~siretart/ubuntu/utopic/blender/libav10

/* Generated by: ../../../genfft/gen_hc2hc -fma -reorder-insns -schedule-for-pipeline -compact -variables 4 -pipeline-latency 4 -sign 1 -n 12 -dif -name hb_12 -include hb.h */

* This function contains 118 FP additions, 68 FP multiplications,

* (or, 72 additions, 22 multiplications, 46 fused multiply/add),

* 66 stack variables, and 48 memory accesses

* Generator Id's :

* $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $

* $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $

* $Id: gen_hc2hc.ml,v 1.16 2006-02-12 23:34:12 athena Exp $

#include "hb.h"

static const R *hb_12(R *rio, R *iio, const R *W, stride ios, INT m, INT dist)

{

DK(KP866025403, +0.866025403784438646763723170752936183471402627);

DK(KP500000000, +0.500000000000000000000000000000000000000000000);

INT i;

for (i = m - 2; i > 0; i = i - 2, rio = rio + dist, iio = iio - dist, W = W + 22, MAKE_VOLATILE_STRIDE(ios)) {

E T18, T20, T21, T1b, T2a, T1s, T29, T1p, Tz, T11, TD, Tb, Tg, T23, T1f;

E Tl, TN, TI, T1i, T24, T1z, T2d, T1w, T2c;

{

E T5, Ta, Ty, Tt;

{

E T1, Tp, T6, Tu, T7, T1o, T4, T17, Ts, T8, Tv, Tw;

T1 = rio[0];

Tp = iio[0];

T6 = iio[-WS(ios, 6)];

Tu = rio[WS(ios, 6)];

{

E T2, T3, Tq, Tr;

T2 = rio[WS(ios, 4)];

T3 = iio[-WS(ios, 8)];

Tq = iio[-WS(ios, 4)];

Tr = rio[WS(ios, 8)];

T7 = iio[-WS(ios, 10)];

T1o = T2 - T3;

T4 = T2 + T3;

T17 = Tr + Tq;

Ts = Tq - Tr;

T8 = rio[WS(ios, 2)];

Tv = rio[WS(ios, 10)];

Tw = iio[-WS(ios, 2)];

}

{

E T1r, T1a, T19, T1q, T9, Tx, T16, T1n;

T5 = T1 + T4;

T16 = FNMS(KP500000000, T4, T1);

T1r = T7 - T8;

T9 = T7 + T8;

T1a = Tv + Tw;

Tx = Tv - Tw;

T18 = FMA(KP866025403, T17, T16);

T20 = FNMS(KP866025403, T17, T16);

T19 = FNMS(KP500000000, T9, T6);

Ta = T6 + T9;

Ty = Tu + Tx;

T1q = FNMS(KP500000000, Tx, Tu);

T1n = FNMS(KP500000000, Ts, Tp);

Tt = Tp + Ts;

T21 = FMA(KP866025403, T1a, T19);

T1b = FNMS(KP866025403, T1a, T19);

T2a = FNMS(KP866025403, T1r, T1q);

T1s = FMA(KP866025403, T1r, T1q);

T29 = FMA(KP866025403, T1o, T1n);

T1p = FNMS(KP866025403, T1o, T1n);

}

{

100

E Tc, TE, Th, TM, Ti, Tf, T1v, TH, T1e, Tj, TJ, TK;

101

Tc = rio[WS(ios, 3)];

102

Tz = Tt + Ty;

103

T11 = Tt - Ty;

104

TE = iio[-WS(ios, 3)];

105

TD = T5 - Ta;

106

Tb = T5 + Ta;

107

Th = iio[-WS(ios, 9)];

108

TM = rio[WS(ios, 9)];

109

{

110

E Td, Te, TF, TG;

111

Td = iio[-WS(ios, 7)];

112

Te = iio[-WS(ios, 11)];

113

TF = rio[WS(ios, 7)];

114

TG = rio[WS(ios, 11)];

115

Ti = rio[WS(ios, 1)];

116

Tf = Td + Te;

117

T1v = Td - Te;

118

TH = TF + TG;

119

T1e = TF - TG;

120

Tj = rio[WS(ios, 5)];

121

TJ = iio[-WS(ios, 5)];

122

TK = iio[-WS(ios, 1)];

123

}

124

{

125

E T1y, T1h, T1g, T1x, Tk, TL, T1d, T1u;

126

T1d = FNMS(KP500000000, Tf, Tc);

127

Tg = Tc + Tf;

128

Tk = Ti + Tj;

129

T1y = Ti - Tj;

130

TL = TJ + TK;

