dlsch_llr_computation.c 441 KB
Newer Older
1
/*******************************************************************************
ghaddab's avatar
ghaddab committed
2 3
    OpenAirInterface 
    Copyright(c) 1999 - 2014 Eurecom
4

ghaddab's avatar
ghaddab committed
5 6 7 8
    OpenAirInterface is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.
9 10


ghaddab's avatar
ghaddab committed
11 12 13 14
    OpenAirInterface is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.
15

ghaddab's avatar
ghaddab committed
16 17 18 19
    You should have received a copy of the GNU General Public License
    along with OpenAirInterface.The full GNU General Public License is 
   included in this distribution in the file called "COPYING". If not, 
   see <http://www.gnu.org/licenses/>.
20 21

  Contact Information
ghaddab's avatar
ghaddab committed
22 23 24 25
  OpenAirInterface Admin: openair_admin@eurecom.fr
  OpenAirInterface Tech : openair_tech@eurecom.fr
  OpenAirInterface Dev  : openair4g-devel@eurecom.fr
  
ghaddab's avatar
ghaddab committed
26
  Address      : Eurecom, Campus SophiaTech, 450 Route des Chappes, CS 50193 - 06904 Biot Sophia Antipolis cedex, FRANCE
27

ghaddab's avatar
ghaddab committed
28
 *******************************************************************************/
29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44

/*! \file PHY/LTE_TRANSPORT/dlsch_llr_computation.c
* \brief Top-level routines for LLR computation of the PDSCH physical channel from 36-211, V8.6 2009-03
* \author R. Knopp, F. Kaltenberger,A. Bhamri, S. Aubert, S. Wagner
* \date 2011
* \version 0.1
* \company Eurecom
* \email: knopp@eurecom.fr,florian.kaltenberger@eurecom.fr,ankit.bhamri@eurecom.fr,sebastien.aubert@eurecom.fr, sebastian.wagner@eurecom.fr
* \note
* \warning
*/

#include "PHY/defs.h"
#include "PHY/extern.h"
#include "defs.h"
#include "extern.h"
45
#include "PHY/sse_intrin.h"
46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639

#ifndef USER_MODE
#define NOCYGWIN_STATIC static
#else
#define NOCYGWIN_STATIC 
#endif

NOCYGWIN_STATIC short zero[8] __attribute__ ((aligned(16))) = {0,0,0,0,0,0,0,0};
NOCYGWIN_STATIC short ones[8] __attribute__ ((aligned(16))) = {0xffff,0xffff,0xffff,0xffff,0xffff,0xffff,0xffff,0xffff};
NOCYGWIN_STATIC __m128i rho_rpi __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_1_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_1_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_1_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_1_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_3_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_3_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_3_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_3_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_5_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_5_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_5_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_5_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_7_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_7_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_7_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rpi_7_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_1_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_1_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_1_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_1_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_3_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_3_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_3_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_3_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_5_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_5_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_5_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_5_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_7_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_7_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_7_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i rho_rmi_7_7 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i psi_r_m7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m7_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_m1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_r_p7_p7 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i psi_i_m7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m7_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_m1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_i_p7_p7 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i a_r_m7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m7_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_m1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_r_p7_p7 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i a_i_m7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m7_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_m1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_i_p7_p7 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i psi_a_m7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m7_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_m1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i psi_a_p7_p7 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i a_sq_m7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m7_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_m1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i a_sq_p7_p7 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i bit_met_m7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m7_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_m1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p1_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p1_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p1_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p1_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p1_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p1_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p1_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p1_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p3_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p3_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p3_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p3_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p3_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p3_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p3_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p3_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p5_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p5_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p5_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p5_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p5_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p5_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p5_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p5_p7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p7_m7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p7_m5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p7_m3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p7_m1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p7_p1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p7_p3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p7_p5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i bit_met_p7_p7 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i  y0_p_1_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_1_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_1_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_1_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_3_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_3_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_3_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_3_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_5_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_5_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_5_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_5_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_7_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_7_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_7_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_p_7_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_1_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_1_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_1_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_1_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_3_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_3_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_3_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_3_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_5_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_5_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_5_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_5_7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_7_1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_7_3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_7_5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0_m_7_7 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i  xmm0 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  xmm1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  xmm2 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  xmm3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  xmm4 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  xmm5 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  xmm6 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  xmm7 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  xmm8 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i  y0r __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0i __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y1r __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y1i __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y2r __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y2i __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i  logmax_num_re0 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  logmax_num_im0 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  logmax_den_re0 __attribute__ ((aligned(16))); 
NOCYGWIN_STATIC __m128i  logmax_den_im0 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  logmax_num_re1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  logmax_num_im1 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  logmax_den_re1 __attribute__ ((aligned(16))); 
NOCYGWIN_STATIC __m128i  logmax_den_im1 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i tmp_result  __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i tmp_result2 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i tmp_result3 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i tmp_result4 __attribute__ ((aligned(16)));

//==============================================================================================
// Auxiliary Makros

// calculates psi_a = psi_r*a_r + psi_i*a_i 
#define prodsum_psi_a_epi16(psi_r,a_r,psi_i,a_i,psi_a) tmp_result = _mm_mulhi_epi16(psi_r,a_r); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result2 = _mm_mulhi_epi16(psi_i,a_i); tmp_result2 = _mm_slli_epi16(tmp_result2,1); psi_a = _mm_adds_epi16(tmp_result,tmp_result2);

// calculate interference magnitude
#define interference_abs_epi16(psi,int_ch_mag,int_mag,c1,c2) tmp_result = _mm_cmplt_epi16(psi,int_ch_mag); tmp_result2 = _mm_xor_si128(tmp_result,(*(__m128i*)&ones[0])); tmp_result = _mm_and_si128(tmp_result,c1); tmp_result2 = _mm_and_si128(tmp_result2,c2); int_mag = _mm_or_si128(tmp_result,tmp_result2); 

// calculate interference magnitude
// tmp_result = ones in shorts corr. to interval 2<=x<=4, tmp_result2 interval < 2, tmp_result3 interval 4<x<6 and tmp_result4 interval x>6 
#define interference_abs_64qam_epi16(psi,int_ch_mag,int_two_ch_mag,int_three_ch_mag,a,c1,c3,c5,c7) tmp_result = _mm_cmplt_epi16(psi,int_two_ch_mag); tmp_result3 = _mm_xor_si128(tmp_result,(*(__m128i*)&ones[0])); tmp_result2 = _mm_cmplt_epi16(psi,int_ch_mag); tmp_result = _mm_xor_si128(tmp_result,tmp_result2); tmp_result4 = _mm_cmpgt_epi16(psi,int_three_ch_mag); tmp_result3 = _mm_xor_si128(tmp_result3,tmp_result4); tmp_result = _mm_and_si128(tmp_result,c3); tmp_result2 = _mm_and_si128(tmp_result2,c1); tmp_result3 = _mm_and_si128(tmp_result3,c5); tmp_result4 = _mm_and_si128(tmp_result4,c7); tmp_result = _mm_or_si128(tmp_result,tmp_result2); tmp_result3 = _mm_or_si128(tmp_result3,tmp_result4); a = _mm_or_si128(tmp_result,tmp_result3); 

// calculates a_sq = int_ch_mag*(a_r^2 + a_i^2)*scale_factor 
#define square_a_epi16(a_r,a_i,int_ch_mag,scale_factor,a_sq) tmp_result = _mm_mulhi_epi16(a_r,a_r); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result = _mm_mulhi_epi16(tmp_result,scale_factor); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result = _mm_mulhi_epi16(tmp_result,int_ch_mag); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result2 = _mm_mulhi_epi16(a_i,a_i); tmp_result2 = _mm_slli_epi16(tmp_result2,1); tmp_result2 = _mm_mulhi_epi16(tmp_result2,scale_factor); tmp_result2 = _mm_slli_epi16(tmp_result2,1); tmp_result2 = _mm_mulhi_epi16(tmp_result2,int_ch_mag); tmp_result2 = _mm_slli_epi16(tmp_result2,1); a_sq = _mm_adds_epi16(tmp_result,tmp_result2);

