https://gcc.gnu.org/projects/gomp/

The GOMP project consists of implementation of OpenMP and OpenACC to permit annotating the source code to permit running it concurrently with thread parallelization and on offloading devices (accelerators

such as GPUs), including the associated run-time library and API routines. Both OpenMP and OpenACC are supported with GCC's C, C++ and Fortran compilers.

UsageDocumentationHistory and Project GoalContributingReporting BugsOpenMP Implementation Status:2.5·3.0·3.1·4.0·4.5·5.0·5.1·5.2·6.0·TR 14OpenMP Releases and Status

• To enable , useOpenMP.-fopenmp

can be used to enable only the SIMD vectorization and loop-transformation constructs without creating multiple threads, offloading code or adding a library dependency.-fopenmp-simd

• To enable , useOpenACC.-fopenacc

• If either is enabled, offloading is automatically generated for all offload-device types for which the compiler has been configured. Use to disable or specify the offload-devices to be used. Use-foffload=

to pass device-specific compiler and linker flags.-foffload-options=

Diagnostics

• The flag provides details about optimizations at compile time.-fopt-info

• Environment variables can be used to influence run-time behavior and output more data, useful for debugging or performance tuning. See the GNU libgompmanual for details.

GNU libgompmanual for API routines, environment variables and implementation details.OpenMP specification, including OpenMP API examples documents, reference cards and additional definitions specification.OpenACC specification.- Related GCC wiki pages: openmp,OpenACC,Offloading.

The GOMP project has developed an implementation of OpenMP for the C, C++, and Fortran compilers in the GNU Compiler Collection and is further improving it. GOMP simplifies parallel programming for a broad variety of systems. This effort operates in an open environment to attract developers and ensure applicability across multiple architectures and applications.

Traditionally, programmers have used architecture-specific methods to effectively program tightly-parallelized computers — high band-width clusters, SMP machines, or multi-core processors. Parallel programming has thus been a time-consuming and arcane task.

OpenMP offers a simple way of exploiting parallelism without interfering with algorithm design. An OpenMP program compiles and operates correctly in both parallel and serial execution environments. Using OpenMP's directive-based parallelism also simplifies converting existing serial code to efficient parallel code.

OpenMP additionally permits to offload computations to accelerators such as GPUs, making use of their highly parallel computation support. If no accelerator is available, as fallback, the computation is done on the host.

To remain relevant, free software development tools must support emerging technologies. By implementing OpenMP, GOMP provides a simple tool for creating software targeted at parallel architectures. OpenMP's platform-neutral syntax meshes well with the portability goals of GCC and other GNU projects.

GOMP includes a support library (libgomp) and extensions to target language parsers. A long-term goal is the generation of efficient and small code for OpenMP applications.

When support for OpenACC was added, the project name GOMP

was reinterpreted as GNU Offloading and Multi-Processing

instead of denoting GNU OpenMP

.

We encourage everyone to contribute changes and help test GOMP. GOMP has been merged into mainline GCC.

Please add openmp

or openacc

to the keywords field when filing a bug report.

Implementation status in libgomp manual: Mainline (GCC 16), GCC 15, GCC 14, GCC 13, GCC 12.

Disclaimer: A feature might be only fully supported in a later GCC version than listed, depending on resolved corner cases and optimizations.

• Supported since GCC 4.2

• Supported since GCC 4.4

• Supported since GCC 4.7

**C/C++:Supported since GCC 4.9Fortran:**Supported since GCC 4.9.1

**C/C++:Supported since GCC 6Fortran:**Partial support since GCC 7, full support since GCC 11

depend

clausesmutexinoutset

depend

clausedepobj

construct and depend objects depend

clause on taskwait

teams

construct outside an enclosing target regionif

, nontemporal

and order(concurrent)

in simd

constructdefaultmap

extensionshint

clause on the atomic

constructatomic

and flush

constructomp_pause_resource

and omp_pause_resource_all

runtime routinesnonmonotonic

as default loop schedule modifier for worksharing-loop constructsatomic

constructs in simd

task_reduction

clause with taskgroup

task

modifier to reduction

clausereduction

and in_reduction

clauses on taskloop

and taskloop simd

constructstaskloop

construct cancelable by cancel

constructrequires

directiveatomic_default_mem_order

)(

dynamic_allocators

)complete but no non-host devices provides

unified_address

or unified_shared_memory

complete but no non-host devices provides

unified_shared_memory

complete; see also

conditional

modifier to lastprivate

clausescan

directive and in_scan

modifier for the reduction clause

order(concurrent)

