]>
Commit | Line | Data |
---|---|---|
3721b9e1 DF |
1 | \input texinfo @c -*-texinfo-*- |
2 | ||
3 | @c %**start of header | |
4 | @setfilename libgomp.info | |
5 | @settitle GNU libgomp | |
6 | @c %**end of header | |
7 | ||
8 | ||
9 | @copying | |
10 | Copyright @copyright{} 2006 Free Software Foundation, Inc. | |
11 | ||
12 | Permission is granted to copy, distribute and/or modify this document | |
13 | under the terms of the GNU Free Documentation License, Version 1.1 or | |
14 | any later version published by the Free Software Foundation; with the | |
15 | Invariant Sections being ``GNU General Public License'' and ``Funding | |
16 | Free Software'', the Front-Cover | |
17 | texts being (a) (see below), and with the Back-Cover Texts being (b) | |
18 | (see below). A copy of the license is included in the section entitled | |
19 | ``GNU Free Documentation License''. | |
20 | ||
21 | (a) The FSF's Front-Cover Text is: | |
22 | ||
23 | A GNU Manual | |
24 | ||
25 | (b) The FSF's Back-Cover Text is: | |
26 | ||
27 | You have freedom to copy and modify this GNU Manual, like GNU | |
28 | software. Copies published by the Free Software Foundation raise | |
29 | funds for GNU development. | |
30 | @end copying | |
31 | ||
32 | @ifinfo | |
33 | @dircategory GNU Libraries | |
34 | @direntry | |
35 | * libgomp: (libgomp). GNU OpenMP runtime library | |
36 | @end direntry | |
37 | ||
38 | This manual documents the GNU implementation of the OpenMP API for | |
39 | multi-platform shared-memory parallel programming in C/C++ and Fortran. | |
40 | ||
41 | Published by the Free Software Foundation | |
42 | 51 Franklin Street, Fifth Floor | |
43 | Boston, MA 02110-1301 USA | |
44 | ||
45 | @insertcopying | |
46 | @end ifinfo | |
47 | ||
48 | ||
49 | @setchapternewpage odd | |
50 | ||
51 | @titlepage | |
52 | @title The GNU OpenMP Implementation | |
53 | @page | |
54 | @vskip 0pt plus 1filll | |
55 | @comment For the @value{version-GCC} Version* | |
56 | @sp 1 | |
57 | Published by the Free Software Foundation @* | |
58 | 51 Franklin Street, Fifth Floor@* | |
59 | Boston, MA 02110-1301, USA@* | |
60 | @sp 1 | |
61 | @insertcopying | |
62 | @end titlepage | |
63 | ||
64 | @summarycontents | |
65 | @contents | |
66 | @page | |
67 | ||
68 | ||
69 | @node Top | |
70 | @top Introduction | |
71 | @cindex Introduction | |
72 | ||
73 | This manual documents the usage of libgomp, the GNU implementation of the | |
74 | @uref{http://www.openmp.org, OpenMP} Application Programming Interface (API) | |
75 | for multi-platform shared-memory parallel programming in C/C++ and Fortran. | |
76 | ||
77 | ||
78 | ||
79 | @comment | |
80 | @comment When you add a new menu item, please keep the right hand | |
81 | @comment aligned to the same column. Do not use tabs. This provides | |
82 | @comment better formatting. | |
83 | @comment | |
84 | @menu | |
85 | * Enabling OpenMP:: How to enable OpenMP for your applications. | |
86 | * Runtime Library Routines:: The OpenMP runtime application programming | |
87 | interface. | |
88 | * Environment Variables:: Influencing runtime behavior with environment | |
89 | variables. | |
90 | * The libgomp ABI:: Notes on the external ABI presented by libgomp. | |
91 | * Reporting Bugs:: How to report bugs in GNU OpenMP. | |
92 | * Copying:: GNU general public license says | |
93 | how you can copy and share libgomp. | |
94 | * GNU Free Documentation License:: | |
95 | How you can copy and share this manual. | |
96 | * Funding:: How to help assure continued work for free | |
97 | software. | |
98 | * Index:: Index of this documentation. | |
99 | @end menu | |
100 | ||
101 | ||
102 | @c --------------------------------------------------------------------- | |
103 | @c Enabling OpenMP | |
104 | @c --------------------------------------------------------------------- | |
105 | ||
106 | @node Enabling OpenMP | |
107 | @chapter Enabling OpenMP | |
108 | ||
109 | To activate the OpenMP extensions for C/C++ and Fortran, the compile-time | |
110 | flag @command{-fopenmp} must be specified. This enables the OpenMP directive | |
111 | @code{#pragma omp} in C/C++ and @code{!$omp} directives in free form, | |
112 | @code{c$omp}, @code{*$omp} and @code{!$omp} directives in fixed form, | |
113 | @code{!$} conditional compilation sentinels in free form and @code{c$}, | |
114 | @code{*$} and @code{!$} sentinels in fixed form, for Fortran. The flag also | |
115 | arranges for automatic linking of the OpenMP runtime library | |
116 | (@ref{Runtime Library Routines}). | |
117 | ||
118 | A complete description of all OpenMP directives accepted may be found in | |
119 | the @uref{http://www.openmp.org, OpenMP Application Program Interface} manual, | |
120 | version 2.5. | |
121 | ||
122 | ||
123 | @c --------------------------------------------------------------------- | |
124 | @c Runtime Library Routines | |
125 | @c --------------------------------------------------------------------- | |
126 | ||
127 | @node Runtime Library Routines | |
128 | @chapter Runtime Library Routines | |
129 | ||
130 | The runtime routines described here are defined by section 3 of the OpenMP | |
131 | specifications in version 2.5. | |
132 | ||
133 | Control threads, processors and the parallel environment. | |
134 | ||
135 | @menu | |
136 | * omp_get_dynamic:: Dynamic teams setting | |
137 | * omp_get_max_threads:: Maximum number of threads | |
138 | * omp_get_nested:: Nested parallel regions | |
139 | * omp_get_num_procs:: Number of processors online | |
140 | * omp_get_num_threads:: Size of the active team | |
141 | * omp_get_thread_num:: Current thread ID | |
142 | * omp_in_parallel:: Whether a parallel region is active | |
143 | * omp_set_dynamic:: Enable/disable dynamic teams | |
144 | * omp_set_nested:: Enable/disable nested parallel regions | |
145 | * omp_set_num_threads:: Set upper team size limit | |
146 | @end menu | |
147 | ||
148 | Initialize, set, test, unset and destroy simple and nested locks. | |
149 | ||
150 | @menu | |
151 | * omp_init_lock:: Initialize simple lock | |
152 | * omp_set_lock:: Wait for and set simple lock | |
153 | * omp_test_lock:: Test and set simple lock if available | |
154 | * omp_unset_lock:: Unset simple lock | |
155 | * omp_destroy_lock:: Destroy simple lock | |
