OpenMP API 4.0 C/C++ Page 1 OpenMP 4.0 API C/C++Syntax Quick Reference Card OpenMP tions C/C++ Directives entry at the top and a single exit at the bottom target update2sa到 eou网. 3cordeen5istentwntmth ault shared I none) m听 atel侧 d o simd [28.11 declare target 00p2.7.125.1】 2他化地a网 ared I none) function definition or declaratio redchionlredtcton-identferil step/) distribute [simd] d: nt -list o the c loop simd at can sing SIMD i that th ections [2.7.2)[2.5.2] target [data] distribute parallel for [simd] pra structured-block) Contirued D 2013 OpenMP ARE OMP1013
© 2013 OpenMP ARB OMP1013C OpenMP API 4.0 C/C++ Page 1 OpenMP 4.0 API C/C++ Syntax Quick Reference Card C/C++ OpenMP Application Program Interface (API) is a portable, scalable model that gives parallel programmers a simple and flexible interface for developing portable parallel applications. OpenMP supports multi-platform shared-memory parallel programming in C/C++ and Fortran on all architectures, including Unix platforms and Windows platforms. See www.openmp.org for specifications. 4.0 Refers to functionality new in version 4.0. [n.n.n] refers to sections in the OpenMP API specification version 4.0, and [n.n.n] refers to version 3.1. ® Directives An OpenMP executable directive applies to the succeeding structured block or an OpenMP construct. Each directive starts with #pragma omp. The remainder of the directive follows the conventions of the C and C++ standards for compiler directives. A structured-block is a single statement or a compound statement with a single entry at the top and a single exit at the bottom. parallel [2.5] [2.4] Forms a team of threads and starts parallel execution. #pragma omp parallel [clause[ [, ]clause] ...] structured-block clause: if(scalar-expression) num_threads(integer-expression) default(shared | none) private(list) firstprivate(list) shared(list) copyin(list) reduction(reduction-identifier: list) 4.0 proc_bind(master | close | spread) loop [2.7.1] [2.5.1] Specifies that the iterations of associated loops will be executed in parallel by threads in the team in the context of their implicit tasks. #pragma omp for [clause[ [, ]clause] ...] for-loops clause: private(list) firstprivate(list) lastprivate(list) reduction(reduction-identifier: list) schedule(kind[, chunk_size]) collapse(n) ordered nowait kind: • static: Iterations are divided into chunks of size chunk_size and assigned to threads in the team in round-robin fashion in order of thread number. • dynamic: Each thread executes a chunk of iterations then requests another chunk until none remain. • guided: Each thread executes a chunk of iterations then requests another chunk until no chunks remain to be assigned. • auto: The decision regarding scheduling is delegated to the compiler and/or runtime system. • runtime: The schedule and chunk size are taken from the run-sched-var ICV. sections [2.7.2] [2.5.2] A noniterative worksharing construct that contains a set of structured blocks that are to be distributed among and executed by the threads in a team. #pragma omp sections [clause[ [, ] clause] ...] { [#pragma omp section] structured-block [#pragma omp section structured-block] ... } clause: private(list) firstprivate(list) lastprivate(list) reduction(reduction-identifier: list) nowait single [2.7.3] [2.5.3] Specifies that the associated structured block is executed by only one of the threads in the team. #pragma omp single [clause[ [, ]clause] ...] structured-block clause: private(list) firstprivate(list) copyprivate(list) nowait 4.0 simd [2.8.1] Applied to a loop to indicate that the loop can be transformed into a SIMD loop. #pragma omp simd [clause[ [, ]clause] ...] for-loops clause: safelen(length) linear(list[:linear-step]) aligned(list[:alignment]) private(list) lastprivate(list) reduction(reduction-identifier: list) collapse(n) 4.0 declare simd [2.8.2] Enables the creation of one or more versions that can process multiple arguments using SIMD instructions from a single invocation from a SIMD loop. #pragma omp declare simd [clause[ [, ]clause] ...] [#pragma omp declare simd [clause[ [, ]clause] ...] ] [...] function definition or declaration clause: simdlen(length) linear(argument-list[:constant-linear-step]) aligned(argument-list[:alignment]) uniform(argument-list) inbranch notinbranch 4.0 loop simd [2.8.3] Specifies that a loop that can be executed concurrently using SIMD instructions, and that those iterations will also be executed in parallel by threads in the team. #pragma omp for simd [clause[ [, ]clause] ...] for-loops clause: Any accepted by the simd or for directives with identical meanings and restrictions. 