clauseloop

constructdeclare variant

directiveuse_device_addr

clause on target data

declare target

directiveallocate

clauseOMP_TARGET_OFFLOAD

env variableFortran (partial,

detach

clause to task

constructomp_fulfill_event

runtime routinedeclare target

directiveomp_get_supported_active_levels

routinein_reduction

clause on target

constructsnowait

only stubaffinity

clause to task

constructclose

omp_get_device_num

runtime routineancestor

modifier on device

clauseReverse offload supported

(full)

metadirective

directiveallocate

directivealso static variables; C++ unsupported; see also

target update

constructto

, from

and map

clausesdeclare mapper

directive| Feature | GCC Version | Comments | |---|---|---| | OpenMP directive as C++ attribute specifiers | |

nothing

directiveerror

directivemasked

constructscope

directivestrict

modifier in the grainsize

and num_tasks

clauses of the taskloop

constructalign

clause in allocate

directivealign

modifier in allocate

clausethread_limit

clause to target

constructhas_device_addr

clause to target construct

atomic

directiveseq_cst

clause on a flush

constructprivate

and firstprivate

argument to default

clause in C and C++omp_set_num_teams

, omp_set_teams_thread_limit

, omp_get_max_teams

, omp_get_teams_thread_limit

runtime routinesomp_calloc

, omp_realloc

, omp_aligned_alloc

, and omp_aligned_calloc

runtime routinesomp_alloctrait_key_t

enum: omp_atv_serialized

added, omp_atv_default

changedomp_display_env

runtime routineOMP_PLACES

syntax extensionsOMP_NUM_TEAMS

and OMP_TEAMS_THREAD_LIMIT

environment variablesunconstrained

and reproducible

modifiers on order

clauseassume

and assumes

directivesinoutset

argument to the depend

clauseomp_target_is_accessible

runtime routineomp_target_memcpy_async

and omp_target_memcpy_rect_async

runtime routinesomp_get_mapped_ptr

runtime routineomp_all_memory

reserved locatornowait

clause in taskwait

directivebegin/end declare target

syntax in C/C++target_device

selector set in context selectorsdeclare variant

directive: support elision of preprocessed code and interpret enclosed function definitions as variant functionsdeclare variant

: new clauses adjust_args

and append_args

dispatch

constructtarget update

motion clauses and map clausesinterop

directiveomp_interop_t

object support in runtime routinespresent

argument to defaultmap

clauseompt_scope_endpoint_t

enum: ompt_scope_beginend

USE

, IMPORT

, and IMPLICIT

as invalid#pragma

formindirect

clause in declare target

device_type(nohost)

/device_type(host)

for variablespresent

modifier to the map

, to

and from

clausesdeclare target

and OpenMP contextmetadirective

declare variant

ompt_sync_region_t

enum additionsompt_state_t

enum: ompt_state_wait_barrier_implementation

and ompt_state_wait_barrier_teams

ompt_callback_target_data_op_emi_t

, ompt_callback_target_emi_t

, ompt_callback_target_map_emi_t

and ompt_callback_target_submit_emi_t

ompt_callback_error_t

type| Feature | GCC Version | Comments | |---|---|---| omp_in_explicit_task routine and implicit-task-var ICV | |

omp

/ompx

/omx

sentinels and omp_

/ompx_

namespacesompx

/omx

sentinels end

directive can be on directivedestroy

clause with depobj

destroy

clause on depobj

linear

clause syntax changes and step

modifiermap

modifiers without commadeclare mapper

with iterator and present

modifiersenter

clause as alias for to

on declare target directiveto

clause on declare target directiveallocators

directive for Fortranallocate

directive for Fortran allocatables/pointersend

directive with dispatch

memspace

and traits

modifiers for uses_allocators

uses_allocators

otherwise

clause as alias for default

on metadirectivesdefault

clause on metadirectivesdeclare target

order(concurrent)

allocate

and firstprivate

clauses on scope

ompt_callback_work

map

clause in target enter/exit data

doacross

clause as alias for depend

with source

/sink

modifierdepend

with source

/sink

modifieromp_cur_iteration

keywordomp_initial_device

and omp_invalid_device

enum/PARAMETEROMP_TARGET_OFFLOAD=mandatory

all

as defaultmap

init

clause of the interop

constructmapper

as map-type modifier in declare mapper

(1) The ompx

sentinel as C/C++ pragma and C++ attributes are warned for with -Wunknown-pragmas

(implied by -Wall

) and -Wattributes

(enabled by default), respectively; for Fortran free-source code, there is a warning enabled by default and, for fixed-source code, the omx

sentinel is warned for with -Wsurprising

(enabled by -Wall

). Unknown clauses are always rejected with an error.