156 | * omp_init_nest_lock:: Initialize nested lock | |
157 | * omp_set_nest_lock:: Wait for and set simple lock | |
158 | * omp_test_nest_lock:: Test and set nested lock if available | |
159 | * omp_unset_nest_lock:: Unset nested lock | |
160 | * omp_destroy_nest_lock:: Destroy nested lock | |
161 | @end menu | |
162 | ||
163 | Portable, thread-based, wall clock timer. | |
164 | ||
165 | @menu | |
166 | * omp_get_wtick:: Get timer precision. | |
167 | * omp_get_wtime:: Elapsed wall clock time. | |
168 | @end menu | |
169 | ||
170 | @node omp_get_dynamic | |
171 | @section @code{omp_get_dynamic} -- Dynamic teams setting | |
172 | @table @asis | |
173 | @item @emph{Description}: | |
174 | This function returns @code{true} if enabled, @code{false} otherwise. | |
175 | Here, @code{true} and @code{false} represent their language-specific | |
176 | counterparts. | |
177 | ||
178 | @item @emph{C/C++}: | |
179 | @multitable @columnfractions .20 .80 | |
180 | @item @emph{Prototype}: @tab @code{int omp_get_dynamic();} | |
181 | @end multitable | |
182 | ||
183 | @item @emph{Fortran}: | |
184 | @multitable @columnfractions .20 .80 | |
185 | @item @emph{Interface}: @tab @code{logical function omp_get_dynamic()} | |
186 | @end multitable | |
187 | ||
188 | @item @emph{See also}: | |
189 | @ref{omp_set_dynamic} | |
190 | ||
191 | @item @emph{Reference}: | |
192 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.2.8. | |
193 | @end table | |
194 | ||
195 | ||
196 | ||
197 | @node omp_get_max_threads | |
198 | @section @code{omp_get_max_threads} -- Maximum number of threads | |
199 | @table @asis | |
200 | @item @emph{Description}: | |
201 | Return the maximum number of threads used for parallel regions that do | |
202 | not use the clause @code{num_threads}. | |
203 | ||
204 | @item @emph{C/C++}: | |
205 | @multitable @columnfractions .20 .80 | |
206 | @item @emph{Prototype}: @tab @code{int omp_get_max_threads();} | |
207 | @end multitable | |
208 | ||
209 | @item @emph{Fortran}: | |
210 | @multitable @columnfractions .20 .80 | |
211 | @item @emph{Interface}: @tab @code{integer function omp_get_max_threads()} | |
212 | @end multitable | |
213 | ||
214 | @item @emph{See also}: | |
215 | @ref{omp_set_num_threads}, @ref{omp_set_dynamic} | |
216 | ||
217 | @item @emph{Reference}: | |
218 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.2.3. | |
219 | @end table | |
220 | ||
221 | ||
222 | ||
223 | @node omp_get_nested | |
224 | @section @code{omp_get_nested} -- Nested parallel regions | |
225 | @table @asis | |
226 | @item @emph{Description}: | |
227 | This function returns @code{true} if nested parallel regions are | |
228 | enabled, @code{false} otherwise. Here, @code{true} and @code{false} | |
229 | represent their language-specific counterparts. | |
230 | ||
231 | @item @emph{C/C++}: | |
232 | @multitable @columnfractions .20 .80 | |
233 | @item @emph{Prototype}: @tab @code{int omp_get_nested();} | |
234 | @end multitable | |
235 | ||
236 | @item @emph{Fortran}: | |
237 | @multitable @columnfractions .20 .80 | |
238 | @item @emph{Interface}: @tab @code{integer function omp_get_nested()} | |
239 | @end multitable | |
240 | ||
241 | @item @emph{See also}: | |
242 | @ref{omp_set_nested} | |
243 | ||
244 | @item @emph{Reference}: | |
245 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.2.10. | |
246 | @end table | |
247 | ||
248 | ||
249 | ||
250 | @node omp_get_num_procs | |
251 | @section @code{omp_get_num_procs} -- Number of processors online | |
252 | @table @asis | |
253 | @item @emph{Description}: | |
254 | Returns the number of processors online. | |
255 | ||
256 | @item @emph{C/C++}: | |
257 | @multitable @columnfractions .20 .80 | |
258 | @item @emph{Prototype}: @tab @code{int omp_get_num_procs();} | |
259 | @end multitable | |
260 | ||
261 | @item @emph{Fortran}: | |
262 | @multitable @columnfractions .20 .80 | |
263 | @item @emph{Interface}: @tab @code{integer function omp_get_num_procs()} | |
264 | @end multitable | |
265 | ||
266 | @item @emph{Reference}: | |
267 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.2.5. | |
268 | @end table | |
269 | ||
270 | ||
271 | ||
272 | @node omp_get_num_threads | |
273 | @section @code{omp_get_num_threads} -- Size of the active team | |
274 | @table @asis | |
275 | @item @emph{Description}: | |
276 | The number of threads in the current team. In a sequential section of | |
277 | the program @code{omp_get_num_threads} returns 1. | |
278 | ||
279 | @item @emph{C/C++}: | |
280 | @multitable @columnfractions .20 .80 | |
281 | @item @emph{Prototype}: @tab @code{int omp_get_num_threads();} | |
282 | @end multitable | |
283 | ||
284 | @item @emph{Fortran}: | |
285 | @multitable @columnfractions .20 .80 | |
286 | @item @emph{Interface}: @tab @code{integer function omp_get_num_threads()} | |
287 | @end multitable | |
288 | ||
289 | @item @emph{See also}: | |
290 | @ref{omp_get_max_threads}, @ref{omp_set_num_threads}, @ref{OMP_NUM_THREADS} | |
291 | ||
292 | @item @emph{Reference}: | |
293 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.2.2. | |
294 | @end table | |
295 | ||
296 | ||
297 | ||
298 | @node omp_get_thread_num | |
299 | @section @code{omp_get_thread_num} -- Current thread ID | |
300 | @table @asis | |
301 | @item @emph{Description}: | |
302 | Unique thread identification number. In a sequential parts of the program, | |
303 | @code{omp_get_thread_num} always returns 0. In parallel regions the return | |
304 | value varies from 0 to @code{omp_get_max_threads}-1 inclusive. The return | |
305 | value of the master thread of a team is always 0. | |
306 | ||
307 | @item @emph{C/C++}: | |
308 | @multitable @columnfractions .20 .80 | |
309 | @item @emph{Prototype}: @tab @code{int omp_get_thread_num();} | |
310 | @end multitable | |
311 | ||
312 | @item @emph{Fortran}: | |
313 | @multitable @columnfractions .20 .80 | |
314 | @item @emph{Interface}: @tab @code{integer function omp_get_thread_num()} | |
315 | @end multitable | |
316 | ||
317 | @item @emph{See also}: | |
318 | @ref{omp_get_max_threads} | |
319 | ||
320 | @item @emph{Reference}: | |
321 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.2.4. | |
322 | @end table | |
323 | ||
324 | ||
325 | ||
326 | @node omp_in_parallel | |