4.0 target [data] [2.9.1, 2.9.2] These constructs create a device data environment for the extent of the region. target also starts execution on the device. #pragma omp target data [clause[ [, ]clause] ...] structured-block #pragma omp target [clause[ [, ]clause] ...] structured-block clause: device(integer-expression) map([map-type: ] list) if(scalar-expression) 4.0 target update [2.9.3] Makes the corresponding list items in the device data environment consistent with their original list items, according to the specified motion clauses. #pragma omp target update clause[ [, ]clause] ,...] clause is motion-clause or one of: device(integer-expression) if(scalar-expression) motion-clause: to(list) from(list) 4.0 declare target [2.9.4] A declarative directive that specifies that variables and functions are mapped to a device. #pragma omp declare target declarations-definition-seq #pragma omp end declare target 4.0 teams [2.9.5] Creates a league of thread teams where the master thread of each team executes the region. #pragma omp teams [clause[ [, ]clause] ,...] structured-block clause: num_teams(integer-expression) thread_limit(integer-expression) default(shared | none) private(list) firstprivate(list) shared(list) reduction(reduction-identifier: list) 4.0 distribute [simd] [2.9.6, 2.9.7] distribute specifies loops which are executed by the thread teams. distribute simd specifies loops which are executed concurrently using SIMD instructions. #pragma omp distribute [clause[ [, ]clause] ...] for-loops #pragma omp distribute simd [clause[ [, ]clause] ...] for-loops clause: private(list) firstprivate(list) collapse(n) dist_schedule(kind[, chunk_size]) 4.0 distribute parallel for [simd] [2.9.8, 2.9.9] These constructs specify a loop that can be executed in parallel [using SIMD semantics in the simd case] by multiple threads that are members of multiple teams. #pragma omp distribute parallel for [clause[ [, ]clause] ...] for-loops #pragma omp distribute parallel for simd [clause[ [, ]clause] ...] for-loops clause: See clause for distribute Continued4
OpenMP API4.0C/C++ Page 2 Directives(Continued) allel loop [2.10.1112.61 ask211127.1 DetnesneelatastThe.dbenvionmeatottheasl arallel sections [2.10.21 [26.21 ush2228 te de dependence-type:list) pragma omp parallel loop simd cancel or label pragma omp cancel construct-type-clausecuse) e-clouse onstruct containing a master 2 fo target or teams ifiscalbrepressonl ute5md2o.6,2a.7 critical [2.12.2][2.8.21 ne a iatedstrnvedturedbocktoa ne) ellation point construct-type-clouse e:Any clause used for teams or distribute [simd] ct-type-cause: ms distribu Hpragma omp barrier askwait 12.12 41 taskgroup [2.12.51 skgroup tareet teams distribute simd eatakaskgoepbowa市to e2129 ms distribute parallel for [simd] arallel for simd]. update I capture] stm an capture seq_cst/ pmaa [simd]cons ute or teams distribute D 2013 OpenMP ARB 0MP10130
OpenMP API 4.0 C/C++ Page 2 © 2013 OpenMP ARB OMP1013C Directives (Continued) parallel loop [2.10.1] [2.6.1] Shortcut for specifying a parallel construct containing one or more associated loops and no other statements. #pragma omp parallel for [clause[ [, ]clause] ...] for-loop clause: Any accepted by the parallel or for directives, except the nowait clause, with identical meanings and restrictions. parallel sections [2.10.2] [2.6.2] Shortcut for specifying a parallel construct containing one sections construct and no other statements. #pragma omp parallel sections [clause[ [, ]clause] ...] { [#pragma omp section] structured-block [#pragma omp section structured-block] ... } clause: Any of the clauses accepted by the parallel or sections directives, except the nowait clause, with identical meanings and restrictions. 