131

T1h = TJ - TK;

132

T23 = FMA(KP866025403, T1e, T1d);

133

T1f = FNMS(KP866025403, T1e, T1d);

134

Tl = Th + Tk;

135

T1g = FNMS(KP500000000, Tk, Th);

136

T1x = FMA(KP500000000, TL, TM);

137

TN = TL - TM;

138

TI = TE - TH;

139

T1u = FMA(KP500000000, TH, TE);

140

T1i = FNMS(KP866025403, T1h, T1g);

141

T24 = FMA(KP866025403, T1h, T1g);

142

T1z = FMA(KP866025403, T1y, T1x);

143

T2d = FNMS(KP866025403, T1y, T1x);

144

T1w = FNMS(KP866025403, T1v, T1u);

145

T2c = FMA(KP866025403, T1v, T1u);

146

}

147

}

148

}

149

{

150

E TY, T13, TX, T10;

151

{

152

E Tn, T12, TC, Tm, To, TS, TP, TO;

153

Tn = W[16];

154

T12 = TI + TN;

155

TO = TI - TN;

156

TC = W[17];

157

Tm = Tg + Tl;

158

To = Tg - Tl;

159

TS = TD + TO;

160

TP = TD - TO;

161

{

162

E TV, TU, TW, TT;

163

{

164

E TB, TR, TA, TQ;

165

TV = Tz - To;

166

TA = To + Tz;

167

rio[0] = Tb + Tm;

168

TQ = Tn * TP;

169

TB = Tn * TA;

170

TR = W[4];

171

rio[WS(ios, 9)] = FNMS(TC, TA, TQ);

172

TU = W[5];

173

iio[-WS(ios, 2)] = FMA(TC, TP, TB);

174

TW = TR * TV;

175

TT = TR * TS;

176

}

177

iio[-WS(ios, 8)] = FMA(TU, TS, TW);

178

rio[WS(ios, 3)] = FNMS(TU, TV, TT);

179

TY = Tb - Tm;

180

T13 = T11 - T12;

181

TX = W[10];

182

T10 = W[11];

183

iio[-WS(ios, 11)] = T11 + T12;

184

}

185

}

186

{

187

E T1c, T1A, T1t, T1j, T22, T2e, T2b, T2B, T2q, T25, T2s, T2y, T2C, T2z, T2w;

188

E T2A;

189

{

190

E T1X, T1M, T1O, T1U, T1Y, T1V, T1S, T1W, T1P, T1Q;

191

{

192

E T1K, TZ, T14, T1L;

193

T1c = T18 + T1b;

194

T1K = T18 - T1b;

195

TZ = TX * TY;

196

T14 = T10 * TY;

197

T1L = T1w + T1z;

198

T1A = T1w - T1z;

199

T1t = T1p - T1s;

200

T1P = T1p + T1s;

201

rio[WS(ios, 6)] = FNMS(T10, T13, TZ);

202

iio[-WS(ios, 5)] = FMA(TX, T13, T14);

203

T1X = T1K + T1L;

204

T1M = T1K - T1L;

205

T1Q = T1f - T1i;

206

T1j = T1f + T1i;

207

}

208

{

209

E T1J, T1T, T1R, T1N;

210

T1J = W[8];

211

T1O = W[9];

212

T1T = W[20];

213

T1U = T1P - T1Q;

214

T1R = T1P + T1Q;

215

T1N = T1J * T1M;

216

T1Y = T1T * T1X;

217

T1V = T1T * T1U;

218

T1S = T1J * T1R;

219

rio[WS(ios, 5)] = FNMS(T1O, T1R, T1N);

220

T1W = W[21];

221

}

222

{

223

E T2t, T2u, T2o, T2p;

224

T2o = T20 - T21;

225

T22 = T20 + T21;

226

iio[-WS(ios, 6)] = FMA(T1O, T1M, T1S);

227

T2p = T2c + T2d;

228

T2e = T2c - T2d;

229

rio[WS(ios, 11)] = FNMS(T1W, T1U, T1Y);

230

iio[0] = FMA(T1W, T1X, T1V);

231

T2b = T29 - T2a;

232

T2t = T29 + T2a;

233

T2B = T2o + T2p;

234

T2q = T2o - T2p;

235

T2u = T23 - T24;

236

T25 = T23 + T24;

237

{

238

E T2n, T2x, T2v, T2r;

239

T2n = W[0];

240

T2s = W[1];

241

T2x = W[12];

242

T2y = T2t - T2u;

243

T2v = T2t + T2u;

244

T2r = T2n * T2q;