327 | @section @code{omp_in_parallel} -- Whether a parallel region is active | |
328 | @table @asis | |
329 | @item @emph{Description}: | |
330 | This function returns @code{true} if currently running in parallel, | |
331 | @code{false} otherwise. Here, @code{true} and @code{false} represent | |
332 | their language-specific counterparts. | |
333 | ||
334 | @item @emph{C/C++}: | |
335 | @multitable @columnfractions .20 .80 | |
336 | @item @emph{Prototype}: @tab @code{int omp_in_parallel();} | |
337 | @end multitable | |
338 | ||
339 | @item @emph{Fortran}: | |
340 | @multitable @columnfractions .20 .80 | |
341 | @item @emph{Interface}: @tab @code{logical function omp_in_parallel()} | |
342 | @end multitable | |
343 | ||
344 | @item @emph{Reference}: | |
345 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.2.6. | |
346 | @end table | |
347 | ||
348 | ||
349 | @node omp_set_dynamic | |
350 | @section @code{omp_set_dynamic} -- Enable/disable dynamic teams | |
351 | @table @asis | |
352 | @item @emph{Description}: | |
353 | Enable or disable the dynamic adjustment of the number of threads | |
354 | within a team. The function takes the language-specific equivalent | |
355 | of @code{true} and @code{false}, where @code{true} enables dynamic | |
356 | adjustment of team sizes and @code{false} disables it. | |
357 | ||
358 | @item @emph{C/C++}: | |
359 | @multitable @columnfractions .20 .80 | |
360 | @item @emph{Prototype}: @tab @code{void omp_set_dynamic(int);} | |
361 | @end multitable | |
362 | ||
363 | @item @emph{Fortran}: | |
364 | @multitable @columnfractions .20 .80 | |
365 | @item @emph{Interface}: @tab @code{subroutine omp_set_dynamic(set)} | |
366 | @item @tab @code{integer, intent(in) :: set} | |
367 | @end multitable | |
368 | ||
369 | @item @emph{See also}: | |
370 | @ref{OMP_DYNAMIC}, @ref{omp_get_dynamic} | |
371 | ||
372 | @item @emph{Reference}: | |
373 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.2.7. | |
374 | @end table | |
375 | ||
376 | ||
377 | ||
378 | @node omp_set_nested | |
379 | @section @code{omp_set_nested} -- Enable/disable nested parallel regions | |
380 | @table @asis | |
381 | @item @emph{Description}: | |
382 | Enable or disable nested parallel regions, i. e. whether team members | |
383 | are allowed to create new teams. The function takes the language-specific | |
384 | equivalent of @code{true} and @code{false}, where @code{true} enables | |
385 | dynamic adjustment of team sizes and @code{false} disables it. | |
386 | ||
387 | @item @emph{C/C++}: | |
388 | @multitable @columnfractions .20 .80 | |
389 | @item @emph{Prototype}: @tab @code{void omp_set_dynamic(int);} | |
390 | @end multitable | |
391 | ||
392 | @item @emph{Fortran}: | |
393 | @multitable @columnfractions .20 .80 | |
394 | @item @emph{Interface}: @tab @code{subroutine omp_set_dynamic(set)} | |
395 | @item @tab @code{integer, intent(in) :: set} | |
396 | @end multitable | |
397 | ||
398 | @item @emph{See also}: | |
399 | @ref{OMP_NESTED}, @ref{omp_get_nested} | |
400 | ||
401 | @item @emph{Reference}: | |
402 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.2.9. | |
403 | @end table | |
404 | ||
405 | ||
406 | ||
407 | @node omp_set_num_threads | |
408 | @section @code{omp_set_num_threads} -- Set upper team size limit | |
409 | @table @asis | |
410 | @item @emph{Description}: | |
411 | Specifies the number of threads used by default in subsequent parallel | |
412 | sections, if those do not specify a @code{num_threads} clause. The | |
413 | argument of @code{omp_set_num_threads} shall be a positive integer. | |
414 | ||
415 | If the argument is negative integer or zero, the application will crash or | |
416 | stop, respectively. An enhancement request was filed, | |
417 | @uref{http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29949, PR29949}. | |
418 | ||
419 | @item @emph{C/C++}: | |
420 | @multitable @columnfractions .20 .80 | |
421 | @item @emph{Prototype}: @tab @code{void omp_set_num_threads(int);} | |
422 | @end multitable | |
423 | ||
424 | @item @emph{Fortran}: | |
425 | @multitable @columnfractions .20 .80 | |
426 | @item @emph{Interface}: @tab @code{subroutine omp_set_num_threads(set)} | |
427 | @item @tab @code{integer, intent(in) :: set} | |
428 | @end multitable | |
429 | ||
430 | @item @emph{See also}: | |
431 | @ref{OMP_NUM_THREADS}, @ref{omp_get_num_threads}, @ref{omp_get_max_threads} | |
432 | ||
433 | @item @emph{Reference}: | |
434 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.2.1. | |
435 | @end table | |
436 | ||
437 | ||
438 | ||
439 | @node omp_init_lock | |
440 | @section @code{omp_init_lock} -- Initialize simple lock | |
441 | @table @asis | |
442 | @item @emph{Description}: | |
443 | Initialize a simple lock. After initialization, the lock is in | |
444 | an unlocked state. | |
445 | ||
446 | @item @emph{C/C++}: | |
447 | @multitable @columnfractions .20 .80 | |
448 | @item @emph{Prototype}: @tab @code{void omp_init_lock(omp_lock_t *lock);} | |
449 | @end multitable | |
450 | ||
451 | @item @emph{Fortran}: | |
452 | @multitable @columnfractions .20 .80 | |
453 | @item @emph{Interface}: @tab @code{subroutine omp_init_lock(lock)} | |
454 | @item @tab @code{integer(omp_lock_kind), intent(out) :: lock} | |
455 | @end multitable | |
456 | ||
457 | @item @emph{See also}: | |
458 | @ref{omp_destroy_lock} | |
459 | ||
460 | @item @emph{Reference}: | |
461 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.3.1. | |
462 | @end table | |
463 | ||
464 | ||
465 | ||
466 | @node omp_set_lock | |
467 | @section @code{omp_set_lock} -- Wait for and set simple lock | |
468 | @table @asis | |
469 | @item @emph{Description}: | |
470 | Before setting a simple lock, the lock variable must be initialized by | |
471 | @code{omp_init_lock}. The calling thread is blocked until the lock | |
472 | is available. If the lock is already held by the current thread, | |
473 | a deadlock occurs. | |
474 | ||
475 | @item @emph{C/C++}: | |
476 | @multitable @columnfractions .20 .80 | |
477 | @item @emph{Prototype}: @tab @code{void omp_set_lock(omp_lock_t *lock);} | |
478 | @end multitable | |
479 | ||
480 | @item @emph{Fortran}: | |
481 | @multitable @columnfractions .20 .80 | |