4.0 parallel loop simd [2.10.4] Shortcut for specifying a parallel construct containing one loop SIMD construct and no other statements. #pragma omp parallel for simd [clause[ [, ]clause] ...] for-loops clause: Any accepted by the parallel, for or simd directives, except the nowait clause, with identical meanings and restrictions. 4.0 target teams [2.10.5] Shortcut for specifying a target construct containing a teams construct. #pragma omp target teams [clause[ [, ]clause] ...] structured-block clause: See clause for target or teams 4.0 teams distribute [simd] [2.10.6, 2.10.7] Shortcuts for specifying a teams construct containing a distribute [simd] construct. #pragma omp teams distribute [clause[ [, ]clause] ...] for-loops #pragma omp teams distribute simd [clause[ [, ]clause] ...] for-loops clause: Any clause used for teams or distribute [simd] 4.0 target teams distribute [simd] [2.10.8, 2.10.9] Shortcuts for specifying a target construct containing a teams distribute [simd] construct. #pragma omp target teams distribute [clause[ [, ]clause] ...] for-loops #pragma omp target teams distribute simd [clause[ [, ]clause] ...] for-loops clause: Any clause used for target or teams distribute [simd] 4.0 teams distribute parallel for [simd] [2.10.10, 12] Shortcuts for specifying a teams construct containing a distribute parallel for [simd] construct. #pragma omp teams distribute parallel for [clause[ [, ]clause] ...] for-loops #pragma omp teams distribute parallel for simd [clause[ [, ]clause] ...] for-loops clause: Any clause used for teams or distribute parallel for [simd] 4.0 target teams distribute parallel for [simd] [2.10.11, 13] Shortcut for specifying a target construct containing a teams distribute parallel for [simd] construct. #pragma omp target teams distribute parallel for \ [clause[ [, ]clause] ...] for-loops #pragma omp target teams distribute parallel for simd \ [clause[ [, ]clause] ...] for-loops clause: Any clause used for target or teams distribute parallel for [simd] task [2.11.1] [2.7.1] Defines an explicit task. The data environment of the task is created according to data-sharing attribute clauses on task construct and any defaults that apply. #pragma omp task [clause[ [, ]clause] ...] structured-block clause: if(scalar-expression) final(scalar-expression) untied default(shared | none) mergeable private(list) firstprivate(list) shared(list) 4.0 depend(dependence-type: list) The list items that appear in the depend clause may include array sections. dependence-type: The generated task will be a dependent task of all previously generated sibling tasks that reference at least one of the list items... • in: ...in an out or inout clause. • out and inout: ...in an in, out, or inout clause. taskyield [2.11.2] [2.7.2] Specifies that the current task can be suspended in favor of execution of a different task. #pragma omp taskyield master [2.12.1] [2.8.1] Specifies a structured block that is executed by the master thread of the team. #pragma omp master structured-block critical [2.12.2] [2.8.2] Restricts execution of the associated structured block to a single thread at a time. #pragma omp critical [(name)] structured-block barrier [2.12.3] [2.8.3] Specifies an explicit barrier at the point at which the construct appears. #pragma omp barrier taskwait [2.12.4] [2.8.4], 4.0 taskgroup [2.12.5] These constructs each specify a wait on the completion of child tasks of the current task. taskgroup also waits for descendant tasks. #pragma omp taskwait #pragma omp taskgroup structured-block atomic [2.12.6] [2.8.5] Ensures that a specific storage location is accessed atomically. [seq_cst] is 4.0 . #pragma omp atomic [read | write | update | capture] [seq_cst] expression-stmt #pragma omp atomic capture [seq_cst] structured-block where expression-stmt may be one of: if clause is... expression-stmt: read v = x; write x = expr; update or is not present x++; x--; ++x; --x; x binop= expr; x = x binop expr; x = expr binop x; capture v=x++; v=x--; v=++x; v= --x; v=x