// calculates a_sq = int_ch_mag*(a_r^2 + a_i^2)*scale_factor for 64-QAM
#define square_a_64qam_epi16(a_r,a_i,int_ch_mag,scale_factor,a_sq)  tmp_result = _mm_mulhi_epi16(a_r,a_r); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result = _mm_mulhi_epi16(tmp_result,scale_factor); tmp_result = _mm_slli_epi16(tmp_result,3); tmp_result = _mm_mulhi_epi16(tmp_result,int_ch_mag); tmp_result = _mm_slli_epi16(tmp_result,1); tmp_result2 = _mm_mulhi_epi16(a_i,a_i); tmp_result2 = _mm_slli_epi16(tmp_result2,1); tmp_result2 = _mm_mulhi_epi16(tmp_result2,scale_factor); tmp_result2 = _mm_slli_epi16(tmp_result2,3); tmp_result2 = _mm_mulhi_epi16(tmp_result2,int_ch_mag); tmp_result2 = _mm_slli_epi16(tmp_result2,1); a_sq = _mm_adds_epi16(tmp_result,tmp_result2);

//==============================================================================================
// SINGLE-STREAM
//==============================================================================================

//----------------------------------------------------------------------------------------------
// QPSK
//----------------------------------------------------------------------------------------------

int dlsch_qpsk_llr(LTE_DL_FRAME_PARMS *frame_parms,
                   int **rxdataF_comp,
                   short *dlsch_llr,
                   unsigned char symbol,
gauthier's avatar
gauthier committed
640 641 642
                   uint8_t first_symbol_flag,
                   uint16_t nb_rb,
                   uint16_t pbch_pss_sss_adjust,
643 644
                   short **llr32p) {

gauthier's avatar
gauthier committed
645 646
  uint32_t *rxF = (uint32_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)];
  uint32_t *llr32;
647
  int i,len;
gauthier's avatar
gauthier committed
648
  uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol;
649 650

  if (first_symbol_flag==1) {
gauthier's avatar
gauthier committed
651
      llr32 = (uint32_t*)dlsch_llr;
652 653
  }
  else {
gauthier's avatar
gauthier committed
654
      llr32 = (uint32_t*)(*llr32p);
655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695
  }
 
  if (!llr32) {
      msg("dlsch_qpsk_llr: llr is null, symbol %d, llr32=%p\n",symbol, llr32);
      return(-1);
  } 


  if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) {
      if (frame_parms->mode1_flag==0)
          len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3);
      else
          len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6);
  }
  else {
      len = (nb_rb*12) - pbch_pss_sss_adjust;
  }
  //  printf("dlsch_qpsk_llr: symbol %d,len %d,pbch_pss_sss_adjust %d\n",symbol,len,pbch_pss_sss_adjust);
  for (i=0;i<len;i++) {
      *llr32 = *rxF;
      rxF++;
      llr32++;
  }

  *llr32p = (short *)llr32;

  _mm_empty();
  _m_empty();

  return(0);
}

//----------------------------------------------------------------------------------------------
// 16-QAM
//----------------------------------------------------------------------------------------------

void dlsch_16qam_llr(LTE_DL_FRAME_PARMS *frame_parms,
                     int **rxdataF_comp,
                     short *dlsch_llr,
                     int **dl_ch_mag,
                     unsigned char symbol,
gauthier's avatar
gauthier committed
696
                     uint8_t first_symbol_flag,
697
                     unsigned short nb_rb,
gauthier's avatar
gauthier committed
698 699
                     uint16_t pbch_pss_sss_adjust,
                     int16_t **llr32p) {
700 701 702 703 704 705

    __m128i *rxF = (__m128i*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)];
    __m128i *ch_mag;
    __m128i llr128[2];
    int i,len;
    unsigned char symbol_mod,len_mod4=0;
gauthier's avatar
gauthier committed
706
    uint32_t *llr32;
knopp's avatar
 
knopp committed
707 708


709
    if (first_symbol_flag==1) {
gauthier's avatar
gauthier committed
710
        llr32 = (uint32_t*)dlsch_llr;
711 712
    }
    else {
gauthier's avatar
gauthier committed
713
        llr32 = (uint32_t*)*llr32p;
714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747
    }
  
    symbol_mod = (symbol>=(7-frame_parms->Ncp)) ? symbol-(7-frame_parms->Ncp) : symbol;
    
    ch_mag = (__m128i*)&dl_ch_mag[0][(symbol*frame_parms->N_RB_DL*12)];
    
    if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) {
        if (frame_parms->mode1_flag==0)
            len = nb_rb*8 - (2*pbch_pss_sss_adjust/3);
        else
            len = nb_rb*10 - (5*pbch_pss_sss_adjust/6);
    }
    else {
        len = nb_rb*12 - pbch_pss_sss_adjust;
    }
    
  // update output pointer according to number of REs in this symbol (<<2 because 4 bits per RE)
    if (first_symbol_flag == 1)
        *llr32p = dlsch_llr + (len<<2);
    else
        *llr32p += (len<<2);
    
    len_mod4 = len&3;
    len>>=2;  // length in quad words (4 REs)
    len+=(len_mod4==0 ? 0 : 1);
    
    for (i=0;i<len;i++) {
        
        xmm0 = _mm_abs_epi16(rxF[i]);
        xmm0 = _mm_subs_epi16(ch_mag[i],xmm0);

        // lambda_1=y_R, lambda_2=|y_R|-|h|^2, lamda_3=y_I, lambda_4=|y_I|-|h|^2
        llr128[0] = _mm_unpacklo_epi32(rxF[i],xmm0); 
        llr128[1] = _mm_unpackhi_epi32(rxF[i],xmm0);
gauthier's avatar
gauthier committed
748 749 750 751 752 753 754 755
        llr32[0] = _mm_extract_epi32(llr128[0],0); //((uint32_t *)&llr128[0])[0];
        llr32[1] = _mm_extract_epi32(llr128[0],1); //((uint32_t *)&llr128[0])[1];
        llr32[2] = _mm_extract_epi32(llr128[0],2); //((uint32_t *)&llr128[0])[2];
        llr32[3] = _mm_extract_epi32(llr128[0],3); //((uint32_t *)&llr128[0])[3];
        llr32[4] = _mm_extract_epi32(llr128[1],0); //((uint32_t *)&llr128[1])[0];
        llr32[5] = _mm_extract_epi32(llr128[1],1); //((uint32_t *)&llr128[1])[1];
        llr32[6] = _mm_extract_epi32(llr128[1],2); //((uint32_t *)&llr128[1])[2];
        llr32[7] = _mm_extract_epi32(llr128[1],3); //((uint32_t *)&llr128[1])[3];
756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771
        llr32+=8;
  }
  _mm_empty();
  _m_empty();
}

//----------------------------------------------------------------------------------------------
// 64-QAM
//----------------------------------------------------------------------------------------------

void dlsch_64qam_llr(LTE_DL_FRAME_PARMS *frame_parms,
                     int **rxdataF_comp,
                     short *dlsch_llr,
                     int **dl_ch_mag,
                     int **dl_ch_magb,
                     unsigned char symbol,
gauthier's avatar
gauthier committed
772
                     uint8_t first_symbol_flag,
773
                     unsigned short nb_rb,
gauthier's avatar
gauthier committed
774
                     uint16_t pbch_pss_sss_adjust,
775 776 777 778
                     short **llr_save) {