482 | @item @emph{Interface}: @tab @code{subroutine omp_set_lock(lock)} | |
483 | @item @tab @code{integer(omp_lock_kind), intent(out) :: lock} | |
484 | @end multitable | |
485 | ||
486 | @item @emph{See also}: | |
487 | @ref{omp_init_lock}, @ref{omp_test_lock}, @ref{omp_unset_lock} | |
488 | ||
489 | @item @emph{Reference}: | |
490 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.3.3. | |
491 | @end table | |
492 | ||
493 | ||
494 | ||
495 | @node omp_test_lock | |
496 | @section @code{omp_test_lock} -- Test and set simple lock if available | |
497 | @table @asis | |
498 | @item @emph{Description}: | |
499 | Before setting a simple lock, the lock variable must be initialized by | |
500 | @code{omp_init_lock}. Contrary to @code{omp_set_lock}, @code{omp_test_lock} | |
501 | does not block if the lock is not available. This function returns | |
502 | @code{true} upon success,@code{false} otherwise. Here, @code{true} and | |
503 | @code{false} represent their language-specific counterparts. | |
504 | ||
505 | @item @emph{C/C++}: | |
506 | @multitable @columnfractions .20 .80 | |
507 | @item @emph{Prototype}: @tab @code{int omp_test_lock(omp_lock_t *lock);} | |
508 | @end multitable | |
509 | ||
510 | @item @emph{Fortran}: | |
511 | @multitable @columnfractions .20 .80 | |
512 | @item @emph{Interface}: @tab @code{subroutine omp_test_lock(lock)} | |
513 | @item @tab @code{logical(omp_logical_kind) :: omp_test_lock} | |
514 | @item @tab @code{integer(omp_lock_kind), intent(out) :: lock} | |
515 | @end multitable | |
516 | ||
517 | @item @emph{See also}: | |
518 | @ref{omp_init_lock}, @ref{omp_set_lock}, @ref{omp_set_lock} | |
519 | ||
520 | @item @emph{Reference}: | |
521 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.3.5. | |
522 | @end table | |
523 | ||
524 | ||
525 | ||
526 | @node omp_unset_lock | |
527 | @section @code{omp_unset_lock} -- Unset simple lock | |
528 | @table @asis | |
529 | @item @emph{Description}: | |
530 | A simple lock about to be unset must have been locked by @code{omp_set_lock} | |
531 | or @code{omp_test_lock} before. In addition, the lock must be held by the | |
532 | thread calling @code{omp_unset_lock}. Then, the lock becomes unlocked. If one | |
533 | ore more threads attempted to set the lock before, one of them is chosen to, | |
534 | again, set the lock for itself. | |
535 | ||
536 | @item @emph{C/C++}: | |
537 | @multitable @columnfractions .20 .80 | |
538 | @item @emph{Prototype}: @tab @code{void omp_unset_lock(omp_lock_t *lock);} | |
539 | @end multitable | |
540 | ||
541 | @item @emph{Fortran}: | |
542 | @multitable @columnfractions .20 .80 | |
543 | @item @emph{Interface}: @tab @code{subroutine omp_unset_lock(lock)} | |
544 | @item @tab @code{integer(omp_lock_kind), intent(out) :: lock} | |
545 | @end multitable | |
546 | ||
547 | @item @emph{See also}: | |
548 | @ref{omp_set_lock}, @ref{omp_test_lock} | |
549 | ||
550 | @item @emph{Reference}: | |
551 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.3.4. | |
552 | @end table | |
553 | ||
554 | ||
555 | ||
556 | @node omp_destroy_lock | |
557 | @section @code{omp_destroy_lock} -- Destroy simple lock | |
558 | @table @asis | |
559 | @item @emph{Description}: | |
560 | Destroy a simple lock. In order to be destroyed, a simple lock must be | |
561 | in the unlocked state. | |
562 | ||
563 | @item @emph{C/C++}: | |
564 | @multitable @columnfractions .20 .80 | |
565 | @item @emph{Prototype}: @tab @code{void omp_destroy_lock(omp_lock_t *);} | |
566 | @end multitable | |
567 | ||
568 | @item @emph{Fortran}: | |
569 | @multitable @columnfractions .20 .80 | |
570 | @item @emph{Interface}: @tab @code{subroutine omp_destroy_lock(lock)} | |
571 | @item @tab @code{integer(omp_lock_kind), intent(inout) :: lock} | |
572 | @end multitable | |
573 | ||
574 | @item @emph{See also}: | |
575 | @ref{omp_init_lock} | |
576 | ||
577 | @item @emph{Reference}: | |
578 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.3.2. | |
579 | @end table | |
580 | ||
581 | ||
582 | ||
583 | @node omp_init_nest_lock | |
584 | @section @code{omp_init_nest_lock} -- Initialize nested lock | |
585 | @table @asis | |
586 | @item @emph{Description}: | |
587 | Initialize a nested lock. After initialization, the lock is in | |
588 | an unlocked state and the nesting count is set to zero. | |
589 | ||
590 | @item @emph{C/C++}: | |
591 | @multitable @columnfractions .20 .80 | |
592 | @item @emph{Prototype}: @tab @code{void omp_init_nest_lock(omp_nest_lock_t *lock);} | |
593 | @end multitable | |
594 | ||
595 | @item @emph{Fortran}: | |
596 | @multitable @columnfractions .20 .80 | |
597 | @item @emph{Interface}: @tab @code{subroutine omp_init_nest_lock(lock)} | |
598 | @item @tab @code{integer(omp_nest_lock_kind), intent(out) :: lock} | |
599 | @end multitable | |
600 | ||
601 | @item @emph{See also}: | |
602 | @ref{omp_destroy_nest_lock} | |
603 | ||
604 | @item @emph{Reference}: | |
605 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.3.1. | |
606 | @end table | |
607 | ||
608 | ||
609 | @node omp_set_nest_lock | |
610 | @section @code{omp_set_nest_lock} -- Wait for and set simple lock | |
611 | @table @asis | |
612 | @item @emph{Description}: | |
613 | Before setting a nested lock, the lock variable must be initialized by | |
614 | @code{omp_init_nest_lock}. The calling thread is blocked until the lock | |
615 | is available. If the lock is already held by the current thread, the | |
616 | nesting count for the lock in incremented. | |
617 | ||
618 | @item @emph{C/C++}: | |
619 | @multitable @columnfractions .20 .80 | |
620 | @item @emph{Prototype}: @tab @code{void omp_set_nest_lock(omp_nest_lock_t *lock);} | |
621 | @end multitable | |
622 | ||
623 | @item @emph{Fortran}: | |
624 | @multitable @columnfractions .20 .80 | |
625 | @item @emph{Interface}: @tab @code{subroutine omp_set_nest_lock(lock)} | |
626 | @item @tab @code{integer(omp_nest_lock_kind), intent(out) :: lock} | |
627 | @end multitable | |
628 | ||
629 | @item @emph{See also}: | |
630 | @ref{omp_init_nest_lock}, @ref{omp_unset_nest_lock} | |
631 | ||
632 | @item @emph{Reference}: | |
633 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.3.3. | |
634 | @end table | |
635 | ||
636 | ||
637 | ||
638 | @node omp_test_nest_lock | |