binop= expr; v=x = x binop expr; v=x = expr binop x; (Continued >) atomic (continued) and where structured-block may be one of the following forms: {v = x; x binop= expr;} {x binop= expr; v = x;} {v = x; x = x binop expr;} {v = x; x = expr binop x;} {x = x binop expr; v = x;} {x = expr binop x; v = x;} {v = x; x = expr;} {v = x; x++;} {v = x; ++x;} {++x; v = x;} {x++; v = x;} {v = x; x--;} {v = x; --x;} {--x; v = x;} {x--; v = x;} flush [2.12.7] [2.8.6] Executes the OpenMP flush operation, which makes a thread’s temporary view of memory consistent with memory, and enforces an order on the memory operations of the variables. #pragma omp flush [(list)] ordered [2.12.8] [2.8.7] Specifies a structured block in a loop region that will be executed in the order of the loop iterations. #pragma omp ordered structured-block 4.0 cancel [2.13.1] Requests cancellation of the innermost enclosing region of the type specified. The cancel directive may not be used in place of the statement following an if, while, do, switch, or label. #pragma omp cancel construct-type-clause[ [, ] if-clause] construct-type-clause: parallel sections for taskgroup if-clause: if(scalar-expression) 4.0 cancellation point [2.13.2] Introduces a user-defined cancellation point at which tasks check if cancellation of the innermost enclosing region of the type specified has been requested. #pragma omp cancellation point construct-type-clause construct-type-clause: parallel sections for taskgroup threadprivate [2.14.2] [2.9.2] Specifies that variables are replicated, with each thread having its own copy. Each copy of a threadprivate variable is initialized once prior to the first reference to that copy. #pragma omp threadprivate(list) list: A comma-separated list of file-scope, namespacescope, or static block-scope variables that do not have incomplete types. 4.0 declare reduction [2.15] Declares a reduction-identifier that can be used in a reduction clause. #pragma omp declare reduction(reduction-identifier : typename-list : combiner) [initializer-clause] reduction-identifier: A base language identifier or one of the following operators: +, -, *, &, |, ^, && and || In C++, this may also be an operator-function-id typename-list: A list of type names combiner: An expression initializer-clause: initializer ( omp_priv = initializer | function-name (argument-list ))
OpenMP API 4.0 C/C++ Page 3 Runtime Library Routines Retumn types are show in greer Execution environment routines affect and monitor threads.processors,and the parallel environment.The library routines are external functions with"C"linkage cution Environm nent Routines mp_set_num_thn mp get_num ot sp omp_get_num_,teams [32z间 See kind above. void omp_set_num_thre 22 omp get_team_num [3.2.27) mostcdosing nt omp get thread limit void 1.Ihe int_um): omp _getmax threads int omp_get_team_num(void): h void omptmaxactive leveb(int m) omp get thread_num it omp get_ma Lock Routines omp tet_num_procs region,ret the nt omp get levellyoidl: ck t"lockh omp_in_parallel 2632.6 ICVs greaterthan void omp_init nest___) e2321周 rov lock omp set dynamic void omp_destroy nest lock(ompest lock) Setlo omp_get_dynamic Unset lock [3.B3a uted Test lock [.3.5]3.3.5] nt omp_get_ca elaiontkoid omp_get_proc_bind ism,by setting the int omp_nest lock(omp_nest lock_ Timing Routines omp_set_schedule iihhetethartapiedhennuntime格tsed omp-set ometwme double omp_get_wtimelvoid): vod omp_set_sc r(seconds betweer omp_get_default_device cke)us omewtime. double omp get wtick void): D 2013 DocnMP ARB OMP1013