    __m128i *rxF = (__m128i*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)];
    __m128i *ch_mag,*ch_magb;
779
    int i,len,len2;
780 781
    unsigned char symbol_mod,len_mod4;
    short *llr;
gauthier's avatar
gauthier committed
782
    int16_t *llr2;
783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810
    
    if (first_symbol_flag==1)
        llr = dlsch_llr;
    else
        llr = *llr_save;
    
    symbol_mod = (symbol>=(7-frame_parms->Ncp)) ? symbol-(7-frame_parms->Ncp) : symbol;
    
    ch_mag = (__m128i*)&dl_ch_mag[0][(symbol*frame_parms->N_RB_DL*12)];
    ch_magb = (__m128i*)&dl_ch_magb[0][(symbol*frame_parms->N_RB_DL*12)];

    if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) {
        if (frame_parms->mode1_flag==0)
            len = nb_rb*8 - (2*pbch_pss_sss_adjust/3);
        else
            len = nb_rb*10 - (5*pbch_pss_sss_adjust/6);
    }
    else {
        len = nb_rb*12 - pbch_pss_sss_adjust;
    }

    llr2 = llr;
    llr += (len*6);

    len_mod4 =len&3;
    len2=len>>2;  // length in quad words (4 REs)
    len2+=(len_mod4?0:1);

knopp's avatar
 
knopp committed
811
    for (i=0;i<len2;i++) {
812 813
        
        xmm1 = _mm_abs_epi16(rxF[i]);
knopp's avatar
 
knopp committed
814
        xmm1 = _mm_subs_epi16(ch_mag[i],xmm1);
815 816 817 818
        xmm2 = _mm_abs_epi16(xmm1);
        xmm2 = _mm_subs_epi16(ch_magb[i],xmm2);
        
        // loop over all LLRs in quad word (24 coded bits)
819
	/*
820 821 822 823 824 825 826 827 828 829
        for (j=0;j<8;j+=2) {
            llr2[0] = ((short *)&rxF[i])[j];
            llr2[1] = ((short *)&rxF[i])[j+1];
            llr2[2] = ((short *)&xmm1)[j];
            llr2[3] = ((short *)&xmm1)[j+1];
            llr2[4] = ((short *)&xmm2)[j];
            llr2[5] = ((short *)&xmm2)[j+1];
            
            llr2+=6;
        }
830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863
	*/
      llr2[0] = ((short *)&rxF[i])[0];
      llr2[1] = ((short *)&rxF[i])[1];
      llr2[2] = _mm_extract_epi16(xmm1,0);
      llr2[3] = _mm_extract_epi16(xmm1,1);//((short *)&xmm1)[j+1];
      llr2[4] = _mm_extract_epi16(xmm2,0);//((short *)&xmm2)[j];
      llr2[5] = _mm_extract_epi16(xmm2,1);//((short *)&xmm2)[j+1];
      
      llr2+=6;
      llr2[0] = ((short *)&rxF[i])[2];
      llr2[1] = ((short *)&rxF[i])[3];
      llr2[2] = _mm_extract_epi16(xmm1,2);
      llr2[3] = _mm_extract_epi16(xmm1,3);//((short *)&xmm1)[j+1];
      llr2[4] = _mm_extract_epi16(xmm2,2);//((short *)&xmm2)[j];
      llr2[5] = _mm_extract_epi16(xmm2,3);//((short *)&xmm2)[j+1];
      
      llr2+=6;
      llr2[0] = ((short *)&rxF[i])[4];
      llr2[1] = ((short *)&rxF[i])[5];
      llr2[2] = _mm_extract_epi16(xmm1,4);
      llr2[3] = _mm_extract_epi16(xmm1,5);//((short *)&xmm1)[j+1];
      llr2[4] = _mm_extract_epi16(xmm2,4);//((short *)&xmm2)[j];
      llr2[5] = _mm_extract_epi16(xmm2,5);//((short *)&xmm2)[j+1];
      
      llr2+=6;
      llr2[0] = ((short *)&rxF[i])[6];
      llr2[1] = ((short *)&rxF[i])[7];
      llr2[2] = _mm_extract_epi16(xmm1,6);
      llr2[3] = _mm_extract_epi16(xmm1,7);//((short *)&xmm1)[j+1];
      llr2[4] = _mm_extract_epi16(xmm2,6);//((short *)&xmm2)[j];
      llr2[5] = _mm_extract_epi16(xmm2,7);//((short *)&xmm2)[j+1];
      
      llr2+=6;

864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901
    }
    *llr_save = llr;
    _mm_empty();
    _m_empty();
}


//==============================================================================================
// DUAL-STREAM
//==============================================================================================

//----------------------------------------------------------------------------------------------
// QPSK
//----------------------------------------------------------------------------------------------

NOCYGWIN_STATIC __m128i  y0r_over2 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0i_over2 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y1r_over2 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y1i_over2 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i  A __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  B __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  C __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  D __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  E __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  F __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  G __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  H __attribute__ ((aligned(16))); 


int dlsch_qpsk_qpsk_llr(LTE_DL_FRAME_PARMS *frame_parms,
                        int **rxdataF_comp,
                        int **rxdataF_comp_i,
                        int **rho_i,
                        short *dlsch_llr,
                        unsigned char symbol,
                        unsigned char first_symbol_flag,
                        unsigned short nb_rb,
gauthier's avatar
gauthier committed
902
                        uint16_t pbch_pss_sss_adjust,
903 904
                        short **llr16p) {
    
gauthier's avatar
gauthier committed
905 906 907 908
    int16_t *rxF=(int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *rxF_i=(int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *rho=(int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *llr16;
909
    int len;
gauthier's avatar
gauthier committed
910
    uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol;
911 912
    
    if (first_symbol_flag == 1) {
gauthier's avatar
gauthier committed
913
        llr16 = (int16_t*)dlsch_llr;
914 915
    }
    else {
gauthier's avatar
gauthier committed
916
        llr16 = (int16_t*)(*llr16p);
917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
    }
    
    if (!llr16) {
        msg("dlsch_qpsk_qpsk_llr: llr is null, symbol %d\n",symbol);
        return -1;
    }
    
    if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) {
        // if symbol has pilots
        if (frame_parms->mode1_flag==0)
            // in 2 antenna ports we have 8 REs per symbol per RB
            len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3);
        else
            // for 1 antenna port we have 10 REs per symbol per RB 
            len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6);
    }
    else {
        // symbol has no pilots
        len = (nb_rb*12) - pbch_pss_sss_adjust;
    }

    //    printf("qpsk_qpsk: len %d, llr16 %p\n",len,llr16);
    qpsk_qpsk((short *)rxF,
              (short *)rxF_i,
              (short *)llr16,
              (short *)rho,
              len);
    
    llr16 += (len<<1);
    *llr16p = (short *)llr16;
    
    return(0);
}

NOCYGWIN_STATIC __m128i ONE_OVER_SQRT_8 __attribute__((aligned(16))); 

void qpsk_qpsk(short *stream0_in,
               short *stream1_in, 
               short *stream0_out,
               short *rho01,
               int length
               ) {

/* 
This function computes the LLRs of stream 0 (s_0) in presence of the interfering stream 1 (s_1) assuming that both symbols are QPSK. It can be used for both MU-MIMO interference-aware receiver or for SU-MIMO receivers.

