root / include / hpl_pauxil.h @ 9
Historique | Voir | Annoter | Télécharger (19,13 ko)
| 1 |
/*
|
|---|---|
| 2 |
* -- High Performance Computing Linpack Benchmark (HPL)
|
| 3 |
* HPL - 2.0 - September 10, 2008
|
| 4 |
* Antoine P. Petitet
|
| 5 |
* University of Tennessee, Knoxville
|
| 6 |
* Innovative Computing Laboratory
|
| 7 |
* (C) Copyright 2000-2008 All Rights Reserved
|
| 8 |
*
|
| 9 |
* -- Copyright notice and Licensing terms:
|
| 10 |
*
|
| 11 |
* Redistribution and use in source and binary forms, with or without
|
| 12 |
* modification, are permitted provided that the following conditions
|
| 13 |
* are met:
|
| 14 |
*
|
| 15 |
* 1. Redistributions of source code must retain the above copyright
|
| 16 |
* notice, this list of conditions and the following disclaimer.
|
| 17 |
*
|
| 18 |
* 2. Redistributions in binary form must reproduce the above copyright
|
| 19 |
* notice, this list of conditions, and the following disclaimer in the
|
| 20 |
* documentation and/or other materials provided with the distribution.
|
| 21 |
*
|
| 22 |
* 3. All advertising materials mentioning features or use of this
|
| 23 |
* software must display the following acknowledgement:
|
| 24 |
* This product includes software developed at the University of
|
| 25 |
* Tennessee, Knoxville, Innovative Computing Laboratory.
|
| 26 |
*
|
| 27 |
* 4. The name of the University, the name of the Laboratory, or the
|
| 28 |
* names of its contributors may not be used to endorse or promote
|
| 29 |
* products derived from this software without specific written
|
| 30 |
* permission.
|
| 31 |
*
|
| 32 |
* -- Disclaimer:
|
| 33 |
*
|
| 34 |
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
| 35 |
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
| 36 |
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
|
| 37 |
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE UNIVERSITY
|
| 38 |
* OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
| 39 |
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
| 40 |
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
| 41 |
* DATA OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
| 42 |
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
| 43 |
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
| 44 |
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
| 45 |
*/
|
| 46 |
#ifndef HPL_PAUXIL_H
|
| 47 |
#define HPL_PAUXIL_H
|
| 48 |
/*
|
| 49 |
* ---------------------------------------------------------------------
|
| 50 |
* Include files
|
| 51 |
* ---------------------------------------------------------------------
|
| 52 |
*/
|
| 53 |
#include "hpl_misc.h" |
| 54 |
#include "hpl_blas.h" |
| 55 |
#include "hpl_auxil.h" |
| 56 |
|
| 57 |
#include "hpl_pmisc.h" |
| 58 |
#include "hpl_grid.h" |
| 59 |
/*
|
| 60 |
* ---------------------------------------------------------------------
|
| 61 |
* #define macros definitions
|
| 62 |
* ---------------------------------------------------------------------
|
| 63 |
*/
|
| 64 |
/*
|
| 65 |
* Mindxg2p returns the process coodinate owning the entry globally in-
|
| 66 |
* dexed by ig_.