639 | @section @code{omp_test_nest_lock} -- Test and set nested lock if available | |
640 | @table @asis | |
641 | @item @emph{Description}: | |
642 | Before setting a nested lock, the lock variable must be initialized by | |
643 | @code{omp_init_nest_lock}. Contrary to @code{omp_set_nest_lock}, | |
644 | @code{omp_test_nest_lock} does not block if the lock is not available. | |
645 | If the lock is already held by the current thread, the new nesting count | |
646 | is returned. Otherwise, the return value equals zero. | |
647 | ||
648 | @item @emph{C/C++}: | |
649 | @multitable @columnfractions .20 .80 | |
650 | @item @emph{Prototype}: @tab @code{int omp_test_nest_lock(omp_nest_lock_t *lock);} | |
651 | @end multitable | |
652 | ||
653 | @item @emph{Fortran}: | |
654 | @multitable @columnfractions .20 .80 | |
655 | @item @emph{Interface}: @tab @code{integer function omp_test_nest_lock(lock)} | |
656 | @item @tab @code{integer(omp_integer_kind) :: omp_test_nest_lock} | |
657 | @item @tab @code{integer(omp_nest_lock_kind), intent(inout) :: lock} | |
658 | @end multitable | |
659 | ||
660 | ||
661 | @item @emph{See also}: | |
662 | @ref{omp_init_lock}, @ref{omp_set_lock}, @ref{omp_set_lock} | |
663 | ||
664 | @item @emph{Reference}: | |
665 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.3.5. | |
666 | @end table | |
667 | ||
668 | ||
669 | ||
670 | @node omp_unset_nest_lock | |
671 | @section @code{omp_unset_nest_lock} -- Unset nested lock | |
672 | @table @asis | |
673 | @item @emph{Description}: | |
674 | A nested lock about to be unset must have been locked by @code{omp_set_nested_lock} | |
675 | or @code{omp_test_nested_lock} before. In addition, the lock must be held by the | |
676 | thread calling @code{omp_unset_nested_lock}. If the nesting count drops to zero, the | |
677 | lock becomes unlocked. If one ore more threads attempted to set the lock before, | |
678 | one of them is chosen to, again, set the lock for itself. | |
679 | ||
680 | @item @emph{C/C++}: | |
681 | @multitable @columnfractions .20 .80 | |
682 | @item @emph{Prototype}: @tab @code{void omp_unset_nest_lock(omp_nest_lock_t *lock);} | |
683 | @end multitable | |
684 | ||
685 | @item @emph{Fortran}: | |
686 | @multitable @columnfractions .20 .80 | |
687 | @item @emph{Interface}: @tab @code{subroutine omp_unset_nest_lock(lock)} | |
688 | @item @tab @code{integer(omp_nest_lock_kind), intent(out) :: lock} | |
689 | @end multitable | |
690 | ||
691 | @item @emph{See also}: | |
692 | @ref{omp_set_nest_lock} | |
693 | ||
694 | @item @emph{Reference}: | |
695 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.3.4. | |
696 | @end table | |
697 | ||
698 | ||
699 | ||
700 | @node omp_destroy_nest_lock | |
701 | @section @code{omp_destroy_nest_lock} -- Destroy nested lock | |
702 | @table @asis | |
703 | @item @emph{Description}: | |
704 | Destroy a nested lock. In order to be destroyed, a nested lock must be | |
705 | in the unlocked state and its nesting count must equal zero. | |
706 | ||
707 | @item @emph{C/C++}: | |
708 | @multitable @columnfractions .20 .80 | |
709 | @item @emph{Prototype}: @tab @code{void omp_destroy_nest_lock(omp_nest_lock_t *);} | |
710 | @end multitable | |
711 | ||
712 | @item @emph{Fortran}: | |
713 | @multitable @columnfractions .20 .80 | |
714 | @item @emph{Interface}: @tab @code{subroutine omp_destroy_nest_lock(lock)} | |
715 | @item @tab @code{integer(omp_nest_lock_kind), intent(inout) :: lock} | |
716 | @end multitable | |
717 | ||
718 | @item @emph{See also}: | |
719 | @ref{omp_init_lock} | |
720 | ||
721 | @item @emph{Reference}: | |
722 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.3.2. | |
723 | @end table | |
724 | ||
725 | ||
726 | ||
727 | @node omp_get_wtick | |
728 | @section @code{omp_get_wtick} -- Get timer precision | |
729 | @table @asis | |
730 | @item @emph{Description}: | |
731 | Gets the timer precision, i. e. the number of seconds between two | |
732 | successive clock ticks. | |
733 | ||
734 | @item @emph{C/C++}: | |
735 | @multitable @columnfractions .20 .80 | |
736 | @item @emph{Prototype}: @tab @code{double omp_get_wtick();} | |
737 | @end multitable | |
738 | ||
739 | @item @emph{Fortran}: | |
740 | @multitable @columnfractions .20 .80 | |
741 | @item @emph{Interface}: @tab @code{double precision function omp_get_wtick()} | |
742 | @end multitable | |
743 | ||
744 | @item @emph{See also}: | |
745 | @ref{omp_get_wtime} | |
746 | ||
747 | @item @emph{Reference}: | |
748 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.4.2. | |
749 | @end table | |
750 | ||
751 | ||
752 | ||
753 | @node omp_get_wtime | |
754 | @section @code{omp_get_wtime} -- Elapsed wall clock time | |
755 | @table @asis | |
756 | @item @emph{Description}: | |
757 | Elapsed wall clock time in seconds. The time is measured per thread, no | |
758 | guarantee can bee made that two distinct threads measure the same time. | |
759 | Time is measured from some "time in the past". On POSIX compliant systems | |
760 | the seconds since the Epoch (00:00:00 UTC, January 1, 1970) are returned. | |
761 | ||
762 | @item @emph{C/C++}: | |
763 | @multitable @columnfractions .20 .80 | |
764 | @item @emph{Prototype}: @tab @code{double omp_get_wtime();} | |
765 | @end multitable | |
766 | ||
767 | @item @emph{Fortran}: | |
768 | @multitable @columnfractions .20 .80 | |
769 | @item @emph{Interface}: @tab @code{double precision function omp_get_wtime()} | |
770 | @end multitable | |
771 | ||
772 | @item @emph{See also}: | |
773 | @ref{omp_get_wtick} | |
774 | ||
775 | @item @emph{Reference}: | |
776 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 3.4.1. | |
777 | @end table | |
778 | ||
779 | ||
780 | ||
781 | @c --------------------------------------------------------------------- | |
782 | @c Environment Variables | |
783 | @c --------------------------------------------------------------------- | |
784 | ||
785 | @node Environment Variables | |
786 | @chapter Environment Variables | |
787 | ||
788 | The variables @env{OMP_DYNAMIC}, @env{OMP_NESTED}, @env{OMP_NUM_THREADS} and | |
789 | @env{OMP_SCHEDULE} are defined by section 4 of the OpenMP specifications in | |
790 | version 2.5, while @env{GOMP_CPU_AFFINITY} and @env{GOMP_STACKSIZE} are GNU | |
791 | extensions. | |