-std=c2x

, since GCC 14 -std=c23

)-std=c++2b

, since GCC 14 -std=c++23

)-std=f2023

)_ALL

suffix to the device-scope environment variablesnum_threads

now accepts a listOMP_NUM_THREADS

, OMP_THREAD_LIMIT

and OMP_TEAMS_THREAD_LIMIT

OMP_PLACES

OMP_DEFAULT_DEVICE

and new OMP_AVAILABLE_DEVICES

environment varsuid

trait for target devices and for OMP_AVAILABLE_DEVICES

and OMP_DEFAULT_DEVICE

OMP_THREADS_RESERVE

environment variabledecl

attribute was added to the C++ attribute syntaxdo concurrent

adjust_args

clause was extended to specify the argument by position and supports variadic argumentsinterop

operation of append_args

, allowing all modifiers of the init

clausedepobj

with repeatable clauses and the init

clausedepend

clause of the argument version of the depend

constructlooprange

clauseprivate

not supportedtaskgraph

construct including saved

modifier and replayable

clausedefault

clause on the target directive and accepting variable categoriesuse_device_ptr

and use_device_addr

reduction

clauseinit_complete

clause to the scan

directiveref

modifier to the map

clausestorage

map-type modifier; context-dependent alloc

and release

are aliasesself

modifier to map

and self

as defaultmap

argumentdelete

as delete-modifier not as map typeautomap

modifier to the enter

clause of declare_target

groupprivate

directivelocal

clause to declare_target

directivepart_size

allocator trait for interleaved

allocator partitionspin_device

, preferred_device

and target_access

allocator traitsaccess

allocator trait changespartitioner

value to partition

allocator traituses_allocators

need_device_addr

modifier to adjust_args

clauseinterop

clause to dispatch

declare_target

message

and severity

clauses to parallel

directiveself_maps

clause to requires

directiveno_openmp_constructs

assumptions clauseordered

regarding loop-transforming directivesapply

clause to loop-transforming constructsfuse

loop-transformation constructinterchange

loop-transformation constructreverse

loop-transformation constructsplit

loop-transformation constructstripe

loop-transformation constructtile

permitting association of grid and inter-tile loopsstrict

modifier keyword to num_threads

safesync

clause to the parallel

constructomp_curr_progress_width

identifieromp_get_max_progress_width

runtime routineorder(concurrent)

and, hence, the loop

constructatomic

permitted in a construct with order(concurrent)

order(concurrent)

workdistribute

directive for FortranDO CONCURRENT

as associated loop in a loop

constructtask_iteration

directive inside taskloop

threadset

clause in task-generating constructspriority

clause to target

, target_enter_data

, target_data

, target_exit_data

and target_update

device_type

clause to the target

directivetarget_data

as composite constructnowait

clause with reverse-offload target

directivesinit

clausenowait

and nogroup

may be non constantmemscope

clause to atomic

and flush

transparent

clause for multi-generational task-dependence graphscancel

construct now completes tasks with unfulfilled eventsomp_fulfill_event

routine was restricted regarding fulfillment of event variablesomp_is_free_agent

and omp_ancestor_is_free_agent

routinesomp_get_device_from_uid

and omp_get_uid_from_device

routinesomp_get_device_num_teams

, omp_set_device_num_teams

, omp_get_device_teams_thread_limit

, and omp_set_device_teams_thread_limit

routinesomp_target_memset

and omp_target_memset_async

routinesomp_get_memspace_num_resources

routineomp_get_memspace_pagesize

routineomp_get_submemspace

routineomp_init_mempartitioner

, omp_destroy_mempartitioner

, omp_init_mempartition

, omp_destroy_mempartition

, omp_mempartition_set_part

, omp_mempartition_get_user_data

routinestarget_data_op

, target

, target_map

and target_submit

callbacks and as values that set_callback

must returnompt_target_data_transfer

and ompt_target_data_transfer_async

values in ompt_target_data_op_t

enumompt_target_data_transfer_to_device

, ompt_target_data_transfer_from_device

, ompt_target_data_transfer_to_device_async

and ompt_target_data_transfer_from_device_async

of the target_data_op

OMPT type were deprecatedompt_get_buffer_limits

OMPT routine| Feature | GCC Version | Comments | |---|---|---| | Deprecation of omitting the optional white space to separate adjacent keywords in the directive-name in Fortran (fixed and free source form) | No | | Deprecation of the combiner expression in the declare_reduction argument | No | | Deprecation of the Fortran include file omp_lib.h | No |

