openmpi/oshmem/shmem/man/man3/intro_shmem.3in

.\" -*- nroff -*-
.\" Copyright (c) 2015      University of Houston.  All rights reserved.
.\" Copyright (c) 2015      Mellanox Technologies, Inc.
.\" $COPYRIGHT$
.de Vb
.ft CW
.nf
..
.de Ve
.ft R

.fi
..
.TH "INTRO\\_SHMEM" "3" "#OMPI_DATE#" "#PACKAGE_VERSION#" "#PACKAGE_NAME#"
.SH NAME

intro_shmem \- Introduction to the OpenSHMEM programming model
.PP
.SH DESCRIPTION

The SHMEM programming model consists of library routines that provide low\-latency,
high\-bandwidth communication for use in highly parallelized scalable programs. The
routines in the OpenSHMEM application programming interface (API) provide a programming
model for exchanging data between cooperating parallel processes. The resulting programs
are similar in style to Message Passing Interface (MPI) programs. The SHMEM API can
be used either alone or in combination with MPI routines in the same parallel program.
.PP
An OpenSHMEM program is SPMD (single program, multiple data) in style. The SHMEM
processes, called processing elements or PEs, all start at the same time and they all run the
same program. Usually the PEs perform computation on their own subdomains of the larger
problem and periodically communicate with other PEs to exchange information on
which the next computation phase depends.
.PP
The OpenSHMEM routines minimize the overhead associated with data transfer requests,
maximize bandwidth and minimize data latency. Data latency is the period of time that
starts when a PE initiates a transfer of data and ends when a PE can use the data.
OpenSHMEM routines support remote data transfer through put operations, which transfer
data to a different PE, get operations, which transfer data from a different PE, and remote
pointers, which allow direct references to data objects owned by another PE. Other
operations supported are collective broadcast and reduction, barrier synchronization, and
atomic memory operations. An atomic memory operation is an atomic read\-and\-update
operation, such as a fetch\-and\-increment, on a remote or local data object.
.PP
.SH OPENSHMEM ROUTINES