Parameters:
stream0_in = Matched filter output y0' = (h0*g0)*y0
stream1_in = Matched filter output y1' = (h0*g1)*y0
stream0_out = LLRs
rho01 = Correlation between the two effective channels \rho_{10} = (h1*g1)*(h0*g0)
length = number of resource elements
*/

    __m128i *rho01_128i = (__m128i *)rho01;
    __m128i *stream0_128i_in = (__m128i *)stream0_in;
    __m128i *stream1_128i_in = (__m128i *)stream1_in;
    __m128i *stream0_128i_out = (__m128i *)stream0_out;

#ifdef DEBUG_LLR
    print_shorts2("rho01_128i:\n",rho01_128i);
#endif

    int i;
    ONE_OVER_SQRT_8 = _mm_set1_epi16(23170); //round(2^16/sqrt(8))

    for (i=0;i<length>>2;i+=2) {
        // in each iteration, we take 8 complex samples

        xmm0 = rho01_128i[i]; // 4 symbols
        xmm1 = rho01_128i[i+1]; 
    
        // put (rho_r + rho_i)/2sqrt2 in rho_rpi
        // put (rho_r - rho_i)/2sqrt2 in rho_rmi
        xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)]
        //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)]
        xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho)
        xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho)
        rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho)
        rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho)
         
        // divide by sqrt(8), no shift needed ONE_OVER_SQRT_8 = Q1.16
        rho_rpi = _mm_mulhi_epi16(rho_rpi,ONE_OVER_SQRT_8);
        rho_rmi = _mm_mulhi_epi16(rho_rmi,ONE_OVER_SQRT_8);

        // Compute LLR for first bit of stream 0
        
        // Compute real and imaginary parts of MF output for stream 0
        xmm0 = stream0_128i_in[i];
        xmm1 = stream0_128i_in[i+1];

        xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)]
        //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)]
        y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)]
        y0i = _mm_unpackhi_epi64(xmm0,xmm1); 

        y0r_over2  = _mm_srai_epi16(y0r,1);   // divide by 2                    
        y0i_over2  = _mm_srai_epi16(y0i,1);   // divide by 2 
    
        // Compute real and imaginary parts of MF output for stream 1
        xmm0 = stream1_128i_in[i];
        xmm1 = stream1_128i_in[i+1];

        xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)]
        //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)]
        y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)]
        y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)]
        
        y1r_over2  = _mm_srai_epi16(y1r,1);   // divide by 2                    
        y1i_over2  = _mm_srai_epi16(y1i,1);   // divide by 2
    
        // Compute the terms for the LLR of first bit
    
1048
        xmm0 = _mm_setzero_si128(); // ZERO
1049
    
1050
        // 1 term for numerator of LLR
1051 1052 1053 1054 1055 1056 1057
        xmm3 = _mm_subs_epi16(y1r_over2,rho_rpi);
        A = _mm_abs_epi16(xmm3); // A = |y1r/2 - rho/sqrt(8)|
        xmm2 = _mm_adds_epi16(A,y0i_over2); // = |y1r/2 - rho/sqrt(8)| + y0i/2
        xmm3 = _mm_subs_epi16(y1i_over2,rho_rmi); 
        B = _mm_abs_epi16(xmm3); // B = |y1i/2 - rho*/sqrt(8)|
        logmax_num_re0 = _mm_adds_epi16(B,xmm2); // = |y1r/2 - rho/sqrt(8)|+|y1i/2 - rho*/sqrt(8)| + y0i/2

1058
        // 2 term for numerator of LLR        
1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130
        xmm3 = _mm_subs_epi16(y1r_over2,rho_rmi); 
        C = _mm_abs_epi16(xmm3); // C = |y1r/2 - rho*/4|
        xmm2 = _mm_subs_epi16(C,y0i_over2); // = |y1r/2 - rho*/4| - y0i/2
        xmm3 = _mm_adds_epi16(y1i_over2,rho_rpi); 
        D = _mm_abs_epi16(xmm3); // D = |y1i/2 + rho/4|      
        xmm2 = _mm_adds_epi16(xmm2,D); // |y1r/2 - rho*/4| + |y1i/2 + rho/4| - y0i/2
        logmax_num_re0 = _mm_max_epi16(logmax_num_re0,xmm2); // max, numerator done

        // 1 term for denominator of LLR
        xmm3 = _mm_adds_epi16(y1r_over2,rho_rmi); 
        E = _mm_abs_epi16(xmm3); // E = |y1r/2 + rho*/4|
        xmm2 = _mm_adds_epi16(E,y0i_over2); // = |y1r/2 + rho*/4| + y0i/2
        xmm3 = _mm_subs_epi16(y1i_over2,rho_rpi); 
        F = _mm_abs_epi16(xmm3); // F = |y1i/2 - rho/4|
        logmax_den_re0 = _mm_adds_epi16(F,xmm2); // = |y1r/2 + rho*/4| + |y1i/2 - rho/4| + y0i/2

        // 2 term for denominator of LLR
        xmm3 = _mm_adds_epi16(y1r_over2,rho_rpi); 
        G = _mm_abs_epi16(xmm3); // G = |y1r/2 + rho/4|      
        xmm2 = _mm_subs_epi16(G,y0i_over2); // = |y1r/2 + rho/4| - y0i/2
        xmm3 = _mm_adds_epi16(y1i_over2,rho_rmi); 
        H = _mm_abs_epi16(xmm3); // H = |y1i/2 + rho*/4|
        xmm2 = _mm_adds_epi16(xmm2,H); // = |y1r/2 + rho/4| + |y1i/2 + rho*/4| - y0i/2    
        logmax_den_re0 = _mm_max_epi16(logmax_den_re0,xmm2); // max, denominator done

        // Compute the terms for the LLR of first bit

        // 1 term for nominator of LLR    
        xmm2 = _mm_adds_epi16(A,y0r_over2); 
        logmax_num_im0 = _mm_adds_epi16(B,xmm2); // = |y1r/2 - rho/4| + |y1i/2 - rho*/4| + y0r/2
 
        // 2 term for nominator of LLR
        xmm2 = _mm_subs_epi16(E,y0r_over2); 
        xmm2 = _mm_adds_epi16(xmm2,F); // = |y1r/2 + rho*/4| + |y1i/2 - rho/4| - y0r/2

        logmax_num_im0 = _mm_max_epi16(logmax_num_im0,xmm2); // max, nominator done
    
        // 1 term for denominator of LLR
        xmm2 = _mm_adds_epi16(C,y0r_over2); 
        logmax_den_im0 = _mm_adds_epi16(D,xmm2); // = |y1r/2 - rho*/4| + |y1i/2 + rho/4| - y0r/2

        xmm2 = _mm_subs_epi16(G,y0r_over2); 
        xmm2 = _mm_adds_epi16(xmm2,H); // = |y1r/2 + rho/4| + |y1i/2 + rho*/4| - y0r/2
    
        logmax_den_im0 = _mm_max_epi16(logmax_den_im0,xmm2); // max, denominator done

        // LLR of first bit [L1(1), L1(2), L1(3), L1(4)]
        y0r = _mm_adds_epi16(y0r,logmax_num_re0);
        y0r = _mm_subs_epi16(y0r,logmax_den_re0);
    
        // LLR of second bit [L2(1), L2(2), L2(3), L2(4)]
        y0i = _mm_adds_epi16(y0i,logmax_num_im0);
        y0i = _mm_subs_epi16(y0i,logmax_den_im0);

	_mm_storeu_si128(&stream0_128i_out[i],_mm_unpacklo_epi16(y0r,y0i)); // = [L1(1), L2(1), L1(2), L2(2)]
	if (i<((length>>1) - 1)) // false if only 2 REs remain
	  _mm_storeu_si128(&stream0_128i_out[i+1],_mm_unpackhi_epi16(y0r,y0i));
	