245

T2C = T2x * T2B;

246

T2z = T2x * T2y;

247

T2w = T2n * T2v;

248

rio[WS(ios, 1)] = FNMS(T2s, T2v, T2r);

249

T2A = W[13];

250

}

251

}

252

}

253

{

254

E T2i, T2h, T2l, T2j, T2k, T26;

255

iio[-WS(ios, 10)] = FMA(T2s, T2q, T2w);

256

rio[WS(ios, 7)] = FNMS(T2A, T2y, T2C);

257

iio[-WS(ios, 4)] = FMA(T2A, T2B, T2z);

258

T2i = T22 + T25;

259

T26 = T22 - T25;

260

{

261

E T1Z, T28, T2f, T27, T2g;

262

T1Z = W[18];

263

T28 = W[19];

264

T2h = W[6];

265

T2l = T2b + T2e;

266

T2f = T2b - T2e;

267

T27 = T1Z * T26;

268

T2g = T28 * T26;

269

T2j = T2h * T2i;

270

T2k = W[7];

271

rio[WS(ios, 10)] = FNMS(T28, T2f, T27);

272

iio[-WS(ios, 1)] = FMA(T1Z, T2f, T2g);

273

}

274

{

275

E T1k, T1E, T1H, T1B, T2m, T15, T1m;

276

rio[WS(ios, 4)] = FNMS(T2k, T2l, T2j);

277

T2m = T2k * T2i;

278

iio[-WS(ios, 7)] = FMA(T2h, T2l, T2m);

279

T1k = T1c - T1j;

280

T1E = T1c + T1j;

281

T1H = T1t + T1A;

282

T1B = T1t - T1A;

283

T15 = W[2];

284

T1m = W[3];

285

{

286

E T1D, T1G, T1l, T1C, T1F, T1I;

287

T1D = W[14];

288

T1G = W[15];

289

T1l = T15 * T1k;

290

T1C = T1m * T1k;

291

T1F = T1D * T1E;

292

T1I = T1G * T1E;

293

rio[WS(ios, 2)] = FNMS(T1m, T1B, T1l);

294

iio[-WS(ios, 9)] = FMA(T15, T1B, T1C);

295

rio[WS(ios, 8)] = FNMS(T1G, T1H, T1F);

296

iio[-WS(ios, 3)] = FMA(T1D, T1H, T1I);

297

}

298

}

299

}

300

}

301

}

302

}

303

return W;

304

}

305

306

static const tw_instr twinstr[] = {

307

{TW_FULL, 0, 12},

308

{TW_NEXT, 1, 0}

309

};

310

311

static const hc2hc_desc desc = { 12, "hb_12", twinstr, &GENUS, {72, 22, 46, 0}, 0, 0, 0 };

312

313

void X(codelet_hb_12) (planner *p) {

314

X(khc2hc_register) (p, hb_12, &desc);

315

}

316

#else /* HAVE_FMA */

317

318

/* Generated by: ../../../genfft/gen_hc2hc -compact -variables 4 -pipeline-latency 4 -sign 1 -n 12 -dif -name hb_12 -include hb.h */

319

320

321

* This function contains 118 FP additions, 60 FP multiplications,

322

* (or, 88 additions, 30 multiplications, 30 fused multiply/add),

323

* 39 stack variables, and 48 memory accesses

324

325

326

* Generator Id's :

327

* $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $

328

* $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $

329

* $Id: gen_hc2hc.ml,v 1.16 2006-02-12 23:34:12 athena Exp $

330

331

332

#include "hb.h"

333

334

static const R *hb_12(R *rio, R *iio, const R *W, stride ios, INT m, INT dist)

335

{

336

DK(KP500000000, +0.500000000000000000000000000000000000000000000);

337

DK(KP866025403, +0.866025403784438646763723170752936183471402627);

338

INT i;

339

for (i = m - 2; i > 0; i = i - 2, rio = rio + dist, iio = iio - dist, W = W + 22, MAKE_VOLATILE_STRIDE(ios)) {

340

E T5, Tt, T12, T1M, T1i, T1U, Tl, TM, T1c, T1Y, T1s, T1Q, Ta, Ty, T15;

341

E T1N, T1l, T1V, Tg, TH, T19, T1X, T1p, T1P;

342

{

343

E T1, Tp, T4, T1g, Ts, T11, T10, T1h;

344

T1 = rio[0];

345

Tp = iio[0];

346

{

347

E T2, T3, Tq, Tr;

348

T2 = rio[WS(ios, 4)];

349

T3 = iio[-WS(ios, 8)];