© 2013 OpenMP ARB OMP1013C OpenMP API 4.0 C/C++ Page 3 Runtime Library Routines Return types are shown in green. Execution environment routines affect and monitor threads, processors, and the parallel environment. The library routines are external functions with “C” linkage. Execution Environment Routines omp_set_num_threads [3.2.1] [3.2.1] Affects the number of threads used for subsequent parallel regions not specifying a num_threads clause, by setting the value of the first element of the nthreads-var ICV of the current task to num_threads. void omp_set_num_threads(int num_threads); omp_get_num_threads [3.2.2] [3.2.2] Returns the number of threads in the current team. The binding region for an omp_get_num_threads region is the innermost enclosing parallel region. int omp_get_num_threads(void); omp_get_max_threads [3.2.3] [3.2.3] Returns an upper bound on the number of threads that could be used to form a new team if a parallel construct without a num_threads clause were encountered after execution returns from this routine. int omp_get_max_threads(void); omp_get_thread_num [3.2.4] [3.2.4] Returns the thread number of the calling thread within the current team. int omp_get_thread_num(void); omp_get_num_procs [3.2.5] [3.2.5] Returns the number of processors that are available to the device at the time the routine is called. int omp_get_num_procs(void); omp_in_parallel [3.2.6] [3.2.6] Returns true if the active-levels-var ICV is greater than zero; otherwise it returns false. int omp_in_parallel(void); omp_set_dynamic [3.2.7] [3.2.7] Returns the value of the dyn-var ICV, which indicates if dynamic adjustment of the number of threads is enabled or disabled. void omp_set_dynamic(int dynamic_threads); omp_get_dynamic [3.2.8] [3.2.8] This routine returns the value of the dyn-var ICV, which is true if dynamic adjustment of the number of threads is enabled for the current task. int omp_get_dynamic(void); 4.0 omp_get_cancellation [3.2.9] Returns the value of the cancel-var ICV, which controls the behavior of cancel construct and cancellation points. int omp_get_cancellation(void); omp_set_nested [3.2.10] [3.2.9] Enables or disables nested parallelism, by setting the nest-var ICV. void omp_set_nested(int nested); omp_get_nested [3.2.11] [3.2.10] Returns the value of the nest-var ICV, which indicates if nested parallelism is enabled or disabled. int omp_get_nested(void); omp_set_schedule [3.2.12] [3.2.11] Affects the schedule that is applied when runtime is used as schedule kind. void omp_set_schedule(omp_sched_t kind, int modifier); kind: one of the folowing, or an implementation-defined schedule: omp_sched_static = 1 omp_sched_dynamic = 2 omp_sched_guided = 3 omp_sched_auto = 4 omp_get_schedule [3.2.13] [3.2.12] Returns the value of run-sched-var ICV, which is the schedule applied when runtime schedule is used. void omp_get_schedule( omp_sched_t *kind, int *modifier); See kind above. omp_get_thread_limit [3.2.14] [3.2.13] Returns the value of the thread-limit-var ICV, which is the maximum number of OpenMP threads available. int omp_get_thread_limit(void); omp_set_max_active_levels [3.2.15] [3.2.14] Limits the number of nested active parallel regions, by setting max-active-levels-var ICV. void omp_set_max_active_levels(int max_levels); omp_get_max_active_levels [3.2.16] [3.2.15] Returns the value of max-active-levels-var ICV, which determines the maximum number of nested active parallel regions. int omp_get_max_active_levels(void); omp_get_level [3.2.17] [3.2.16] For the enclosing device region, returns the levels-vars ICV, which is the number of nested parallel regions that enclose the task containing the call. int omp_get_level(void); omp_get_ancestor_thread_num [3.2.18] [3.2.17] Returns, for a given nested level of the current thread, the thread number of the ancestor of the current thread. int omp_get_ancestor_thread_num(int level); omp_get_team_size [3.2.19] [3.2.18] Returns, for a given nested level of the current thread, the size of the thread team to which the ancestor or the current thread belongs. int omp_get_team_size(int level); omp_get_active_level [3.2.20] [3.2.19] Returns the value of the active-level-vars ICV, which determines the number of active, nested parallel regions enclosing the task that contains the call. int omp_get_active_level(void); omp_in_final [3.2.21] [3.2.20] Returns true if the routine is executed in a final task region; otherwise, it returns false. int omp_in_final(void); 4.0 omp_get_proc_bind [3.2.22] Returns the thread affinity policy to be used for the subsequent nested parallel regions that do not specify a proc_bind clause. omp_proc_bind_t omp_get_proc_bind(void); Returns one of: omp_proc_bind_false = 0 