    }
    _mm_empty();
    _m_empty();
}

int dlsch_qpsk_16qam_llr(LTE_DL_FRAME_PARMS *frame_parms,
                         int **rxdataF_comp,
                         int **rxdataF_comp_i,
                         int **dl_ch_mag_i, //|h_1|^2*(2/sqrt{10})
                         int **rho_i,
                         short *dlsch_llr,
                         unsigned char symbol,
                         unsigned char first_symbol_flag,
                         unsigned short nb_rb,
gauthier's avatar
gauthier committed
1131
                         uint16_t pbch_pss_sss_adjust,
1132 1133
                         short **llr16p) {
    
gauthier's avatar
gauthier committed
1134 1135 1136 1137 1138
    int16_t *rxF=(int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *rxF_i=(int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *ch_mag_i = (int16_t*)&dl_ch_mag_i[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *rho=(int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *llr16;
1139
    int len;
gauthier's avatar
gauthier committed
1140
    uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol;
1141 1142
    
    if (first_symbol_flag == 1) {
gauthier's avatar
gauthier committed
1143
        llr16 = (int16_t*)dlsch_llr;
1144 1145
    }
    else {
gauthier's avatar
gauthier committed
1146
        llr16 = (int16_t*)(*llr16p);
1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291
    }
    
    if (!llr16) {
        msg("dlsch_qpsk_qpsk_llr: llr is null, symbol %d\n",symbol);
        return -1;
    }
    
    if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) {
        // if symbol has pilots
        if (frame_parms->mode1_flag==0)
            // in 2 antenna ports we have 8 REs per symbol per RB
            len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3);
        else
            // for 1 antenna port we have 10 REs per symbol per RB 
            len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6);
    }
    else {
        // symbol has no pilots
        len = (nb_rb*12) - pbch_pss_sss_adjust;
    }
    
    qpsk_qam16((short *)rxF,
               (short *)rxF_i,
               (short *)ch_mag_i,
               (short *)llr16,
               (short *)rho,
               len);
    
    llr16 += (len<<1);
    *llr16p = (short *)llr16;
    
    return(0);
}

NOCYGWIN_STATIC __m128i ONE_OVER_SQRT_2 __attribute__((aligned(16))); 
NOCYGWIN_STATIC __m128i ONE_OVER_SQRT_10 __attribute__((aligned(16)));
NOCYGWIN_STATIC __m128i THREE_OVER_SQRT_10 __attribute__((aligned(16)));
NOCYGWIN_STATIC __m128i ONE_OVER_SQRT_10_Q15 __attribute__((aligned(16)));
NOCYGWIN_STATIC __m128i SQRT_10_OVER_FOUR __attribute__((aligned(16)));
NOCYGWIN_STATIC __m128i ch_mag_int __attribute__((aligned(16))); 

void qpsk_qam16(short *stream0_in,
                short *stream1_in, 
                short *ch_mag_i,
                short *stream0_out,
                short *rho01,
                int length
                ) {

/* 
This function computes the LLRs of stream 0 (s_0) in presence of the interfering stream 1 (s_1) assuming that both symbols are QPSK. It can be used for both MU-MIMO interference-aware receiver or for SU-MIMO receivers.

Parameters:
stream0_in = Matched filter output y0' = (h0*g0)*y0
stream1_in = Matched filter output y1' = (h0*g1)*y0
stream0_out = LLRs
rho01 = Correlation between the two effective channels \rho_{10} = (h1*g1)*(h0*g0)
length = number of resource elements
*/

    __m128i *rho01_128i = (__m128i *)rho01;
    __m128i *stream0_128i_in = (__m128i *)stream0_in;
    __m128i *stream1_128i_in = (__m128i *)stream1_in;
    __m128i *stream0_128i_out = (__m128i *)stream0_out;
    __m128i *ch_mag_128i_i    = (__m128i *)ch_mag_i;

#ifdef DEBUG_LLR
    print_shorts2("rho01_128i:\n",rho01_128i);
#endif

    int i;

    ONE_OVER_SQRT_2 = _mm_set1_epi16(23170); // round(1/sqrt(2)*2^15)
    ONE_OVER_SQRT_10_Q15 = _mm_set1_epi16(10362); // round(1/sqrt(10)*2^15)
    THREE_OVER_SQRT_10 = _mm_set1_epi16(31086); // round(3/sqrt(10)*2^15)
    SQRT_10_OVER_FOUR = _mm_set1_epi16(25905); // round(sqrt(10)/4*2^15)

    for (i=0;i<length>>2;i+=2) {
        // in each iteration, we take 8 complex samples

        xmm0 = rho01_128i[i]; // 4 symbols
        xmm1 = rho01_128i[i+1]; 
    
        // put (rho_r + rho_i)/2sqrt2 in rho_rpi
        // put (rho_r - rho_i)/2sqrt2 in rho_rmi
        xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)]
        //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)]
        xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho)
        xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho)
        rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho)
        rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho)
        
        // divide by sqrt(2)
        rho_rpi = _mm_mulhi_epi16(rho_rpi, ONE_OVER_SQRT_2);
        rho_rmi = _mm_mulhi_epi16(rho_rmi, ONE_OVER_SQRT_2);
        rho_rpi = _mm_slli_epi16(rho_rpi,1);
        rho_rmi = _mm_slli_epi16(rho_rmi,1);
 
        // Compute LLR for first bit of stream 0
        
        // Compute real and imaginary parts of MF output for stream 0
        xmm0 = stream0_128i_in[i];
        xmm1 = stream0_128i_in[i+1];

        xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)]
        //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)]
        y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)]
        y0i = _mm_unpackhi_epi64(xmm0,xmm1); 

        // divide by sqrt(2)
        y0r_over2 = _mm_mulhi_epi16(y0r, ONE_OVER_SQRT_2);
        y0i_over2 = _mm_mulhi_epi16(y0i, ONE_OVER_SQRT_2);
        y0r_over2  = _mm_slli_epi16(y0r,1);
        y0i_over2  = _mm_slli_epi16(y0i,1);
    
        y0_p_1_1 = _mm_adds_epi16(y0r_over2, y0i_over2);
        y0_m_1_1 = _mm_subs_epi16(y0r_over2, y0i_over2);

        // Compute real and imaginary parts of MF output for stream 1
        xmm0 = stream1_128i_in[i];
        xmm1 = stream1_128i_in[i+1];

        xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)]
        //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)]
        y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)]
        y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)]
    
1292
        xmm0 = _mm_setzero_si128(); // ZERO
1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389
    
        // compute psi
        xmm3 = _mm_subs_epi16(y1r,rho_rpi);
        psi_r_p1_p1 = _mm_abs_epi16(xmm3);    
        xmm3 = _mm_subs_epi16(y1i,rho_rmi);
        psi_i_p1_p1 = _mm_abs_epi16(xmm3);     
        xmm3 = _mm_subs_epi16(y1r,rho_rmi);
        psi_r_p1_m1 = _mm_abs_epi16(xmm3);           
        xmm3 = _mm_adds_epi16(y1i,rho_rpi);
        psi_i_p1_m1 = _mm_abs_epi16(xmm3);     
        xmm3 = _mm_adds_epi16(y1r,rho_rmi);
        psi_r_m1_p1 = _mm_abs_epi16(xmm3);     
        xmm3 = _mm_subs_epi16(y1i,rho_rpi);
        psi_i_m1_p1 = _mm_abs_epi16(xmm3); 
        xmm3 = _mm_adds_epi16(y1r,rho_rpi);
        psi_r_m1_m1 = _mm_abs_epi16(xmm3);         
        xmm3 = _mm_adds_epi16(y1i,rho_rmi);
        psi_i_m1_m1 = _mm_abs_epi16(xmm3);      
    
        // Rearrange interfering channel magnitudes
        xmm2 = ch_mag_128i_i[i];   
        xmm3 = ch_mag_128i_i[i+1]; 

        xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3));

        ch_mag_int = _mm_unpacklo_epi64(xmm2,xmm3); 
        
        // calculate optimal interference amplitudes		
        interference_abs_epi16(psi_r_p1_p1 , ch_mag_int, a_r_p1_p1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10);
        interference_abs_epi16(psi_i_p1_p1 , ch_mag_int, a_i_p1_p1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10);	
        interference_abs_epi16(psi_r_p1_m1 , ch_mag_int, a_r_p1_m1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10);
        interference_abs_epi16(psi_i_p1_m1 , ch_mag_int, a_i_p1_m1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10);	
        interference_abs_epi16(psi_r_m1_p1 , ch_mag_int, a_r_m1_p1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10);
        interference_abs_epi16(psi_i_m1_p1 , ch_mag_int, a_i_m1_p1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10);
        interference_abs_epi16(psi_r_m1_m1 , ch_mag_int, a_r_m1_m1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10);
        interference_abs_epi16(psi_i_m1_m1 , ch_mag_int, a_i_m1_m1 , ONE_OVER_SQRT_10_Q15, THREE_OVER_SQRT_10);
        