|
| 67 |
*/
|
| 68 |
#define Mindxg2p( ig_, inb_, nb_, proc_, src_, nprocs_ ) \
|
| 69 |
{ \
|
| 70 |
if( ( (ig_) >= (inb_) ) && ( (src_) >= 0 ) && \ |
| 71 |
( (nprocs_) > 1 ) ) \
|
| 72 |
{ \
|
| 73 |
proc_ = (src_) + 1 + ( (ig_)-(inb_) ) / (nb_); \
|
| 74 |
proc_ -= ( proc_ / (nprocs_) ) * (nprocs_); \ |
| 75 |
} \ |
| 76 |
else \
|
| 77 |
{ \
|
| 78 |
proc_ = (src_); \ |
| 79 |
} \ |
| 80 |
} |
| 81 |
|
| 82 |
#define Mindxg2l( il_, ig_, inb_, nb_, proc_, src_, nprocs_ ) \
|
| 83 |
{ \
|
| 84 |
if( ( (ig_) < (inb_) ) || ( (src_) == -1 ) || \ |
| 85 |
( (nprocs_) == 1 ) ) { il_ = (ig_); } \
|
| 86 |
else \
|
| 87 |
{ \
|
| 88 |
int i__, j__; \
|
| 89 |
j__ = ( i__ = ( (ig_)-(inb_) ) / (nb_) ) / (nprocs_); \ |
| 90 |
il_ = (nb_)*( j__ - i__ ) + \ |
| 91 |
( (i__ + 1 - ( j__ + 1 ) * (nprocs_) ) ? \ |
| 92 |
(ig_) - (inb_) : (ig_) ); \ |
| 93 |
} \ |
| 94 |
} |
| 95 |
|
| 96 |
#define Mindxg2lp( il_, proc_, ig_, inb_, nb_, src_, nprocs_ ) \
|
| 97 |
{ \
|
| 98 |
if( ( (ig_) < (inb_) ) || ( (src_) == -1 ) || \ |
| 99 |
( (nprocs_) == 1 ) ) \
|
| 100 |
{ il_ = (ig_); proc_ = (src_); } \
|
| 101 |
else \
|
| 102 |
{ \
|
| 103 |
int i__, j__; \
|
| 104 |
j__ = ( i__ = ( (ig_)-(inb_) ) / (nb_) ) / (nprocs_); \ |
| 105 |
il_ = (nb_)*(j__-i__) + \ |
| 106 |
( ( i__ + 1 - ( j__ + 1 ) * (nprocs_) ) ? \ |
| 107 |
(ig_) - (inb_) : (ig_) ); \ |
| 108 |
proc_ = (src_) + 1 + i__; \
|
| 109 |
proc_ -= ( proc_ / (nprocs_) ) * (nprocs_); \ |
| 110 |
} \ |
| 111 |
} |
| 112 |
/*
|
| 113 |
* Mindxl2g computes the global index ig_ corresponding to the local
|
| 114 |
* index il_ in process proc_.
|
| 115 |
*/
|
| 116 |
#define Mindxl2g( ig_, il_, inb_, nb_, proc_, src_, nprocs_ ) \
|
| 117 |
{ \
|
| 118 |
if( ( (src_) >= 0 ) && ( (nprocs_) > 1 ) ) \ |
| 119 |
{ \
|
| 120 |
if( (proc_) == (src_) ) \
|
| 121 |
{ \
|
| 122 |
if( (il_) < (inb_) ) ig_ = (il_); \
|
| 123 |
else ig_ = (il_) + \
|
| 124 |
(nb_)*((nprocs_)-1)*(((il_)-(inb_))/(nb_) + 1); \ |
| 125 |
} \ |
| 126 |
else if( (proc_) < (src_) ) \ |
| 127 |
{ \
|
| 128 |
ig_ = (il_) + (inb_) + \ |
| 129 |
(nb_)*( ((nprocs_)-1)*((il_)/(nb_)) + \
|
| 130 |
(proc_)-(src_)-1+(nprocs_) ); \
|
| 131 |
} \ |
| 132 |
else \
|
| 133 |
{ \
|
| 134 |
ig_ = (il_) + (inb_) + \ |
| 135 |
(nb_)*( ((nprocs_)-1)*((il_)/(nb_)) + \
|
| 136 |
(proc_)-(src_)-1 ); \
|
| 137 |
} \ |
| 138 |
} \ |
| 139 |
else \
|
| 140 |
{ \
|
| 141 |
ig_ = (il_); \ |
| 142 |
} \ |
| 143 |
} |
| 144 |
/*
|
| 145 |
* MnumrocI computes the # of local indexes np_ residing in the process
|
| 146 |
* of coordinate proc_ corresponding to the interval of global indexes
|
| 147 |
* i_:i_+n_-1 assuming that the global index 0 resides in the process
|
| 148 |
* src_, and that the indexes are distributed from src_ using the para-
|
| 149 |
* meters inb_, nb_ and nprocs_.