omp_proc_bind_true = 1 omp_proc_bind_master = 2 omp_proc_bind_close = 3 omp_proc_bind_spread = 4 4.0 omp_set_default_device [3.2.23] Controls the default target device by assigning the value of the default-device-var ICV. void omp_set_default_device(int device_num); 4.0 omp_get_default_device [3.2.24] Returns the default target device. int omp_get_default_device(void); 4.0 omp_get_num_devices [3.2.25] Returns the number of target devices. int omp_get_num_devices(void); 4.0 omp_get_num_teams [3.2.26] Returns the number of teams in the current teams region, or 1 if called from outside of a teams region. int omp_get_num_teams(void); 4.0 omp_get_team_num [3.2.27] Returns the team number of calling thread. The team number is an integer between 0 and one less than the value returned by omp_get_num_teams, inclusive. int omp_get_team_num(void); 4.0 omp_is_initial_device [3.2.28] Returns true if the current task is executing on the host device; otherwise, it returns false. int omp_is_initial_device(void); Lock Routines General-purpose lock routines. Two types of locks are supported: simple locks and nestable locks. A nestable lock can be set multiple times by the same task before being unset; a simple lock cannot be set if it is already owned by the task trying to set it. Initialize lock [3.3.1] [3.3.1] Initialize an OpenMP lock. void omp_init_lock(omp_lock_t *lock); void omp_init_nest_lock(omp_nest_lock_t *lock); Destroy lock [3.3.2] [3.3.2] Ensure that the OpenMP lock is uninitialized. void omp_destroy_lock(omp_lock_t *lock); void omp_destroy_nest_lock(omp_nest_lock_t *lock); Set lock [3.3.3] [3.3.3] Sets an OpenMP lock. The calling task region is suspended until the lock is set. void omp_set_lock(omp_lock_t *lock); void omp_set_nest_lock(omp_nest_lock_t *lock); Unset lock [3.3.4] [3.3.4] Unsets an OpenMP lock. void omp_unset_lock(omp_lock_t *lock); void omp_unset_nest_lock(omp_nest_lock_t *lock); Test lock [3.3.5] [3.3.5] Attempt to set an OpenMP lock but do not suspend execution of the task executing the routine. int omp_test_lock(omp_lock_t *lock); int omp_test_nest_lock(omp_nest_lock_t *lock); Timing Routines Timing routines support a portable wall clock timer.These record elapsed time per-thread and are not guaranteed to be globally consistent across all the threads participating in an application. omp_get_wtime [3.4.1] [3.4.1] Returns elapsed wall clock time in seconds. double omp_get_wtime(void); omp_get_wtick [3.4.2] [3.4.2] Returns the precision of the timer (seconds between ticks) used by omp_get_wtime. double omp_get_wtick(void);
OpenMP API 4.0 C/C++ Page4 Environment Variables [4] Environment variable names are upper case,and the values assigned to them are case insensitive and may have leading and trailing white space. [4.11]OMP_CANCELLATION policy [4.9][4.8]OMP_MAX_ACTIVE_LEVELS levels [4.1][4.1]OMP_SCHEDULE typel,chunk] s the ca el-var ICV.policy may be true or false Sets the rICy that c ols the ets (46]45]OMP_NESTED nested [4.13]OMP_DEFAULT_DEVICE device [4.2][4.2]OMP_NUM_THREADS ist OpenMp [4.12]OMP DISPLAY ENV var kilobytes (K). -vor ICV that defines the OpenMP ols the number of able to the execution envi nment.places is threads participating in the OpenMP program. is displayed. [4.8][4.7]OMP_WAIT_POLICY policy [4.3][4.3]OMP_DYNAMIC dynamic [4.4][4.4]OMP_PROC_BIND policy is of master, and PASSIVE. Clauses 二2装aa的贤 ibed with the direc with the directive Data Sharing Attribute Clauses [214 Operators for reduction (initialization values) SIMD Clauses [2.8.1] 01 afelen(ength) attributes to be shared shared(list) collapse(n) min (Largest representable em type A constant nositive inte ession that specifies how Data Copying Clauses [214]2.) aligned(argument-list[:olignment]) firstprivate(ist) the target plaform Map Clause 2.14.5 original item has when the c tisencoUuntere map(mop-type:ist) ment-list) more arguments to have an invariant kroaSMge man-tyne that the always be called from inearlstfinearstepl inside aconditional statement of a SIMD loop. Specifies that the function will never be called from inside (Continued >a conditional statement of a SIMD loop. cture R MPArchitecture noweonoe o tatement: hand the e of OpenMP Language Application Program Interface Specification. 2013 OpenMP ARB OMP1013C