792 | ||
793 | @menu | |
794 | * OMP_DYNAMIC:: Dynamic adjustment of threads | |
795 | * OMP_NESTED:: Nested parallel regions | |
796 | * OMP_NUM_THREADS:: Specifies the number of threads to use | |
797 | * OMP_SCHEDULE:: How threads are scheduled | |
798 | * GOMP_CPU_AFFINITY:: Bind threads to specific CPUs | |
799 | * GOMP_STACKSIZE:: Set default thread stack size | |
800 | @end menu | |
801 | ||
802 | ||
803 | @node OMP_DYNAMIC | |
804 | @section @env{OMP_DYNAMIC} -- Dynamic adjustment of threads | |
805 | @cindex Environment Variable | |
806 | @table @asis | |
807 | @item @emph{Description}: | |
808 | Enable or disable the dynamic adjustment of the number of threads | |
809 | within a team. The value of this environment variable shall be | |
810 | @code{TRUE} or @code{FALSE}. | |
811 | ||
812 | @item @emph{See also}: | |
813 | @ref{omp_set_dynamic} | |
814 | ||
815 | @item @emph{Reference}: | |
816 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 4.3 | |
817 | @end table | |
818 | ||
819 | ||
820 | ||
821 | @node OMP_NESTED | |
822 | @section @env{OMP_NESTED} -- Nested parallel regions | |
823 | @cindex Environment Variable | |
824 | @table @asis | |
825 | @item @emph{Description}: | |
826 | Enable or disable nested parallel regions, i. e. whether team members | |
827 | are allowed to create new teams. The value of this environment variable | |
828 | shall be @code{TRUE} or @code{FALSE}. | |
829 | ||
830 | @item @emph{See also}: | |
831 | @ref{omp_set_nested} | |
832 | ||
833 | @item @emph{Reference}: | |
834 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 4.4 | |
835 | @end table | |
836 | ||
837 | ||
838 | ||
839 | @node OMP_NUM_THREADS | |
840 | @section @env{OMP_NUM_THREADS} -- Specifies the number of threads to use | |
841 | @cindex Environment Variable | |
842 | @table @asis | |
843 | @item @emph{Description}: | |
844 | Specifies the number of threads to use in parallel regions. If undefined | |
845 | one thread per CPU online is used. The value of this variable shall be | |
846 | positive integer. | |
847 | ||
848 | @item @emph{See also}: | |
849 | @ref{omp_set_num_threads} | |
850 | ||
851 | @item @emph{Reference}: | |
852 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 4.2 | |
853 | @end table | |
854 | ||
855 | ||
856 | ||
857 | @node OMP_SCHEDULE | |
858 | @section @env{OMP_SCHEDULE} -- How threads are scheduled | |
859 | @cindex Environment Variable | |
860 | @table @asis | |
861 | @item @emph{Description}: | |
862 | Allows to specify @code{schedule type} and @code{chunk size}. | |
863 | The value of the variable shall have the form: @code{type[,chunk]} where | |
864 | @code{type} is one of @code{static}, @code{dynamic} or @code{guided}. | |
865 | The optional @code{chunk size} shall be a positive integer. | |
866 | ||
867 | @item @emph{Reference}: | |
868 | @uref{http://www.openmp.org/, OpenMP specifications v2.5}, section 4.1 | |
869 | @end table | |
870 | ||
871 | ||
872 | ||
873 | @node GOMP_CPU_AFFINITY | |
874 | @section @env{GOMP_CPU_AFFINITY} -- Bind threads to specific CPUs | |
875 | @cindex Environment Variable | |
876 | @table @asis | |
877 | @item @emph{Description}: | |
878 | A patch for this extension has been submitted, but was not yet applied at the | |
879 | time of writing. | |
880 | ||
881 | @item @emph{Reference}: | |
882 | @uref{http://gcc.gnu.org/ml/gcc-patches/2006-05/msg00982.html, | |
883 | GCC Patches Mailinglist} | |
884 | @uref{http://gcc.gnu.org/ml/gcc-patches/2006-05/msg01133.html, | |
885 | GCC Patches Mailinglist} | |
886 | @end table | |
887 | ||
888 | ||
889 | ||
890 | @node GOMP_STACKSIZE | |
891 | @section @env{GOMP_STACKSIZE} -- Set default thread stack size | |
892 | @cindex Environment Variable | |
893 | @table @asis | |
894 | @item @emph{Description}: | |
895 | Set the default thread stack size in kilobytes. This is in opposition | |
896 | to @code{pthread_attr_setstacksize} which gets the number of bytes as an | |
897 | argument. If the stacksize can not be set due to system constraints, an | |
898 | error is reported and the initial stacksize is left unchanged. | |
899 | ||
900 | @item @emph{Reference}: | |
901 | @uref{http://gcc.gnu.org/ml/gcc-patches/2006-06/msg00493.html, | |
902 | GCC Patches Mailinglist}, | |
903 | @uref{http://gcc.gnu.org/ml/gcc-patches/2006-06/msg00496.html, | |
904 | GCC Patches Mailinglist} | |
905 | @end table | |
906 | ||
907 | ||
908 | ||
909 | @c --------------------------------------------------------------------- | |
910 | @c The libgomp ABI | |
911 | @c --------------------------------------------------------------------- | |
912 | ||
913 | @node The libgomp ABI | |
914 | @chapter The libgomp ABI | |
915 | ||
916 | The following sections present notes on the external ABI as | |
917 | presented by libgomp. Only maintainers should need them. | |
918 | ||
919 | @menu | |
920 | * Implementing MASTER construct:: | |
921 | * Implementing CRITICAL construct:: | |
922 | * Implementing ATOMIC construct:: | |
923 | * Implementing FLUSH construct:: | |
924 | * Implementing BARRIER construct:: | |
925 | * Implementing THREADPRIVATE construct:: | |
926 | * Implementing PRIVATE clause:: | |
927 | * Implementing FIRSTPRIVATE LASTPRIVATE COPYIN and COPYPRIVATE clauses:: | |
928 | * Implementing REDUCTION clause:: | |
929 | * Implementing PARALLEL construct:: | |
930 | * Implementing FOR construct:: | |
931 | * Implementing ORDERED construct:: | |
932 | * Implementing SECTIONS construct:: | |
933 | * Implementing SINGLE construct:: | |
934 | @end menu | |
935 | ||
936 | ||
937 | @node Implementing MASTER construct | |
938 | @section Implementing MASTER construct | |
939 | ||
940 | @smallexample | |
941 | if (omp_get_thread_num () == 0) | |
942 | block | |
943 | @end smallexample | |
944 | ||
945 | Alternately, we generate two copies of the parallel subfunction | |
946 | and only include this in the version run by the master thread. | |
947 | Surely that's not worthwhile though... | |
948 | ||
949 | ||
950 | ||
951 | @node Implementing CRITICAL construct | |
952 | @section Implementing CRITICAL construct | |
953 | ||
954 | Without a specified name, | |
955 | ||
956 | @smallexample | |
957 | void GOMP_critical_start (void); | |
958 | void GOMP_critical_end (void); | |