|
| 150 |
*/
|
| 151 |
#define MnumrocI( np_, n_, i_, inb_, nb_, proc_, src_, nprocs_ ) \
|
| 152 |
{ \
|
| 153 |
if( ( (src_) >= 0 ) && ( (nprocs_) > 1 ) ) \ |
| 154 |
{ \
|
| 155 |
int inb__, mydist__, n__, nblk__, quot__, src__; \
|
| 156 |
if( ( inb__ = (inb_) - (i_) ) <= 0 ) \ |
| 157 |
{ \
|
| 158 |
nblk__ = (-inb__) / (nb_) + 1; \
|
| 159 |
src__ = (src_) + nblk__; \ |
| 160 |
src__ -= ( src__ / (nprocs_) ) * (nprocs_); \ |
| 161 |
inb__ += nblk__*(nb_); \ |
| 162 |
if( ( n__ = (n_) - inb__ ) <= 0 ) \ |
| 163 |
{ \
|
| 164 |
if( (proc_) == src__ ) np_ = (n_); \
|
| 165 |
else np_ = 0; \ |
| 166 |
} \ |
| 167 |
else \
|
| 168 |
{ \
|
| 169 |
if( ( mydist__ = (proc_) - src__ ) < 0 ) \ |
| 170 |
mydist__ += (nprocs_); \ |
| 171 |
nblk__ = n__ / (nb_) + 1; \
|
| 172 |
mydist__ -= nblk__ - \ |
| 173 |
(quot__ = (nblk__ / (nprocs_))) * (nprocs_); \ |
| 174 |
if( mydist__ < 0 ) \ |
| 175 |
{ \
|
| 176 |
if( (proc_) != src__ ) \
|
| 177 |
np_ = (nb_) + (nb_) * quot__; \ |
| 178 |
else \
|
| 179 |
np_ = inb__ + (nb_) * quot__; \ |
| 180 |
} \ |
| 181 |
else if( mydist__ > 0 ) \ |
| 182 |
{ \
|
| 183 |
np_ = (nb_) * quot__; \ |
| 184 |
} \ |
| 185 |
else \
|
| 186 |
{ \
|
| 187 |
if( (proc_) != src__ ) \
|
| 188 |
np_ = n__ +(nb_)+(nb_)*(quot__ - nblk__); \ |
| 189 |
else \
|
| 190 |
np_ = (n_)+ (nb_)*(quot__ - nblk__); \ |
| 191 |
} \ |
| 192 |
} \ |
| 193 |
} \ |
| 194 |
else \
|
| 195 |
{ \
|
| 196 |
if( ( n__ = (n_) - inb__ ) <= 0 ) \ |
| 197 |
{ \
|
| 198 |
if( (proc_) == (src_) ) np_ = (n_); \
|
| 199 |
else np_ = 0; \ |
| 200 |
} \ |
| 201 |
else \
|
| 202 |
{ \
|
| 203 |
if( ( mydist__ = (proc_) - (src_) ) < 0 ) \ |
| 204 |
mydist__ += (nprocs_); \ |
| 205 |
nblk__ = n__ / (nb_) + 1; \
|
| 206 |
mydist__ -= nblk__ - \ |
| 207 |
( quot__ = (nblk__ / (nprocs_)) )*(nprocs_); \ |
| 208 |
if( mydist__ < 0 ) \ |
| 209 |
{ \
|
| 210 |
if( (proc_) != (src_) ) \
|
| 211 |
np_ = (nb_) + (nb_) * quot__; \ |
| 212 |
else \
|
| 213 |
np_ = inb__ + (nb_) * quot__; \ |
| 214 |
} \ |
| 215 |
else if( mydist__ > 0 ) \ |
| 216 |
{ \
|
| 217 |
np_ = (nb_) * quot__; \ |
| 218 |
} \ |
| 219 |
else \
|
| 220 |
{ \
|
| 221 |
if( (proc_) != (src_) ) \
|
| 222 |
np_ = n__ +(nb_)+(nb_)*(quot__ - nblk__); \ |
| 223 |
else \
|
| 224 |
np_ = (n_)+ (nb_)*(quot__ - nblk__); \ |
| 225 |
} \ |
| 226 |
} \ |
| 227 |
} \ |
| 228 |
} \ |
| 229 |
else \
|
| 230 |
{ \
|
| 231 |
np_ = (n_); \ |
| 232 |
} \ |
| 233 |
} |
| 234 |
|
| 235 |
#define Mnumroc( np_, n_, inb_, nb_, proc_, src_, nprocs_ ) \
|
| 236 |
MnumrocI( np_, n_, 0, inb_, nb_, proc_, src_, nprocs_ )
|
| 237 |
/*
|
| 238 |
* ---------------------------------------------------------------------
|
| 239 |
* Function prototypes
|
| 240 |
* ---------------------------------------------------------------------
|
| 241 |
*/
|
