)]), T30, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 962 ST(&(x[WS(vs, 7) + WS(rs, 5)]), T31, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 963 }
Chris@42 964 {
Chris@42 965 V T44, T45, T42, T43;
Chris@42 966 T42 = VBYI(VADD(T3T, T3Q));
Chris@42 967 T43 = VADD(T3X, T3Y);
Chris@42 968 T44 = BYTW(&(W[0]), VADD(T42, T43));
Chris@42 969 T45 = BYTW(&(W[TWVL * 12]), VSUB(T43, T42));
Chris@42 970 ST(&(x[WS(vs, 1) + WS(rs, 7)]), T44, ms, &(x[WS(vs, 1) + WS(rs, 1)]));
Chris@42 971 ST(&(x[WS(vs, 7) + WS(rs, 7)]), T45, ms, &(x[WS(vs, 7) + WS(rs, 1)]));
Chris@42 972 }
Chris@42 973 }
Chris@42 974 }
Chris@42 975 VLEAVE();
Chris@42 976 }
Chris@42 977
Chris@42 978 static const tw_instr twinstr[] = {
Chris@42 979 VTW(0, 1),
Chris@42 980 VTW(0, 2),
Chris@42 981 VTW(0, 3),
Chris@42 982 VTW(0, 4),
Chris@42 983 VTW(0, 5),
Chris@42 984 VTW(0, 6),
Chris@42 985 VTW(0, 7),
Chris@42 986 {TW_NEXT, VL, 0}
Chris@42 987 };
Chris@42 988
Chris@42 989 static const ct_desc desc = { 8, XSIMD_STRING("q1bv_8"), twinstr, &GENUS, {264, 128, 0, 0}, 0, 0, 0 };
Chris@42 990
Chris@42 991 void XSIMD(codelet_q1bv_8) (planner *p) {
Chris@42 992 X(kdft_difsq_register) (p, q1bv_8, &desc);
Chris@42 993 }
Chris@42 994 #endif /* HAVE_FMA */