gcc/libgomp/icv.c
Jakub Jelinek 07dd3bcda1 openmp: Add omp_set_num_teams, omp_get_max_teams, omp_[gs]et_teams_thread_limit
OpenMP 5.1 adds env vars and functions to set and query new ICVs used
as fallback if thread_limit or num_teams clauses aren't specified on
teams construct.

The following patch implements those, though further work will be needed:
1) OpenMP 5.1 also changed the num_teams clause, so that it can specify
   both lower and upper limit for how many teams should be created and
   changed the meaning when only one expression is provided, instead of
   num_teams(expr) in 5.0 meaning num_teams(1:expr) in 5.1, it now means
   num_teams(expr:expr), i.e. while previously we could create 1 to expr
   teams, in 5.1 we have some low limit by default equal to the single
   expression provided and may not create fewer teams.
   For host teams (which we don't currently implement efficiently for
   NUMA hosts) we trivially satisfy it now by always honoring what the
   user asked for, but for the offloading teams I think we'll need to
   rethink the APIs; currently teams construct is just a call that returns
   and possibly lowers the number of teams; and whenever possible we try
   to evaluate num_teams/thread_limit already on the target construct
   and the GOMP_teams call just sets the number of teams to the minimum
   of provided and requested teams; for some cases e.g. where target
   is not combined with teams and num_teams expression calls some functions
   etc., we need to call those functions in the target region and so it is
   late to figure number of teams, but also hw could just limit what it
   is willing to create; in that case I'm afraid we need to run the target
   body multiple times and arrange for omp_get_team_num () returning the
   right values
2) we need to finally implement the NUMA handling for GOMP_teams_reg
3) I now realize I haven't added some testcase coverage, will do that
   incrementally
4) libgomp.texi needs updates for these new APIs, but also others like
   the allocator

2021-10-11  Jakub Jelinek  <jakub@redhat.com>

gcc/
	* omp-low.c (omp_runtime_api_call): Handle omp_get_max_teams,
	omp_[sg]et_teams_thread_limit and omp_set_num_teams.
libgomp/
	* omp.h.in (omp_set_num_teams, omp_get_max_teams,
	omp_set_teams_thread_limit, omp_get_teams_thread_limit): Declare.
	* omp_lib.f90.in (omp_set_num_teams, omp_get_max_teams,
	omp_set_teams_thread_limit, omp_get_teams_thread_limit): Declare.
	* omp_lib.h.in (omp_set_num_teams, omp_get_max_teams,
	omp_set_teams_thread_limit, omp_get_teams_thread_limit): Declare.
	* libgomp.h (gomp_nteams_var, gomp_teams_thread_limit_var): Declare.
	* libgomp.map (OMP_5.1): Export omp_get_max_teams{,_},
	omp_get_teams_thread_limit{,_}, omp_set_num_teams{,_,_8_} and
	omp_set_teams_thread_limit{,_,_8_}.
	* icv.c (omp_set_num_teams, omp_get_max_teams,
	omp_set_teams_thread_limit, omp_get_teams_thread_limit): New
	functions.
	* env.c (gomp_nteams_var, gomp_teams_thread_limit_var): Define.
	(omp_display_env): Print OMP_NUM_TEAMS and OMP_TEAMS_THREAD_LIMIT.
	(initialize_env): Handle OMP_NUM_TEAMS and OMP_TEAMS_THREAD_LIMIT env
	vars.
	* teams.c (GOMP_teams_reg): If thread_limit is not specified, use
	gomp_teams_thread_limit_var as fallback if not zero.  If num_teams
	is not specified, use gomp_nteams_var.
	* fortran.c (omp_set_num_teams, omp_get_max_teams,
	omp_set_teams_thread_limit, omp_get_teams_thread_limit): Add
	ialias_redirect.
	(omp_set_num_teams_, omp_set_num_teams_8_, omp_get_max_teams_,
	omp_set_teams_thread_limit_, omp_set_teams_thread_limit_8_,
	omp_get_teams_thread_limit_): New functions.
2021-10-11 12:20:22 +02:00