350

T4 = T2 + T3;

351

T1g = KP866025403 * (T2 - T3);

352

Tq = rio[WS(ios, 8)];

353

Tr = iio[-WS(ios, 4)];

354

Ts = Tq - Tr;

355

T11 = KP866025403 * (Tq + Tr);

356

}

357

T5 = T1 + T4;

358

Tt = Tp - Ts;

359

T10 = FNMS(KP500000000, T4, T1);

360

T12 = T10 - T11;

361

T1M = T10 + T11;

362

T1h = FMA(KP500000000, Ts, Tp);

363

T1i = T1g + T1h;

364

T1U = T1h - T1g;

365

}

366

{

367

E Th, TL, Tk, T1a, TK, T1r, T1b, T1q;

368

Th = iio[-WS(ios, 9)];

369

TL = rio[WS(ios, 9)];

370

{

371

E Ti, Tj, TI, TJ;

372

Ti = rio[WS(ios, 1)];

373

Tj = rio[WS(ios, 5)];

374

Tk = Ti + Tj;

375

T1a = KP866025403 * (Ti - Tj);

376

TI = iio[-WS(ios, 5)];

377

TJ = iio[-WS(ios, 1)];

378

TK = TI + TJ;

379

T1r = KP866025403 * (TI - TJ);

380

}

381

Tl = Th + Tk;

382

TM = TK - TL;

383

T1b = FMA(KP500000000, TK, TL);

384

T1c = T1a - T1b;

385

T1Y = T1a + T1b;

386

T1q = FNMS(KP500000000, Tk, Th);

387

T1s = T1q + T1r;

388

T1Q = T1q - T1r;

389

}

390

{

391

E T6, Tx, T9, T1j, Tw, T14, T13, T1k;

392

T6 = iio[-WS(ios, 6)];

393

Tx = rio[WS(ios, 6)];

394

{

395

E T7, T8, Tu, Tv;

396

T7 = iio[-WS(ios, 10)];

397

T8 = rio[WS(ios, 2)];

398

T9 = T7 + T8;

399

T1j = KP866025403 * (T7 - T8);

400

Tu = rio[WS(ios, 10)];

401

Tv = iio[-WS(ios, 2)];

402

Tw = Tu - Tv;

403

T14 = KP866025403 * (Tu + Tv);

404

}

405

Ta = T6 + T9;

406

Ty = Tw + Tx;

407

T13 = FNMS(KP500000000, T9, T6);

408

T15 = T13 + T14;

409

T1N = T13 - T14;

410

T1k = FMS(KP500000000, Tw, Tx);

411

T1l = T1j + T1k;

412

T1V = T1k - T1j;

413

}

414

{

415

E Tc, TD, Tf, T17, TG, T1o, T18, T1n;

416

Tc = rio[WS(ios, 3)];

417

TD = iio[-WS(ios, 3)];

418

{

419

E Td, Te, TE, TF;

420

Td = iio[-WS(ios, 7)];

421

Te = iio[-WS(ios, 11)];

422

Tf = Td + Te;

423

T17 = KP866025403 * (Td - Te);

424

TE = rio[WS(ios, 7)];

425

TF = rio[WS(ios, 11)];

426

TG = TE + TF;

427

T1o = KP866025403 * (TE - TF);

428

}

429

Tg = Tc + Tf;

430

TH = TD - TG;

431

T18 = FMA(KP500000000, TG, TD);

432

T19 = T17 + T18;

433

T1X = T18 - T17;

434

T1n = FNMS(KP500000000, Tf, Tc);

435

T1p = T1n + T1o;

436

T1P = T1n - T1o;

437

}

438

{

439

E Tb, Tm, TU, TW, TX, TY, TT, TV;

440

Tb = T5 + Ta;

441

Tm = Tg + Tl;

442

TU = Tb - Tm;

443

TW = Tt - Ty;

444

TX = TH + TM;

445

TY = TW - TX;

446

rio[0] = Tb + Tm;

447

iio[-WS(ios, 11)] = TW + TX;

448

TT = W[10];

449

TV = W[11];

450

rio[WS(ios, 6)] = FNMS(TV, TY, TT * TU);

451

iio[-WS(ios, 5)] = FMA(TV, TU, TT * TY);

452

}

453

{

454

E T28, T2g, T2c, T2e;

455

{

456

E T26, T27, T2a, T2b;

457

T26 = T1M - T1N;

458

T27 = T1X + T1Y;

459

T28 = T26 - T27;

460

T2g = T26 + T27;

461

T2a = T1U - T1V;