959 | @end smallexample | |
960 | ||
961 | so that we don't get COPY relocations from libgomp to the main | |
962 | application. | |
963 | ||
964 | With a specified name, use omp_set_lock and omp_unset_lock with | |
965 | name being transformed into a variable declared like | |
966 | ||
967 | @smallexample | |
968 | omp_lock_t gomp_critical_user_<name> __attribute__((common)) | |
969 | @end smallexample | |
970 | ||
971 | Ideally the ABI would specify that all zero is a valid unlocked | |
972 | state, and so we wouldn't actually need to initialize this at | |
973 | startup. | |
974 | ||
975 | ||
976 | ||
977 | @node Implementing ATOMIC construct | |
978 | @section Implementing ATOMIC construct | |
979 | ||
980 | The target should implement the @code{__sync} builtins. | |
981 | ||
982 | Failing that we could add | |
983 | ||
984 | @smallexample | |
985 | void GOMP_atomic_enter (void) | |
986 | void GOMP_atomic_exit (void) | |
987 | @end smallexample | |
988 | ||
989 | which reuses the regular lock code, but with yet another lock | |
990 | object private to the library. | |
991 | ||
992 | ||
993 | ||
994 | @node Implementing FLUSH construct | |
995 | @section Implementing FLUSH construct | |
996 | ||
997 | Expands to the @code{__sync_synchronize} builtin. | |
998 | ||
999 | ||
1000 | ||
1001 | @node Implementing BARRIER construct | |
1002 | @section Implementing BARRIER construct | |
1003 | ||
1004 | @smallexample | |
1005 | void GOMP_barrier (void) | |
1006 | @end smallexample | |
1007 | ||
1008 | ||
1009 | @node Implementing THREADPRIVATE construct | |
1010 | @section Implementing THREADPRIVATE construct | |
1011 | ||
1012 | In _most_ cases we can map this directly to @code{__thread}. Except | |
1013 | that OMP allows constructors for C++ objects. We can either | |
1014 | refuse to support this (how often is it used?) or we can | |
1015 | implement something akin to .ctors. | |
1016 | ||
1017 | Even more ideally, this ctor feature is handled by extensions | |
1018 | to the main pthreads library. Failing that, we can have a set | |
1019 | of entry points to register ctor functions to be called. | |
1020 | ||
1021 | ||
1022 | ||
1023 | @node Implementing PRIVATE clause | |
1024 | @section Implementing PRIVATE clause | |
1025 | ||
1026 | In association with a PARALLEL, or within the lexical extent | |
1027 | of a PARALLEL block, the variable becomes a local variable in | |
1028 | the parallel subfunction. | |
1029 | ||
1030 | In association with FOR or SECTIONS blocks, create a new | |
1031 | automatic variable within the current function. This preserves | |
1032 | the semantic of new variable creation. | |
1033 | ||
1034 | ||
1035 | ||
1036 | @node Implementing FIRSTPRIVATE LASTPRIVATE COPYIN and COPYPRIVATE clauses | |
1037 | @section Implementing FIRSTPRIVATE LASTPRIVATE COPYIN and COPYPRIVATE clauses | |
1038 | ||
1039 | Seems simple enough for PARALLEL blocks. Create a private | |
1040 | struct for communicating between parent and subfunction. | |
1041 | In the parent, copy in values for scalar and "small" structs; | |
1042 | copy in addresses for others TREE_ADDRESSABLE types. In the | |
1043 | subfunction, copy the value into the local variable. | |
1044 | ||
1045 | Not clear at all what to do with bare FOR or SECTION blocks. | |
1046 | The only thing I can figure is that we do something like | |
1047 | ||
1048 | @smallexample | |
1049 | #pragma omp for firstprivate(x) lastprivate(y) | |
1050 | for (int i = 0; i < n; ++i) | |
1051 | body; | |
1052 | @end smallexample | |
1053 | ||
1054 | which becomes | |
1055 | ||
1056 | @smallexample | |
1057 | @{ | |
1058 | int x = x, y; | |
1059 | ||
1060 | // for stuff | |
1061 | ||
1062 | if (i == n) | |
1063 | y = y; | |
1064 | @} | |
1065 | @end smallexample | |
1066 | ||
1067 | where the "x=x" and "y=y" assignments actually have different | |
1068 | uids for the two variables, i.e. not something you could write | |
1069 | directly in C. Presumably this only makes sense if the "outer" | |
1070 | x and y are global variables. | |
1071 | ||
1072 | COPYPRIVATE would work the same way, except the structure | |
1073 | broadcast would have to happen via SINGLE machinery instead. | |
1074 | ||
1075 | ||
1076 | ||
1077 | @node Implementing REDUCTION clause | |
1078 | @section Implementing REDUCTION clause | |
1079 | ||
1080 | The private struct mentioned in the previous section should have | |
1081 | a pointer to an array of the type of the variable, indexed by the | |
1082 | thread's @var{team_id}. The thread stores its final value into the | |
1083 | array, and after the barrier the master thread iterates over the | |
1084 | array to collect the values. | |
1085 | ||
1086 | ||
1087 | @node Implementing PARALLEL construct | |
1088 | @section Implementing PARALLEL construct | |
1089 | ||
1090 | @smallexample | |
1091 | #pragma omp parallel | |
1092 | @{ | |
1093 | body; | |
1094 | @} | |
1095 | @end smallexample | |
1096 | ||
1097 | becomes | |
1098 | ||
1099 | @smallexample | |
1100 | void subfunction (void *data) | |
1101 | @{ | |
1102 | use data; | |
1103 | body; | |
1104 | @} | |
1105 | ||
1106 | setup data; | |
1107 | GOMP_parallel_start (subfunction, &data, num_threads); | |
1108 | subfunction (&data); | |
1109 | GOMP_parallel_end (); | |
1110 | @end smallexample | |
1111 | ||
1112 | @smallexample | |
1113 | void GOMP_parallel_start (void (*fn)(void *), void *data, unsigned num_threads) | |
1114 | @end smallexample | |
1115 | ||
1116 | The @var{FN} argument is the subfunction to be run in parallel. | |
1117 | ||
1118 | The @var{DATA} argument is a pointer to a structure used to | |
1119 | communicate data in and out of the subfunction, as discussed | |
1120 | above wrt FIRSTPRIVATE et al. | |
1121 | ||
1122 | The @var{NUM_THREADS} argument is 1 if an IF clause is present | |
1123 | and false, or the value of the NUM_THREADS clause, if | |
1124 | present, or 0. | |
1125 | ||
1126 | The function needs to create the appropriate number of | |
1127 | threads and/or launch them from the dock. It needs to | |
1128 | create the team structure and assign team ids. | |
1129 | ||
1130 | @smallexample | |
1131 | void GOMP_parallel_end (void) | |
1132 | @end smallexample | |
1133 | ||