OpenMP API 4.0 C/C++ Page 4 © 2013 OpenMP ARB OMP1013C Clauses The set of clauses that is valid on a particular directive is described with the directive. Most clauses accept a comma-separated list of list items. All list items appearing in a clause must be visible, according to the scoping rules of the base language. Not all of the clauses listed in this section are valid on all directives. The set of clauses that is valid on a particular directive is described with the directive. Data Sharing Attribute Clauses [2.14.3] [2.9.3] Data-sharing attribute clauses apply only to variables whose names are visible in the construct on which the clause appears. default(shared | none) Explicitly determines the default data-sharing attributes of variables that are referenced in a parallel, task, or teams construct, causing all variables referenced in the construct that have implicitly determined data-sharing attributes to be shared. shared(list) Declares one or more list items to be shared by tasks generated by a parallel, task, or teams construct. The programmer must ensure that storage shared by an explicit task region does not reach the end of its lifetime before the explicit task region completes its execution. private(list) Declares one or more list items to be private to a task or a SIMD lane. Each task that references a list item that appears in a private clause in any statement in the construct receives a new list item. firstprivate(list) Declares list items to be private to a task, and initializes each of them with the value that the corresponding original item has when the construct is encountered. lastprivate(list) Declares one or more list items to be private to an implicit task or to a SIMD lane, and causes the corresponding original list item to be updated after the end of the region. 4.0 linear(list[:linear-step]) Declares one or more list items to be private to a SIMD lane and to have a linear relationship with respect to the iteration space of a loop. reduction(reduction-identifier:list) Specifies a reduction-identifier and one or more list items. The reduction-identifier must match a previously declared reduction-identifier of the same name and type for each of the list items. Operators for reduction (initialization values) + (0) | (0) * (1) ^ (0) - (0) && (1) & (~0) || (0) max (Least representable number in reduction list item type) min (Largest representable number in reduction list item type) Data Copying Clauses [2.14.4] [2.9.4] copyin(list) Copies the value of the master thread’s threadprivate variable to the threadprivate variable of each other member of the team executing the parallel region. copyprivate(list) Broadcasts a value from the data environment of one implicit task to the data environments of the other implicit tasks belonging to the parallel region. 4.0 Map Clause [2.14.5] map([map-type:]ist) Map a variable from the task’s data environment to the device data environment associated with the construct. map-type: alloc: On entry to the region each new corresponding list item has an undefined initial value. to: On entry to the region each new corresponding list item is initialized with the original list item’s value. from: On exit from the region the corresponding list item’s value is assigned to each original list item. (Continued >) tofrom: (Default) On entry to the region each new corresponding list item is initialized with the original list item’s value, and on exit from the region the corresponding list item’s value is assigned to each original list item. 4.0 SIMD Clauses [2.8.1] safelen(length) If used then no two iterations executed concurrently with SIMD instructions can have a greater distance in the logical iteration space than its value. collapse(n) A constant positive integer expression that specifies how many loops are associated with the loop construct. simdlen(length) A constant positive integer expression that specifies the number of concurrent arguments of the