| 242 |
void HPL_indxg2lp
|
| 243 |
STDC_ARGS( ( |
| 244 |
int *,
|
| 245 |
int *,
|
| 246 |
const int, |
| 247 |
const int, |
| 248 |
const int, |
| 249 |
const int, |
| 250 |
const int |
| 251 |
) ); |
| 252 |
int HPL_indxg2l
|
| 253 |
STDC_ARGS( ( |
| 254 |
const int, |
| 255 |
const int, |
| 256 |
const int, |
| 257 |
const int, |
| 258 |
const int |
| 259 |
) ); |
| 260 |
int HPL_indxg2p
|
| 261 |
STDC_ARGS( ( |
| 262 |
const int, |
| 263 |
const int, |
| 264 |
const int, |
| 265 |
const int, |
| 266 |
const int |
| 267 |
) ); |
| 268 |
int HPL_indxl2g
|
| 269 |
STDC_ARGS( ( |
| 270 |
const int, |
| 271 |
const int, |
| 272 |
const int, |
| 273 |
const int, |
| 274 |
const int, |
| 275 |
const int |
| 276 |
) ); |
| 277 |
void HPL_infog2l
|
| 278 |
STDC_ARGS( ( |
| 279 |
int,
|
| 280 |
int,
|
| 281 |
const int, |
| 282 |
const int, |
| 283 |
const int, |
| 284 |
const int, |
| 285 |
const int, |
| 286 |
const int, |
| 287 |
const int, |
| 288 |
const int, |
| 289 |
const int, |
| 290 |
const int, |
| 291 |
int *,
|
| 292 |
int *,
|
| 293 |
int *,
|
| 294 |
int *
|
| 295 |
) ); |
| 296 |
int HPL_numroc
|
| 297 |
STDC_ARGS( ( |
| 298 |
const int, |
| 299 |
const int, |
| 300 |
const int, |
| 301 |
const int, |
| 302 |
const int, |
| 303 |
const int |
| 304 |
) ); |
| 305 |
int HPL_numrocI
|
| 306 |
STDC_ARGS( ( |
| 307 |
const int, |
| 308 |
const int, |
| 309 |
const int, |
| 310 |
const int, |
| 311 |
const int, |
| 312 |
const int, |
| 313 |
const int |
| 314 |
) ); |
| 315 |
|
| 316 |
void HPL_dlaswp00N
|
| 317 |
STDC_ARGS( ( |
| 318 |
const int, |
| 319 |
const int, |
| 320 |
double *,
|
| 321 |
const int, |
| 322 |
const int * |
| 323 |
) ); |
| 324 |
void HPL_dlaswp10N
|
| 325 |
STDC_ARGS( ( |
| 326 |
const int, |
| 327 |
const int, |
| 328 |
double *,
|
| 329 |
const int, |
| 330 |
const int * |
| 331 |
) ); |
| 332 |
void HPL_dlaswp01N
|
| 333 |
STDC_ARGS( ( |
| 334 |
const int, |
| 335 |
const int, |
| 336 |
double *,
|
| 337 |
const int, |
| 338 |
double *,
|
| 339 |
const int, |
| 340 |
const int *, |
| 341 |
const int * |
| 342 |
) ); |
| 343 |
void HPL_dlaswp01T
|
| 344 |
STDC_ARGS( ( |
| 345 |
const int, |
| 346 |
const int, |
| 347 |
double *,
|
| 348 |
const int, |
| 349 |
double *,
|
| 350 |
const int, |
| 351 |
const int *, |
| 352 |
const int * |
| 353 |
) ); |
| 354 |
void HPL_dlaswp02N
|
| 355 |
STDC_ARGS( ( |
| 356 |
const int, |
| 357 |
const int, |
| 358 |
const double *, |
| 359 |
const int, |
| 360 |
double *,
|
| 361 |
double *,
|
| 362 |
const int, |
| 363 |
const int *, |
| 364 |
const int * |
| 365 |
) ); |
| 366 |
void HPL_dlaswp03N
|
| 367 |
STDC_ARGS( ( |
| 368 |
const int, |
| 369 |
const int, |
| 370 |
double *,
|
| 371 |
const int, |
| 372 |
const double *, |
| 373 |
const double *, |
| 374 |
const int |
| 375 |