        // prodsum
        prodsum_psi_a_epi16(psi_r_p1_p1, a_r_p1_p1, psi_i_p1_p1, a_i_p1_p1, psi_a_p1_p1);
        prodsum_psi_a_epi16(psi_r_p1_m1, a_r_p1_m1, psi_i_p1_m1, a_i_p1_m1, psi_a_p1_m1);
        prodsum_psi_a_epi16(psi_r_m1_p1, a_r_m1_p1, psi_i_m1_p1, a_i_m1_p1, psi_a_m1_p1);
        prodsum_psi_a_epi16(psi_r_m1_m1, a_r_m1_m1, psi_i_m1_m1, a_i_m1_m1, psi_a_m1_m1);
        
        // squares
        square_a_epi16(a_r_p1_p1, a_i_p1_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_p1);
        square_a_epi16(a_r_p1_m1, a_i_p1_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_p1_m1);
        square_a_epi16(a_r_m1_p1, a_i_m1_p1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_p1);
        square_a_epi16(a_r_m1_m1, a_i_m1_m1, ch_mag_int, SQRT_10_OVER_FOUR, a_sq_m1_m1);

       // Computing Metrics
        xmm0 = _mm_subs_epi16(psi_a_p1_p1, a_sq_p1_p1);
        bit_met_p1_p1 = _mm_adds_epi16(xmm0, y0_p_1_1);
        
        xmm0 = _mm_subs_epi16(psi_a_p1_m1, a_sq_p1_m1);
        bit_met_p1_m1 = _mm_adds_epi16(xmm0, y0_m_1_1);
        
        xmm0 = _mm_subs_epi16(psi_a_m1_p1, a_sq_m1_p1);
        bit_met_m1_p1 = _mm_subs_epi16(xmm0, y0_m_1_1);
        
        xmm0 = _mm_subs_epi16(psi_a_m1_m1, a_sq_m1_m1);
        bit_met_m1_m1 = _mm_subs_epi16(xmm0, y0_p_1_1);
        
        // MSB
        logmax_num_re0 = _mm_max_epi16(bit_met_p1_p1,bit_met_p1_m1); // bit=0
        logmax_den_re0 = _mm_max_epi16(bit_met_m1_p1,bit_met_m1_m1); // bit=1
        
        y0r = _mm_subs_epi16(logmax_num_re0,logmax_den_re0);
        
        // LSB
        logmax_num_im0 = _mm_max_epi16(bit_met_p1_p1,bit_met_m1_p1); // bit=0
        logmax_den_im0 = _mm_max_epi16(bit_met_p1_m1,bit_met_m1_m1); // bit=1
        
        y0i = _mm_subs_epi16(logmax_num_im0,logmax_den_im0);
        
        stream0_128i_out[i] = _mm_unpacklo_epi16(y0r,y0i); // = [L1(1), L2(1), L1(2), L2(2)]
        if (i<((length>>1) - 1)) // false if only 2 REs remain
            stream0_128i_out[i+1] = _mm_unpackhi_epi16(y0r,y0i);
   
    }
    _mm_empty();
    _m_empty();
}

int dlsch_qpsk_64qam_llr(LTE_DL_FRAME_PARMS *frame_parms,
                         int **rxdataF_comp,
                         int **rxdataF_comp_i,
                         int **dl_ch_mag_i, //|h_1|^2*(2/sqrt{10})
                         int **rho_i,
                         short *dlsch_llr,
                         unsigned char symbol,
                         unsigned char first_symbol_flag,
                         unsigned short nb_rb,
gauthier's avatar
gauthier committed
1390
                         uint16_t pbch_pss_sss_adjust,
1391 1392
                         short **llr16p) {
    
gauthier's avatar
gauthier committed
1393 1394 1395 1396 1397
    int16_t *rxF=(int16_t*)&rxdataF_comp[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *rxF_i=(int16_t*)&rxdataF_comp_i[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *ch_mag_i = (int16_t*)&dl_ch_mag_i[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *rho=(int16_t*)&rho_i[0][(symbol*frame_parms->N_RB_DL*12)];
    int16_t *llr16;
1398
    int len;
gauthier's avatar
gauthier committed
1399
    uint8_t symbol_mod = (symbol >= (7-frame_parms->Ncp))? (symbol-(7-frame_parms->Ncp)) : symbol;
1400 1401
    
    if (first_symbol_flag == 1) {
gauthier's avatar
gauthier committed
1402
        llr16 = (int16_t*)dlsch_llr;
1403 1404
    }
    else {
gauthier's avatar
gauthier committed
1405
        llr16 = (int16_t*)(*llr16p);
1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555
    }
    
    if (!llr16) {
        msg("dlsch_qpsk_qam64_llr: llr is null, symbol %d\n",symbol);
        return -1;
    }
    
    if ((symbol_mod==0) || (symbol_mod==(4-frame_parms->Ncp))) {
        // if symbol has pilots
        if (frame_parms->mode1_flag==0)
            // in 2 antenna ports we have 8 REs per symbol per RB
            len = (nb_rb*8) - (2*pbch_pss_sss_adjust/3);
        else
            // for 1 antenna port we have 10 REs per symbol per RB 
            len = (nb_rb*10) - (5*pbch_pss_sss_adjust/6);
    }
    else {
        // symbol has no pilots
        len = (nb_rb*12) - pbch_pss_sss_adjust;
    }
    
    qpsk_qam64((short *)rxF,
               (short *)rxF_i,
               (short *)ch_mag_i,
               (short *)llr16,
               (short *)rho,
               len);
    
    llr16 += (len<<1);
    *llr16p = (short *)llr16;
    
    return(0);
}

NOCYGWIN_STATIC __m128i ONE_OVER_SQRT_2_42 __attribute__((aligned(16))); 
NOCYGWIN_STATIC __m128i THREE_OVER_SQRT_2_42 __attribute__((aligned(16))); 
NOCYGWIN_STATIC __m128i FIVE_OVER_SQRT_2_42 __attribute__((aligned(16))); 
NOCYGWIN_STATIC __m128i SEVEN_OVER_SQRT_2_42 __attribute__((aligned(16))); 

NOCYGWIN_STATIC __m128i ch_mag_int_with_sigma2 __attribute__((aligned(16))); 
NOCYGWIN_STATIC __m128i two_ch_mag_int_with_sigma2 __attribute__((aligned(16))); 
NOCYGWIN_STATIC __m128i three_ch_mag_int_with_sigma2 __attribute__((aligned(16))); 
NOCYGWIN_STATIC __m128i SQRT_42_OVER_FOUR __attribute__((aligned(16))); 

void qpsk_qam64(short *stream0_in,
                short *stream1_in, 
                short *ch_mag_i,
                short *stream0_out,
                short *rho01,
                int length
                ) {

/* 
This function computes the LLRs of stream 0 (s_0) in presence of the interfering stream 1 (s_1) assuming that both symbols are QPSK. It can be used for both MU-MIMO interference-aware receiver or for SU-MIMO receivers.