290 lines
6.4 KiB
C

/* Copyright (C) 2005-2021 Free Software Foundation, Inc.
Contributed by Richard Henderson <rth@redhat.com>.
This file is part of the GNU Offloading and Multi Processing Library
(libgomp).
Libgomp 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, or (at your option)
any later version.
Libgomp 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.
Under Section 7 of GPL version 3, you are granted additional
permissions described in the GCC Runtime Library Exception, version
3.1, as published by the Free Software Foundation.
You should have received a copy of the GNU General Public License and
a copy of the GCC Runtime Library Exception along with this program;
see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
<http://www.gnu.org/licenses/>. */
/* This file defines the OpenMP API entry points that operate on internal
control variables. */
#include "libgomp.h"
#include "gomp-constants.h"
#include <limits.h>
ialias_redirect (omp_get_active_level)
void
omp_set_num_threads (int n)
{
struct gomp_task_icv *icv = gomp_icv (true);
icv->nthreads_var = (n > 0 ? n : 1);
}
void
omp_set_dynamic (int val)
{
struct gomp_task_icv *icv = gomp_icv (true);
icv->dyn_var = val;
}
int
omp_get_dynamic (void)
{
struct gomp_task_icv *icv = gomp_icv (false);
return icv->dyn_var;
}
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wdeprecated-declarations"
void
omp_set_nested (int val)
{
struct gomp_task_icv *icv = gomp_icv (true);
if (val)
icv->max_active_levels_var = gomp_supported_active_levels;
else if (icv->max_active_levels_var > 1)
icv->max_active_levels_var = 1;
}
int
omp_get_nested (void)
{
struct gomp_task_icv *icv = gomp_icv (false);
return (icv->max_active_levels_var > 1
&& icv->max_active_levels_var > omp_get_active_level ());
}
#pragma GCC diagnostic pop
void
omp_set_schedule (omp_sched_t kind, int chunk_size)
{
struct gomp_task_icv *icv = gomp_icv (true);
switch (kind & ~omp_sched_monotonic)
{
case omp_sched_static:
if (chunk_size < 1)
chunk_size = 0;
icv->run_sched_chunk_size = chunk_size;
break;
case omp_sched_dynamic:
case omp_sched_guided:
if (chunk_size < 1)
chunk_size = 1;
icv->run_sched_chunk_size = chunk_size;
break;
case omp_sched_auto:
break;
default:
return;
}
icv->run_sched_var = kind;
}
void
omp_get_schedule (omp_sched_t *kind, int *chunk_size)
{
struct gomp_task_icv *icv = gomp_icv (false);
*kind = icv->run_sched_var;
*chunk_size = icv->run_sched_chunk_size;
}
int
omp_get_max_threads (void)
{
struct gomp_task_icv *icv = gomp_icv (false);
return icv->nthreads_var;
}
int
omp_get_thread_limit (void)
{
struct gomp_task_icv *icv = gomp_icv (false);
return icv->thread_limit_var > INT_MAX ? INT_MAX : icv->thread_limit_var;
}
void
omp_set_max_active_levels (int max_levels)
{
if (max_levels >= 0)
{
struct gomp_task_icv *icv = gomp_icv (true);
if (max_levels <= gomp_supported_active_levels)
icv->max_active_levels_var = max_levels;
else
icv->max_active_levels_var = gomp_supported_active_levels;
}
}
int
omp_get_max_active_levels (void)
{
struct gomp_task_icv *icv = gomp_icv (false);
return icv->max_active_levels_var;
}
int
omp_get_supported_active_levels (void)
{
return gomp_supported_active_levels;
}
void
omp_set_num_teams (int num_teams)
{
if (num_teams >= 0)
gomp_nteams_var = num_teams;
}
int
omp_get_max_teams (void)
{
return gomp_nteams_var;
}
void
omp_set_teams_thread_limit (int thread_limit)
{
if (thread_limit >= 0)
gomp_teams_thread_limit_var = thread_limit;
}
int
omp_get_teams_thread_limit (void)
{
return gomp_teams_thread_limit_var;
}
int
omp_get_cancellation (void)
{
return gomp_cancel_var;
}
int
omp_get_max_task_priority (void)
{
return gomp_max_task_priority_var;
}
omp_proc_bind_t
omp_get_proc_bind (void)
{
struct gomp_task_icv *icv = gomp_icv (false);
return icv->bind_var;
}
int
omp_get_num_places (void)
{
return gomp_places_list_len;
}
int
omp_get_place_num (void)
{
if (gomp_places_list == NULL)
return -1;
struct gomp_thread *thr = gomp_thread ();
if (thr->place == 0)
gomp_init_affinity ();
return (int) thr->place - 1;
}
int
omp_get_partition_num_places (void)
{
if (gomp_places_list == NULL)
return 0;
struct gomp_thread *thr = gomp_thread ();
if (thr->place == 0)
gomp_init_affinity ();
return thr->ts.place_partition_len;
}
void
omp_get_partition_place_nums (int *place_nums)
{
if (gomp_places_list == NULL)
return;
struct gomp_thread *thr = gomp_thread ();
if (thr->place == 0)
gomp_init_affinity ();
unsigned int i;
for (i = 0; i < thr->ts.place_partition_len; i++)
*place_nums++ = thr->ts.place_partition_off + i;
}
void
omp_set_default_allocator (omp_allocator_handle_t allocator)
{
struct gomp_thread *thr = gomp_thread ();
if (allocator == omp_null_allocator)
allocator = omp_default_mem_alloc;
thr->ts.def_allocator = (uintptr_t) allocator;
}
omp_allocator_handle_t
omp_get_default_allocator (void)
{
struct gomp_thread *thr = gomp_thread ();
if (thr->ts.def_allocator == omp_null_allocator)
return (omp_allocator_handle_t) gomp_def_allocator;
else
return (omp_allocator_handle_t) thr->ts.def_allocator;
}
ialias (omp_set_dynamic)
ialias (omp_get_dynamic)
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wdeprecated-declarations"
ialias (omp_set_nested)
ialias (omp_get_nested)
#pragma GCC diagnostic pop
ialias (omp_set_num_threads)
ialias (omp_set_schedule)
ialias (omp_get_schedule)
ialias (omp_get_max_threads)
ialias (omp_get_thread_limit)
ialias (omp_set_max_active_levels)
ialias (omp_get_max_active_levels)
ialias (omp_get_supported_active_levels)
ialias (omp_set_num_teams)
ialias (omp_get_max_teams)
ialias (omp_set_teams_thread_limit)
ialias (omp_get_teams_thread_limit)
ialias (omp_get_cancellation)
ialias (omp_get_proc_bind)
ialias (omp_get_max_task_priority)
ialias (omp_get_num_places)
ialias (omp_get_place_num)
ialias (omp_get_partition_num_places)
ialias (omp_get_partition_place_nums)
ialias (omp_set_default_allocator)
ialias (omp_get_default_allocator)