) ); |
| 376 |
void HPL_dlaswp03T
|
| 377 |
STDC_ARGS( ( |
| 378 |
const int, |
| 379 |
const int, |
| 380 |
double *,
|
| 381 |
const int, |
| 382 |
const double *, |
| 383 |
const double *, |
| 384 |
const int |
| 385 |
) ); |
| 386 |
void HPL_dlaswp04N
|
| 387 |
STDC_ARGS( ( |
| 388 |
const int, |
| 389 |
const int, |
| 390 |
const int, |
| 391 |
double *,
|
| 392 |
const int, |
| 393 |
double *,
|
| 394 |
const int, |
| 395 |
const double *, |
| 396 |
const double *, |
| 397 |
const int, |
| 398 |
const int *, |
| 399 |
const int * |
| 400 |
) ); |
| 401 |
void HPL_dlaswp04T
|
| 402 |
STDC_ARGS( ( |
| 403 |
const int, |
| 404 |
const int, |
| 405 |
const int, |
| 406 |
double *,
|
| 407 |
const int, |
| 408 |
double *,
|
| 409 |
const int, |
| 410 |
const double *, |
| 411 |
const double *, |
| 412 |
const int, |
| 413 |
const int *, |
| 414 |
const int * |
| 415 |
) ); |
| 416 |
void HPL_dlaswp05N
|
| 417 |
STDC_ARGS( ( |
| 418 |
const int, |
| 419 |
const int, |
| 420 |
double *,
|
| 421 |
const int, |
| 422 |
const double *, |
| 423 |
const int, |
| 424 |
const int *, |
| 425 |
const int * |
| 426 |
) ); |
| 427 |
void HPL_dlaswp05T
|
| 428 |
STDC_ARGS( ( |
| 429 |
const int, |
| 430 |
const int, |
| 431 |
double *,
|
| 432 |
const int, |
| 433 |
const double *, |
| 434 |
const int, |
| 435 |
const int *, |
| 436 |
const int * |
| 437 |
) ); |
| 438 |
void HPL_dlaswp06N
|
| 439 |
STDC_ARGS( ( |
| 440 |
const int, |
| 441 |
const int, |
| 442 |
double *,
|
| 443 |
const int, |
| 444 |
double *,
|
| 445 |
const int, |
| 446 |
const int * |
| 447 |
) ); |
| 448 |
void HPL_dlaswp06T
|
| 449 |
STDC_ARGS( ( |
| 450 |
const int, |
| 451 |
const int, |
| 452 |
double *,
|
| 453 |
const int, |
| 454 |
double *,
|
| 455 |
const int, |
| 456 |
const int * |
| 457 |
) ); |
| 458 |
|
| 459 |
void HPL_pabort
|
| 460 |
STDC_ARGS( ( |
| 461 |
int,
|
| 462 |
const char *, |
| 463 |
const char *, |
| 464 |
... |
| 465 |
) ); |
| 466 |
void HPL_pwarn
|
| 467 |
STDC_ARGS( ( |
| 468 |
FILE *, |
| 469 |
int,
|
| 470 |
const char *, |
| 471 |
const char *, |
| 472 |
... |
| 473 |
) ); |
| 474 |
void HPL_pdlaprnt
|
| 475 |
STDC_ARGS( ( |
| 476 |
const HPL_T_grid *,
|
| 477 |
const int, |
| 478 |
const int, |
| 479 |
const int, |
| 480 |
double *,
|
| 481 |
const int, |
| 482 |
const int, |
| 483 |
const int, |
| 484 |
const char * |
| 485 |
) ); |
| 486 |
double HPL_pdlamch
|
| 487 |
STDC_ARGS( ( |
| 488 |
MPI_Comm, |
| 489 |
const HPL_T_MACH
|
| 490 |
) ); |
| 491 |
double HPL_pdlange
|
| 492 |
STDC_ARGS( ( |
| 493 |
const HPL_T_grid *,
|
| 494 |
const HPL_T_NORM,
|
| 495 |
const int, |
| 496 |
const int, |
| 497 |
const int, |
| 498 |
const double *, |
| 499 |
const int |
| 500 |
) ); |
| 501 |
|
| 502 |
#endif
|
| 503 |
/*
|
| 504 |
* End of hpl_pauxil.h
|
| 505 |
*/
|