Technical Report (TR) 14 is the first preview for OpenMP 6.1.

omp_target_is_accessible

routine returns zero| Feature | GCC Version | Comments | |---|---|---| | Deprecation of conditional-update-capture structured block without a capture statement | No |

November 13, 2025OpenMP Technical Report 14 (first preview for the OpenMP API Version 6.1) has been released [

PDF,specifications site].November 14, 2024OpenMP Version 6.0has been released.August 1, 2024OpenMP Technical Report 13 (third preview for the OpenMP API Version 6.0)has been released.November 9, 2023OpenMP Technical Report 12 (second preview for the OpenMP API Version 6.0)has been released.November 9, 2022OpenMP Technical Report 11 (first preview for the OpenMP API Version 6.0)has been released.November 9, 2021OpenMP Version 5.2has been released.July 15, 2021OpenMP Technical Report 10: Version 5.2 Public Comment Drafthas been released.November 13, 2020OpenMP Version 5.1has been released.August 20, 2020OpenMP Technical Report 9: Version 5.1 Public Comment Drafthas been released.May 7, 2020GCC 10 has been released. It adds a number of newly implemented OpenMP 5.0 features on top of the GCC 9 release such asconditional

lastprivate

clause,scan

andloop

directives,order(concurrent)

anduse_device_addr

clauses support,if

clause onsimd

construct or partial support for thedeclare variant

directive, getting closer to full support of the OpenMP 5.0 standard.May 3, 2019GCC 9 has been releasedand version 5.0 of the OpenMP specification is now partially supported in the C and C++ compilers. For details which features of OpenMP 5.0 are and which are not supported in the GCC 9 release seethis mail.November 8, 2018The

OpenMP v5.0specification has been released.May 2, 2017GCC 7 has been releasedand version 4.5 of the OpenMP specification is now partially supported in the Fortran compiler; the largest missing item is structure element mapping.April 27, 2016GCC 6 has been releasedand version 4.5 of the OpenMP specification is now supported in the C and C++ compilers.November 14, 2015The final

OpenMP v4.5specification has been released.October 13, 2015The

gomp-4_1-branch

has been merged into mainline, so GCC 6 and later will feature OpenMP v4.5 support for C and C++.June 30, 2014Fortran OpenMP v4.0 support has been backported to

gcc-4_9-branch

and will be available in the upcoming GCC 4.9.1 release.June 18, 2014The last major part of Fortran OpenMP v4.0 support has been committed into mainline.

October 11, 2013The

gomp-4_0-branch

has been merged into mainline, so GCC 4.9 and later will feature OpenMP v4.0 support for C and C++.July 23, 2013The final OpenMP v4.0 specification has been released.

August 2, 2011The

gomp-3_1-branch

has been merged into mainline, so GCC 4.7 and later will feature OpenMP v3.1 support.July 9, 2011The final OpenMP v3.1 specification has been released.

February 6, 2011A draft of the OpenMP v3.1 specification has been released for public review. The

gomp-3_1-branch

branch has been created and work began on implementing v3.1 support.June 6, 2008The

gomp-3_0-branch

has been merged into mainline, so GCC 4.4 and later will feature OpenMP v3.0 support.May 12, 2008The final OpenMP v3.0 specification has been released.

October 22, 2007Draft of the OpenMP v3.0 specification has been released for public review, the

gomp-3_0-branch

branch has been created and work began on implementing v3.0 support.March 9, 2006The branch has been merged into mainline, so starting with GCC 4.2 the compiler supports the OpenMP v2.5 specification.

November 18, 2005The branch is ready to be merged into mainline. All three front ends are functional and there should not be many corners of the standard left to implement. There are 5 main modules to merge into mainline: (1) runtime library, (2) code generation, (3) C front end, (4) C++ front end, and, (5) Fortran front end.

October 20, 2005The runtime library is functionally complete. The syntax parsers for C, C++ and Fortran are complete, though there are still dusty corners with respect to semantic translation to be resolved. Adventurous users who don't mind the compiler crashing on every other source file are encouraged to begin filing bugs.

For questions related to the use of GCC, please consult these web pages and the GCC manuals. If that fails, the gcc-help@gcc.gnu.org mailing list might help. Comments on these web pages and the development of GCC are welcome on our developer list at gcc@gcc.gnu.org. All of our lists have public archives.

Copyright (C) Free Software Foundation, Inc. Verbatim copying and distribution of this entire article is permitted in any medium, provided this notice is preserved.

These pages are maintained by the GCC team. Last modified 2026-01-20.