function. aligned(argument-list[:alignment]) Declares one or more list items to be aligned to the specified number of bytes. alignment, if present, must be a constant positive integer expression. If no optional parameter is specified, implementation-defined default alignments for SIMD instructions on the target platforms are assumed. uniform(argument-list) Declares one or more arguments to have an invariant value for all concurrent invocations of the function in the execution of a single SIMD loop. inbranch Specifies that the function will always be called from inside a conditional statement of a SIMD loop. notinbranch Specifies that the function will never be called from inside a conditional statement of a SIMD loop. Environment Variables [4] Environment variable names are upper case, and the values assigned to them are case insensitive and may have leading and trailing white space. 4.0 [4.11] OMP_CANCELLATION policy Sets the cancel-var ICV. policy may be true or false. If true, the effects of the cancel construct and of cancellation points are enabled and cancellation is activated 4.0 [4.13] OMP_DEFAULT_DEVICE device Sets the default-device-var ICV that controls the default device number to use in device constructs. 4.0 [4.12] OMP_DISPLAY_ENV var If var is TRUE, instructs the runtime to display the OpenMP version number and the value of the ICVs associated with the environment variables as name=value pairs. If var is VERBOSE, the runtime may also display vendor-specific variables. If var is FALSE, no information is displayed. [4.3] [4.3] OMP_DYNAMIC dynamic Sets the dyn-var ICV. If true, the implementation may dynamically adjust the number of threads to use for executing parallel regions. [4.9] [4.8] OMP_MAX_ACTIVE_LEVELS levels Sets the max-active-levels-var ICV that controls the maximum number of nested active parallel regions. [4.6] [4.5] OMP_NESTED nested Sets the nest-var ICV to enable or to disable nested parallelism. Valid values for nested are true or false. [4.2] [4.2] OMP_NUM_THREADS list Sets the nthreads-var ICV for the number of threads to use for parallel regions. 4.0 [4.5] OMP_PLACES places Sets the place-partition-var ICV that defines the OpenMP places available to the execution environment. places is an abstract name (threads, cores, sockets, or implementation-defined), or a list of non-negative numbers. [4.4] [4.4] OMP_PROC_BIND policy Sets the value of the global bind-var ICV, which sets the thread affinity policy to be used for parallel regions at the corresponding nested level. policy can be the values true, false, or a comma-separated list of master, close, or spread in quotes. [4.1] [4.1] OMP_SCHEDULE type[,chunk] Sets the run-sched-var ICV for the runtime schedule type and chunk size. Valid OpenMP schedule types are static, dynamic, guided, or auto. [4.7] [4.6] OMP_STACKSIZE size[B | K | M | G] Sets the stacksize-var ICV that specifies the size of the stack for threads created by the OpenMP implementation. size is a positive integer that specifies stack size. If unit is not specified, size is measured in kilobytes (K). [4.10] [4.9] OMP_THREAD_LIMIT limit Sets the thread-limit-var ICV that controls the number of threads participating in the OpenMP program. [4.8] [4.7] OMP_WAIT_POLICY policy Sets the wait-policy-var ICV that provides a hint to an OpenMP implementation about the desired behavior of waiting threads. Valid values for policy are ACTIVE (waiting threads consume processor cycles while waiting) and PASSIVE. Copyright © 2013 OpenMP Architecture Review Board. Permission to copy without fee all or part of this material is granted, provided the OpenMP Architecture Review Board copyright notice and the title of this document appear. Notice is given that copying is by permission of the OpenMP Architecture Review Board. Products or publications based on one or more of the OpenMP specifications must acknowledge the copyright by displaying the following statement: “OpenMP is a trademark of the OpenMP Architecture Review Board. Portions of this product/ publication may have been derived from the OpenMP Language Application Program Interface Specification.” ®