Parameters:
stream0_in = Matched filter output y0' = (h0*g0)*y0
stream1_in = Matched filter output y1' = (h0*g1)*y0
stream0_out = LLRs
rho01 = Correlation between the two effective channels \rho_{10} = (h1*g1)*(h0*g0)
length = number of resource elements
*/

    __m128i *rho01_128i = (__m128i *)rho01;
    __m128i *stream0_128i_in = (__m128i *)stream0_in;
    __m128i *stream1_128i_in = (__m128i *)stream1_in;
    __m128i *stream0_128i_out = (__m128i *)stream0_out;
    __m128i *ch_mag_128i_i    = (__m128i *)ch_mag_i;

#ifdef DEBUG_LLR
    print_shorts2("rho01_128i:\n",rho01_128i);
#endif

    int i;

    ONE_OVER_SQRT_2 = _mm_set1_epi16(23170); // round(1/sqrt(2)*2^15)
    ONE_OVER_SQRT_2_42 = _mm_set1_epi16(3575); // round(1/sqrt(2*42)*2^15)
    THREE_OVER_SQRT_2_42 = _mm_set1_epi16(10726); // round(3/sqrt(2*42)*2^15)
    FIVE_OVER_SQRT_2_42 = _mm_set1_epi16(17876); // round(5/sqrt(2*42)*2^15)
    SEVEN_OVER_SQRT_2_42 = _mm_set1_epi16(25027); // round(7/sqrt(2*42)*2^15)
    SQRT_42_OVER_FOUR = _mm_set1_epi16(13272); // round(sqrt(42)/4*2^13), Q3.12

    for (i=0;i<length>>2;i+=2) {
        // in each iteration, we take 8 complex samples

        xmm0 = rho01_128i[i]; // 4 symbols
        xmm1 = rho01_128i[i+1]; 
    
        // put (rho_r + rho_i)/sqrt2 in rho_rpi
        // put (rho_r - rho_i)/sqrt2 in rho_rmi
        xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)]
        //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)]
        xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho)
        xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho)
        rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho)
        rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho)
        
        // divide by sqrt(2)
        rho_rpi = _mm_mulhi_epi16(rho_rpi, ONE_OVER_SQRT_2);
        rho_rmi = _mm_mulhi_epi16(rho_rmi, ONE_OVER_SQRT_2);
        rho_rpi = _mm_slli_epi16(rho_rpi,1);
        rho_rmi = _mm_slli_epi16(rho_rmi,1);
 
        // Compute LLR for first bit of stream 0
        
        // Compute real and imaginary parts of MF output for stream 0
        xmm0 = stream0_128i_in[i];
        xmm1 = stream0_128i_in[i+1];

        xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)]
        //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)]
        y0r = _mm_unpacklo_epi64(xmm0,xmm1); // = [y0r(1),y0r(2),y0r(3),y0r(4)]
        y0i = _mm_unpackhi_epi64(xmm0,xmm1); 

        // divide by sqrt(2)
        y0r_over2 = _mm_mulhi_epi16(y0r, ONE_OVER_SQRT_2);
        y0i_over2 = _mm_mulhi_epi16(y0i, ONE_OVER_SQRT_2);
        y0r_over2  = _mm_slli_epi16(y0r,1);
        y0i_over2  = _mm_slli_epi16(y0i,1);
    
        y0_p_1_1 = _mm_adds_epi16(y0r_over2, y0i_over2);
        y0_m_1_1 = _mm_subs_epi16(y0r_over2, y0i_over2);

        // Compute real and imaginary parts of MF output for stream 1
        xmm0 = stream1_128i_in[i];
        xmm1 = stream1_128i_in[i+1];

        xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)]
        //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)]
        y1r = _mm_unpacklo_epi64(xmm0,xmm1); //[y1r(1),y1r(2),y1r(3),y1r(4)]
        y1i = _mm_unpackhi_epi64(xmm0,xmm1); //[y1i(1),y1i(2),y1i(3),y1i(4)]
    
1556
        xmm0 = _mm_setzero_si128(); // ZERO
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766
    
        // compute psi
        xmm3 = _mm_subs_epi16(y1r,rho_rpi);
        psi_r_p1_p1 = _mm_abs_epi16(xmm3);    
        xmm3 = _mm_subs_epi16(y1i,rho_rmi);
        psi_i_p1_p1 = _mm_abs_epi16(xmm3);     
        xmm3 = _mm_subs_epi16(y1r,rho_rmi);
        psi_r_p1_m1 = _mm_abs_epi16(xmm3);           
        xmm3 = _mm_adds_epi16(y1i,rho_rpi);
        psi_i_p1_m1 = _mm_abs_epi16(xmm3);     
        xmm3 = _mm_adds_epi16(y1r,rho_rmi);
        psi_r_m1_p1 = _mm_abs_epi16(xmm3);     
        xmm3 = _mm_subs_epi16(y1i,rho_rpi);
        psi_i_m1_p1 = _mm_abs_epi16(xmm3); 
        xmm3 = _mm_adds_epi16(y1r,rho_rpi);
        psi_r_m1_m1 = _mm_abs_epi16(xmm3);         
        xmm3 = _mm_adds_epi16(y1i,rho_rmi);
        psi_i_m1_m1 = _mm_abs_epi16(xmm3);      
    
        // Rearrange interfering channel magnitudes
        xmm2 = ch_mag_128i_i[i];   
        xmm3 = ch_mag_128i_i[i+1]; 

        xmm2 = _mm_shufflelo_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm2 = _mm_shufflehi_epi16(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm2 = _mm_shuffle_epi32(xmm2,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm3 = _mm_shufflelo_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm3 = _mm_shufflehi_epi16(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm3 = _mm_shuffle_epi32(xmm3,0xd8); //_MM_SHUFFLE(0,2,1,3));

        ch_mag_int = _mm_unpacklo_epi64(xmm2,xmm3); 
        ch_mag_int_with_sigma2       = _mm_srai_epi16(ch_mag_int, 1); // *2
        two_ch_mag_int_with_sigma2   = ch_mag_int; // *4
        three_ch_mag_int_with_sigma2 = _mm_adds_epi16(ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2); // *6

        interference_abs_64qam_epi16(psi_r_p1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, SEVEN_OVER_SQRT_2_42);    
        interference_abs_64qam_epi16(psi_i_p1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, SEVEN_OVER_SQRT_2_42);    
        interference_abs_64qam_epi16(psi_r_p1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_p1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, SEVEN_OVER_SQRT_2_42);    
        interference_abs_64qam_epi16(psi_i_p1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_p1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, SEVEN_OVER_SQRT_2_42);    
        interference_abs_64qam_epi16(psi_r_m1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, SEVEN_OVER_SQRT_2_42);    
        interference_abs_64qam_epi16(psi_i_m1_p1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_p1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, SEVEN_OVER_SQRT_2_42);    
        interference_abs_64qam_epi16(psi_r_m1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_r_m1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, SEVEN_OVER_SQRT_2_42);    
        interference_abs_64qam_epi16(psi_i_m1_m1, ch_mag_int_with_sigma2, two_ch_mag_int_with_sigma2, three_ch_mag_int_with_sigma2, a_i_m1_m1, ONE_OVER_SQRT_2_42, THREE_OVER_SQRT_2_42, FIVE_OVER_SQRT_2_42, SEVEN_OVER_SQRT_2_42);    
        
        // prodsum
        prodsum_psi_a_epi16(psi_r_p1_p1, a_r_p1_p1, psi_i_p1_p1, a_i_p1_p1, psi_a_p1_p1);
        prodsum_psi_a_epi16(psi_r_p1_m1, a_r_p1_m1, psi_i_p1_m1, a_i_p1_m1, psi_a_p1_m1);
        prodsum_psi_a_epi16(psi_r_m1_p1, a_r_m1_p1, psi_i_m1_p1, a_i_m1_p1, psi_a_m1_p1);
        prodsum_psi_a_epi16(psi_r_m1_m1, a_r_m1_m1, psi_i_m1_m1, a_i_m1_m1, psi_a_m1_m1);

        // Multiply by sqrt(2)
        psi_a_p1_p1 = _mm_mulhi_epi16(psi_a_p1_p1, ONE_OVER_SQRT_2);
        psi_a_p1_p1 = _mm_slli_epi16(psi_a_p1_p1, 2);
        psi_a_p1_m1 = _mm_mulhi_epi16(psi_a_p1_m1, ONE_OVER_SQRT_2);
        psi_a_p1_m1 = _mm_slli_epi16(psi_a_p1_m1, 2);
        psi_a_m1_p1 = _mm_mulhi_epi16(psi_a_m1_p1, ONE_OVER_SQRT_2);
        psi_a_m1_p1 = _mm_slli_epi16(psi_a_m1_p1, 2);
        psi_a_m1_m1 = _mm_mulhi_epi16(psi_a_m1_m1, ONE_OVER_SQRT_2);
        psi_a_m1_m1 = _mm_slli_epi16(psi_a_m1_m1, 2);
        
        square_a_64qam_epi16(a_r_p1_p1, a_i_p1_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_p1);
        square_a_64qam_epi16(a_r_p1_m1, a_i_p1_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_p1_m1);
        square_a_64qam_epi16(a_r_m1_p1, a_i_m1_p1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_p1);
        square_a_64qam_epi16(a_r_m1_m1, a_i_m1_m1, ch_mag_int, SQRT_42_OVER_FOUR, a_sq_m1_m1);