1134 | Tears down the team and returns us to the previous @code{omp_in_parallel()} state. | |
1135 | ||
1136 | ||
1137 | ||
1138 | @node Implementing FOR construct | |
1139 | @section Implementing FOR construct | |
1140 | ||
1141 | @smallexample | |
1142 | #pragma omp parallel for | |
1143 | for (i = lb; i <= ub; i++) | |
1144 | body; | |
1145 | @end smallexample | |
1146 | ||
1147 | becomes | |
1148 | ||
1149 | @smallexample | |
1150 | void subfunction (void *data) | |
1151 | @{ | |
1152 | long _s0, _e0; | |
1153 | while (GOMP_loop_static_next (&_s0, &_e0)) | |
1154 | @{ | |
1155 | long _e1 = _e0, i; | |
1156 | for (i = _s0; i < _e1; i++) | |
1157 | body; | |
1158 | @} | |
1159 | GOMP_loop_end_nowait (); | |
1160 | @} | |
1161 | ||
1162 | GOMP_parallel_loop_static (subfunction, NULL, 0, lb, ub+1, 1, 0); | |
1163 | subfunction (NULL); | |
1164 | GOMP_parallel_end (); | |
1165 | @end smallexample | |
1166 | ||
1167 | @smallexample | |
1168 | #pragma omp for schedule(runtime) | |
1169 | for (i = 0; i < n; i++) | |
1170 | body; | |
1171 | @end smallexample | |
1172 | ||
1173 | becomes | |
1174 | ||
1175 | @smallexample | |
1176 | @{ | |
1177 | long i, _s0, _e0; | |
1178 | if (GOMP_loop_runtime_start (0, n, 1, &_s0, &_e0)) | |
1179 | do @{ | |
1180 | long _e1 = _e0; | |
1181 | for (i = _s0, i < _e0; i++) | |
1182 | body; | |
1183 | @} while (GOMP_loop_runtime_next (&_s0, _&e0)); | |
1184 | GOMP_loop_end (); | |
1185 | @} | |
1186 | @end smallexample | |
1187 | ||
1188 | Note that while it looks like there is trickyness to propagating | |
1189 | a non-constant STEP, there isn't really. We're explicitly allowed | |
1190 | to evaluate it as many times as we want, and any variables involved | |
1191 | should automatically be handled as PRIVATE or SHARED like any other | |
1192 | variables. So the expression should remain evaluable in the | |
1193 | subfunction. We can also pull it into a local variable if we like, | |
1194 | but since its supposed to remain unchanged, we can also not if we like. | |
1195 | ||
1196 | If we have SCHEDULE(STATIC), and no ORDERED, then we ought to be | |
1197 | able to get away with no work-sharing context at all, since we can | |
1198 | simply perform the arithmetic directly in each thread to divide up | |
1199 | the iterations. Which would mean that we wouldn't need to call any | |
1200 | of these routines. | |
1201 | ||
1202 | There are separate routines for handling loops with an ORDERED | |
1203 | clause. Bookkeeping for that is non-trivial... | |
1204 | ||
1205 | ||
1206 | ||
1207 | @node Implementing ORDERED construct | |
1208 | @section Implementing ORDERED construct | |
1209 | ||
1210 | @smallexample | |
1211 | void GOMP_ordered_start (void) | |
1212 | void GOMP_ordered_end (void) | |
1213 | @end smallexample | |
1214 | ||
1215 | ||
1216 | ||
1217 | @node Implementing SECTIONS construct | |
1218 | @section Implementing SECTIONS construct | |
1219 | ||
1220 | A block as | |
1221 | ||
1222 | @smallexample | |
1223 | #pragma omp sections | |
1224 | @{ | |
1225 | #pragma omp section | |
1226 | stmt1; | |
1227 | #pragma omp section | |
1228 | stmt2; | |
1229 | #pragma omp section | |
1230 | stmt3; | |
1231 | @} | |
1232 | @end smallexample | |
1233 | ||
1234 | becomes | |
1235 | ||
1236 | @smallexample | |
1237 | for (i = GOMP_sections_start (3); i != 0; i = GOMP_sections_next ()) | |
1238 | switch (i) | |
1239 | @{ | |
1240 | case 1: | |
1241 | stmt1; | |
1242 | break; | |
1243 | case 2: | |
1244 | stmt2; | |
1245 | break; | |
1246 | case 3: | |
1247 | stmt3; | |
1248 | break; | |
1249 | @} | |
1250 | GOMP_barrier (); | |
1251 | @end smallexample | |
1252 | ||
1253 | ||
1254 | @node Implementing SINGLE construct | |
1255 | @section Implementing SINGLE construct | |
1256 | ||
1257 | A block like | |
1258 | ||
1259 | @smallexample | |
1260 | #pragma omp single | |
1261 | @{ | |
1262 | body; | |
1263 | @} | |
1264 | @end smallexample | |
1265 | ||
1266 | becomes | |
1267 | ||
1268 | @smallexample | |
1269 | if (GOMP_single_start ()) | |
1270 | body; | |
1271 | GOMP_barrier (); | |
1272 | @end smallexample | |
1273 | ||
1274 | while | |
1275 | ||
1276 | @smallexample | |
1277 | #pragma omp single copyprivate(x) | |
1278 | body; | |
1279 | @end smallexample | |
1280 | ||
1281 | becomes | |
1282 | ||
1283 | @smallexample | |
1284 | datap = GOMP_single_copy_start (); | |
1285 | if (datap == NULL) | |
1286 | @{ | |
1287 | body; | |
1288 | data.x = x; | |
1289 | GOMP_single_copy_end (&data); | |
1290 | @} | |
1291 | else | |
1292 | x = datap->x; | |
1293 | GOMP_barrier (); | |
1294 | @end smallexample | |
1295 | ||
1296 | ||
1297 | ||
1298 | @c --------------------------------------------------------------------- | |
1299 | @c | |
1300 | @c --------------------------------------------------------------------- | |
1301 | ||
1302 | @node Reporting Bugs | |
1303 | @chapter Reporting Bugs | |
1304 | ||
1305 | Bugs in the GNU OpenMP implementation should be reported via | |
1306 | @uref{http://gcc.gnu.org/bugzilla/, bugzilla}. In all cases, please add | |
1307 | "openmp" to the keywords field in the bug report. | |
1308 | ||
1309 | ||
1310 | ||
1311 | @c --------------------------------------------------------------------- | |
1312 | @c GNU General Public License | |
1313 | @c --------------------------------------------------------------------- | |
1314 | ||
1315 | @include gpl.texi | |
1316 | ||
1317 | ||
1318 | ||
1319 | @c --------------------------------------------------------------------- | |
1320 | @c GNU Free Documentation License | |
1321 | @c --------------------------------------------------------------------- | |
1322 | ||
1323 | @include fdl.texi | |
1324 | ||
1325 | ||
1326 | ||
1327 | @c --------------------------------------------------------------------- | |
1328 | @c Funding Free Software | |
1329 | @c --------------------------------------------------------------------- | |
1330 | ||
1331 | @include funding.texi | |
1332 | ||
1333 | @c --------------------------------------------------------------------- | |
1334 | @c Index | |
1335 | @c --------------------------------------------------------------------- | |
1336 | ||
1337 | @node Index | |
1338 | @unnumbered Index | |
1339 | ||
1340 | @printindex cp | |
1341 | ||
1342 | @bye |