gcc/libgomp/icv.c
Marcel Vollweiler 81476bc4f4 OpenMP: omp_get_max_teams, omp_set_num_teams, and omp_{gs}et_teams_thread_limit on offload devices
This patch adds support for omp_get_max_teams, omp_set_num_teams, and
omp_{gs}et_teams_thread_limit on offload devices. That includes the usage of
device-specific ICV values (specified as environment variables or changed on a
device). In order to reuse device-specific ICV values, a copy back mechanism is
implemented that copies ICV values back from device to the host.

Additionally, a limitation of the number of teams on gcn offload devices is
implemented.  The number of teams is limited by twice the number of compute
units (one team is executed on one compute unit).  This avoids queueing
unnessecary many teams and a corresponding allocation of large amounts of
memory.  Without that limitation the memory allocation for a large number of
user-specified teams can result in an "memory access fault".
A limitation of the number of teams is already also implemented for nvptx
devices (see nvptx_adjust_launch_bounds in libgomp/plugin/plugin-nvptx.c).

gcc/ChangeLog:

	* gimplify.cc (optimize_target_teams): Set initial num_teams_upper
	to "-2" instead of "1" for non-existing num_teams clause in order to
	disambiguate from the case of an existing num_teams clause with value 1.

libgomp/ChangeLog:

	* config/gcn/icv-device.c (omp_get_teams_thread_limit): Added to
	allow processing of device-specific values.
	(omp_set_teams_thread_limit): Likewise.
	(ialias): Likewise.
	* config/nvptx/icv-device.c (omp_get_teams_thread_limit): Likewise.
	(omp_set_teams_thread_limit): Likewise.
	(ialias): Likewise.
	* icv-device.c (omp_get_teams_thread_limit): Likewise.
	(ialias): Likewise.
	(omp_set_teams_thread_limit): Likewise.
	* icv.c (omp_set_teams_thread_limit): Removed.
	(omp_get_teams_thread_limit): Likewise.
	(ialias): Likewise.
	* libgomp.texi: Updated documentation for nvptx and gcn corresponding
	to the limitation of the number of teams.
	* plugin/plugin-gcn.c (limit_teams): New helper function that limits
	the number of teams by twice the number of compute units.
	(parse_target_attributes): Limit the number of teams on gcn offload
	devices.
	* target.c (get_gomp_offload_icvs): Added teams_thread_limit_var
	handling.
	(gomp_load_image_to_device): Added a size check for the ICVs struct
	variable.
	(gomp_copy_back_icvs): New function that is used in GOMP_target_ext to
	copy back the ICV values from device to host.
	(GOMP_target_ext): Update the number of teams and threads in the kernel
	args also considering device-specific values.
	* testsuite/libgomp.c-c++-common/icv-4.c: Fixed an error in the reading
	of OMP_TEAMS_THREAD_LIMIT from the environment.
	* testsuite/libgomp.c-c++-common/icv-5.c: Extended.
	* testsuite/libgomp.c-c++-common/icv-6.c: Extended.
	* testsuite/libgomp.c-c++-common/icv-7.c: Extended.
	* testsuite/libgomp.c-c++-common/icv-9.c: New test.
	* testsuite/libgomp.fortran/icv-5.f90: New test.
	* testsuite/libgomp.fortran/icv-6.f90: New test.

gcc/testsuite/ChangeLog:

	* c-c++-common/gomp/target-teams-1.c: Adapt expected values for
	num_teams from "1" to "-2" in cases without num_teams clause.
	* g++.dg/gomp/target-teams-1.C: Likewise.
	* gfortran.dg/gomp/defaultmap-4.f90: Likewise.
	* gfortran.dg/gomp/defaultmap-5.f90: Likewise.
	* gfortran.dg/gomp/defaultmap-6.f90: Likewise.
2022-12-06 06:03:50 -08:00

260 lines
5.9 KiB
C

/* Copyright (C) 2005-2022 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;
}
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_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)