        // Computing Metrics
        xmm0 = _mm_subs_epi16(psi_a_p1_p1, a_sq_p1_p1);
        bit_met_p1_p1 = _mm_adds_epi16(xmm0, y0_p_1_1);
        
        xmm0 = _mm_subs_epi16(psi_a_p1_m1, a_sq_p1_m1);
        bit_met_p1_m1 = _mm_adds_epi16(xmm0, y0_m_1_1);
        
        xmm0 = _mm_subs_epi16(psi_a_m1_p1, a_sq_m1_p1);
        bit_met_m1_p1 = _mm_subs_epi16(xmm0, y0_m_1_1);
        
        xmm0 = _mm_subs_epi16(psi_a_m1_m1, a_sq_m1_m1);
        bit_met_m1_m1 = _mm_subs_epi16(xmm0, y0_p_1_1);
        
        // MSB
        logmax_num_re0 = _mm_max_epi16(bit_met_p1_p1,bit_met_p1_m1); // bit=0
        logmax_den_re0 = _mm_max_epi16(bit_met_m1_p1,bit_met_m1_m1); // bit=1
        
        y0r = _mm_subs_epi16(logmax_num_re0,logmax_den_re0);
        
        // LSB
        logmax_num_im0 = _mm_max_epi16(bit_met_p1_p1,bit_met_m1_p1); // bit=0
        logmax_den_im0 = _mm_max_epi16(bit_met_p1_m1,bit_met_m1_m1); // bit=1
        
        y0i = _mm_subs_epi16(logmax_num_im0,logmax_den_im0);
        
        stream0_128i_out[i] = _mm_unpacklo_epi16(y0r,y0i); // = [L1(1), L2(1), L1(2), L2(2)]
        if (i<((length>>1) - 1)) // false if only 2 REs remain
            stream0_128i_out[i+1] = _mm_unpackhi_epi16(y0r,y0i);
   
    }
    _mm_empty();
    _m_empty();
}


//----------------------------------------------------------------------------------------------
// 16-QAM
//----------------------------------------------------------------------------------------------

NOCYGWIN_STATIC __m128i ONE_OVER_TWO_SQRT_10 __attribute__((aligned(16)));
NOCYGWIN_STATIC __m128i NINE_OVER_TWO_SQRT_10 __attribute__((aligned(16)));

NOCYGWIN_STATIC __m128i  y0r_over_sqrt10 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0i_over_sqrt10 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0r_three_over_sqrt10 __attribute__ ((aligned(16)));
NOCYGWIN_STATIC __m128i  y0i_three_over_sqrt10 __attribute__ ((aligned(16)));

NOCYGWIN_STATIC __m128i ch_mag_des __attribute__((aligned(16))); 
NOCYGWIN_STATIC __m128i ch_mag_over_10 __attribute__ ((aligned(16))); 
NOCYGWIN_STATIC __m128i ch_mag_over_2 __attribute__ ((aligned(16))); 
NOCYGWIN_STATIC __m128i ch_mag_9_over_10 __attribute__ ((aligned(16))); 

void qam16_qpsk(short *stream0_in,
                short *stream1_in,
                short *ch_mag,
                short *stream0_out,
                short *rho01,
                int length
                ) {

/* 
   Author: Sebastian Wagner
   Date: 2012-06-04

   Input: 
   stream0_in:  MF filter for 1st stream, i.e., y0=h0'*y
   stream!_in:  MF filter for 2nd stream, i.e., y1=h1'*y
   ch_mag:      2*h0/sqrt(00), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc
   ch_mag_i:    2*h1/sqrt(00), [Re0 Im0 Re1 Im1] s.t. Im0=Re0, Im1=Re1, etc
   rho01:       Channel cross correlation, i.e., h1'*h0
   
   Output:
   stream0_out: output LLRs for 1st stream
*/

    __m128i *rho01_128i       = (__m128i *)rho01;
    __m128i *stream0_128i_in  = (__m128i *)stream0_in;
    __m128i *stream1_128i_in  = (__m128i *)stream1_in;
    __m128i *stream0_128i_out = (__m128i *)stream0_out;
    __m128i *ch_mag_128i      = (__m128i *)ch_mag;

    int i;

    ONE_OVER_SQRT_2 = _mm_set1_epi16(23170); // round(1/sqrt(2)*2^15)
    ONE_OVER_SQRT_10 = _mm_set1_epi16(20724); // round(1/sqrt(10)*2^16)
    THREE_OVER_SQRT_10 = _mm_set1_epi16(31086); // round(3/sqrt(10)*2^15)
    SQRT_10_OVER_FOUR = _mm_set1_epi16(25905); // round(sqrt(10)/4*2^15)
    ONE_OVER_TWO_SQRT_10 = _mm_set1_epi16(10362); // round(1/2/sqrt(10)*2^16)
    NINE_OVER_TWO_SQRT_10 = _mm_set1_epi16(23315); // round(9/2/sqrt(10)*2^14)

    for (i=0;i<length>>2;i+=2) {
        // In one iteration, we deal with 8 REs

        // Get rho
        xmm0 = rho01_128i[i];
        xmm1 = rho01_128i[i+1];
        xmm0 = _mm_shufflelo_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shufflehi_epi16(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm0 = _mm_shuffle_epi32(xmm0,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflelo_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shufflehi_epi16(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        xmm1 = _mm_shuffle_epi32(xmm1,0xd8); //_MM_SHUFFLE(0,2,1,3));
        //xmm0 = [Re(0,1) Re(2,3) Im(0,1) Im(2,3)]
        //xmm1 = [Re(4,5) Re(6,7) Im(4,5) Im(6,7)]
        xmm2 = _mm_unpacklo_epi64(xmm0,xmm1); // Re(rho)
        xmm3 = _mm_unpackhi_epi64(xmm0,xmm1); // Im(rho)
        rho_rpi = _mm_adds_epi16(xmm2,xmm3); // rho = Re(rho) + Im(rho)
        rho_rmi = _mm_subs_epi16(xmm2,xmm3); // rho* = Re(rho) - Im(rho)

        // Compute the different rhos
        rho_rpi_1_1 = _mm_mulhi_epi16(rho_rpi,ONE_OVER_SQRT_10);
        rho_rmi_1_1 = _mm_mulhi_epi16(rho_rmi,ONE_OVER_SQRT_10);
        rho_rpi_3_3 = _mm_mulhi_epi16(rho_rpi,THREE_OVER_SQRT_10);
        rho_rmi_3_3 = _mm_mulhi_epi16(rho_rmi,THREE_OVER_SQRT_10);
        rho_rpi_3_3 = _mm_slli_epi16(rho_rpi_3_3,1);
        rho_rmi_3_3 = _mm_slli_epi16(rho_rmi_3_3,1);

        xmm4 = _mm_mulhi_epi16(xmm2,ONE_OVER_SQRT_10); // Re(rho)
        xmm5 = _mm_mulhi_epi16(xmm3,THREE_OVER_SQRT_10); // Im(rho)
        xmm5 = _mm_slli_epi16(xmm5,1);

        rho_rpi_1_3 = _mm_adds_epi16(xmm4,xmm5);
        rho_rmi_1_3 = _mm_subs_epi16(xmm4,xmm5);

        xmm6 = _mm_mulhi_epi16(xmm2,THREE_OVER_SQRT_10); // Re(rho)
        xmm7 = _mm_mulhi_e