462

T2b = T1P - T1Q;

463

T2c = T2a + T2b;

464

T2e = T2a - T2b;

465

}

466

{

467

E T25, T29, T2d, T2f;

468

T25 = W[8];

469

T29 = W[9];

470

rio[WS(ios, 5)] = FNMS(T29, T2c, T25 * T28);

471

iio[-WS(ios, 6)] = FMA(T25, T2c, T29 * T28);

472

T2d = W[20];

473

T2f = W[21];

474

iio[0] = FMA(T2d, T2e, T2f * T2g);

475

rio[WS(ios, 11)] = FNMS(T2f, T2e, T2d * T2g);

476

}

477

}

478

{

479

E TA, TS, TO, TQ;

480

{

481

E To, Tz, TC, TN;

482

To = Tg - Tl;

483

Tz = Tt + Ty;

484

TA = To + Tz;

485

TS = Tz - To;

486

TC = T5 - Ta;

487

TN = TH - TM;

488

TO = TC - TN;

489

TQ = TC + TN;

490

}

491

{

492

E Tn, TB, TP, TR;

493

Tn = W[16];

494

TB = W[17];

495

iio[-WS(ios, 2)] = FMA(Tn, TA, TB * TO);

496

rio[WS(ios, 9)] = FNMS(TB, TA, Tn * TO);

497

TP = W[4];

498

TR = W[5];

499

rio[WS(ios, 3)] = FNMS(TR, TS, TP * TQ);

500

iio[-WS(ios, 8)] = FMA(TP, TS, TR * TQ);

501

}

502

}

503

{

504

E T1S, T22, T20, T24;

505

{

506

E T1O, T1R, T1W, T1Z;

507

T1O = T1M + T1N;

508

T1R = T1P + T1Q;

509

T1S = T1O - T1R;

510

T22 = T1O + T1R;

511

T1W = T1U + T1V;

512

T1Z = T1X - T1Y;

513

T20 = T1W - T1Z;

514

T24 = T1W + T1Z;

515

}

516

{

517

E T1L, T1T, T21, T23;

518

T1L = W[2];

519

T1T = W[3];

520

rio[WS(ios, 2)] = FNMS(T1T, T20, T1L * T1S);

521

iio[-WS(ios, 9)] = FMA(T1T, T1S, T1L * T20);

522

T21 = W[14];

523

T23 = W[15];

524

rio[WS(ios, 8)] = FNMS(T23, T24, T21 * T22);

525

iio[-WS(ios, 3)] = FMA(T23, T22, T21 * T24);

526

}

527

}

528

{

529

E T1C, T1I, T1G, T1K;

530

{

531

E T1A, T1B, T1E, T1F;

532

T1A = T12 + T15;

533

T1B = T1p + T1s;

534

T1C = T1A - T1B;

535

T1I = T1A + T1B;

536

T1E = T1i + T1l;

537

T1F = T19 + T1c;

538

T1G = T1E - T1F;

539

T1K = T1E + T1F;

540

}

541

{

542

E T1z, T1D, T1H, T1J;

543

T1z = W[18];

544

T1D = W[19];

545

rio[WS(ios, 10)] = FNMS(T1D, T1G, T1z * T1C);

546

iio[-WS(ios, 1)] = FMA(T1D, T1C, T1z * T1G);

547

T1H = W[6];

548

T1J = W[7];

549

rio[WS(ios, 4)] = FNMS(T1J, T1K, T1H * T1I);

550

iio[-WS(ios, 7)] = FMA(T1J, T1I, T1H * T1K);

551

}

552

}

553

{

554

E T1e, T1y, T1u, T1w;

555

{

556

E T16, T1d, T1m, T1t;

557

T16 = T12 - T15;

558

T1d = T19 - T1c;

559

T1e = T16 - T1d;

560

T1y = T16 + T1d;

561

T1m = T1i - T1l;

562

T1t = T1p - T1s;

563

T1u = T1m + T1t;

564

T1w = T1m - T1t;

565

}

566

{

567

E TZ, T1f, T1v, T1x;

568

TZ = W[0];

569

T1f = W[1];

570

rio[WS(ios, 1)] = FNMS(T1f, T1u, TZ * T1e);

571

iio[-WS(ios, 10)] = FMA(TZ, T1u, T1f * T1e);

572

T1v = W[12];

573

T1x = W[13];

574

iio[-WS(ios, 4)] = FMA(T1v, T1w, T1x * T1y);

575

rio[WS(ios, 7)] = FNMS(T1x, T1w, T1v * T1y);

576

}

577

}

578

}

579