This section lists the significant OpenSHMEM message\-passing routines.
.TP
PE queries
.PP
.RS
.TP
.B *
C/C++ only:
.RS
.PP
.RS
.RE
.TP
.B *
\fI_num_pes\fP(3)
.TP
.B *
\fI_my_pe\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
Fortran only:
.RS
.PP
.RS
.RE
.TP
.B *
\fINUM_PES\fP(3)
.TP
.B *
\fIMY_PE\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Elemental data put routines
.PP
.RS
.TP
.B *
C/C++ only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_double_p\fP(3)
.TP
.B *
\fIshmem_float_p\fP(3)
.TP
.B *
\fIshmem_int_p\fP(3)
.TP
.B *
\fIshmem_long_p\fP(3)
.TP
.B *
\fIshmem_short_p.\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Block data put routines
.PP
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_put32\fP(3)
.TP
.B *
\fIshmem_put64\fP(3)
.TP
.B *
\fIshmem_put128\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
C/C++ only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_double_put\fP(3)
.TP
.B *
\fIshmem_float_put\fP(3)
.TP
.B *
\fIshmem_int_put\fP(3)
.TP
.B *
\fIshmem_long_put\fP(3)
.TP
.B *
\fIshmem_short_put.\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
Fortran only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_complex_put\fP(3)
.TP
.B *
\fIshmem_integer_put\fP(3)
.TP
.B *
\fIshmem_logical_put\fP(3)
.TP
.B *
\fIshmem_real_put\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Elemental data get routines
.PP
.RS
.TP
.B *
C/C++ only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_double_g\fP(3)
.TP
.B *
\fIshmem_float_g\fP(3)
.TP
.B *
\fIshmem_int_g\fP(3)
.TP
.B *
\fIshmem_long_g\fP(3)
.TP
.B *
\fIshmem_short_g\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Block data get routines
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_get32\fP(3)
.TP
.B *
\fIshmem_get64\fP(3)
.TP
.B *
\fIshmem_get128\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
C/C++ only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_double_get\fP(3)
.TP
.B *
\fIshmem_float_get\fP(3)
.TP
.B *
\fIshmem_int_get\fP(3)
.TP
.B *
\fIshmem_long_get\fP(3)
.TP
.B *
\fIshmem_short_get\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
Fortran only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_complex_get\fP(3)
.TP
.B *
\fIshmem_integer_get\fP(3)
.TP
.B *
\fIshmem_logical_get\fP(3)
.TP
.B *
\fIshmem_real_get\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Strided put routines
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_iput32\fP(3)
.TP
.B *
\fIshmem_iput64\fP(3)
.TP
.B *
\fIshmem_iput128\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
C/C++ only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_double_iput\fP(3)
.TP
.B *
\fIshmem_float_iput\fP(3)
.TP
.B *
\fIshmem_int_iput\fP(3)
.TP
.B *
\fIshmem_long_iput\fP(3)
.TP
.B *
\fIshmem_short_iput\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
Fortran only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_complex_iput\fP(3)
.TP
.B *
\fIshmem_integer_iput\fP(3)
.TP
.B *
\fIshmem_logical_iput\fP(3)
.TP
.B *
\fIshmem_real_iput\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Strided get routines
.PP
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_iget32\fP(3)
.TP
.B *
\fIshmem_iget64\fP(3)
.TP
.B *
\fIshmem_iget128\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
C/C++ only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_double_iget\fP(3)
.TP
.B *
\fIshmem_float_iget\fP(3)
.TP
.B *
\fIshmem_int_iget\fP(3)
.TP
.B *
\fIshmem_long_iget\fP(3)
.TP
.B *
\fIshmem_short_iget\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
Fortran only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_complex_iget\fP(3)
.TP
.B *
\fIshmem_integer_iget\fP(3)
.TP
.B *
\fIshmem_logical_iget\fP(3)
.TP
.B *
\fIshmem_real_iget\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Point\-to\-point synchronization routines
.RS
.TP
.B *
C/C++ only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_int_wait\fP(3)
.TP
.B *
\fIshmem_int_wait_until\fP(3)
.TP
.B *
\fIshmem_long_wait\fP(3)
.TP
.B *
\fIshmem_long_wait_until\fP(3)
.TP
.B *
\fIshmem_longlong_wait\fP(3)
.TP
.B *
\fIshmem_longlong_wait_until\fP(3)
.TP
.B *
\fIshmem_short_wait\fP(3)
.TP
.B *
\fIshmem_short_wait_until\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
Fortran:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_int4_wait\fP(3)
.TP
.B *
\fIshmem_int4_wait_until\fP(3)
.TP
.B *
\fIshmem_int8_wait\fP(3)
.TP
.B *
\fIshmem_int8_wait_until\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Barrier synchronization routines
.PP
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_barrier_all\fP(3)
.TP
.B *
\fIshmem_barrier\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Atomic memory fetch\-and\-operate (fetch\-op) routines
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.TP
.B *
shmem_swap
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Reduction routines
.RS
.TP
.B *
C/C++ only:
.RS
.TP
.B *
\fIshmem_int_and_to_all\fP(3)
.TP
.B *
\fIshmem_long_and_to_all\fP(3)
.TP
.B *
\fIshmem_longlong_and_to_all\fP(3)
.TP
.B *
\fIshmem_short_and_to_all\fP(3)
.TP
.B *
\fIshmem_double_max_to_all\fP(3)
.TP
.B *
\fIshmem_float_max_to_all\fP(3)
.TP
.B *
\fIshmem_int_max_to_all\fP(3)
.TP
.B *
\fIshmem_long_max_to_all\fP(3)
.TP
.B *
\fIshmem_longlong_max_to_all\fP(3)
.TP
.B *
\fIshmem_short_max_to_all\fP(3)
.TP
.B *
\fIshmem_double_min_to_all\fP(3)
.TP
.B *
\fIshmem_float_min_to_all\fP(3)
.TP
.B *
\fIshmem_int_min_to_all\fP(3)
.TP
.B *
\fIshmem_long_min_to_all\fP(3)
.TP
.B *
\fIshmem_longlong_min_to_all\fP(3)
.TP
.B *
\fIshmem_short_min_to_all\fP(3)
.TP
.B *
\fIshmem_double_sum_to_all\fP(3)
.TP
.B *
\fIshmem_float_sum_to_all\fP(3)
.TP
.B *
\fIshmem_int_sum_to_all\fP(3)
.TP
.B *
\fIshmem_long_sum_to_all\fP(3)
.TP
.B *
\fIshmem_longlong_sum_to_all\fP(3)
.TP
.B *
\fIshmem_short_sum_to_all\fP(3)
.TP
.B *
\fIshmem_double_prod_to_all\fP(3)
.TP
.B *
\fIshmem_float_prod_to_all\fP(3)
.TP
.B *
\fIshmem_int_prod_to_all\fP(3)
.TP
.B *
\fIshmem_long_prod_to_all\fP(3)
.TP
.B *
\fIshmem_longlong_prod_to_all\fP(3)
.TP
.B *
\fIshmem_short_prod_to_all\fP(3)
.TP
.B *
\fIshmem_int_or_to_all\fP(3)
.TP
.B *
\fIshmem_long_or_to_all\fP(3)
.TP
.B *
\fIshmem_longlong_or_to_all\fP(3)
.TP
.B *
\fIshmem_short_or_to_all\fP(3)
.TP
.B *
\fIshmem_int_xor_to_all\fP(3)
.TP
.B *
\fIshmem_long_xor_to_all\fP(3)
.TP
.B *
\fIshmem_longlong_xor_to_all\fP(3)
.TP
.B *
\fIshmem_short_xor_to_all\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
Fortran only:
.RS
.TP
.B *
\fIshmem_int4_and_to_all\fP(3)
.TP
.B *
\fIshmem_int8_and_to_all\fP(3)
.TP
.B *
\fIshmem_real4_max_to_all\fP(3)
.TP
.B *
\fIshmem_real8_max_to_all\fP(3)
.TP
.B *
\fIshmem_int4_max_to_all\fP(3)
.TP
.B *
\fIshmem_int8_max_to_all\fP(3)
.TP
.B *
\fIshmem_real4_min_to_all\fP(3)
.TP
.B *
\fIshmem_real8_min_to_all\fP(3)
.TP
.B *
\fIshmem_int4_min_to_all\fP(3)
.TP
.B *
\fIshmem_int8_min_to_all\fP(3)
.TP
.B *
\fIshmem_real4_sum_to_all\fP(3)
.TP
.B *
\fIshmem_real8_sum_to_all\fP(3)
.TP
.B *
\fIshmem_int4_sum_to_all\fP(3)
.TP
.B *
\fIshmem_int8_sum_to_all\fP(3)
.TP
.B *
\fIshmem_real4_prod_to_all\fP(3)
.TP
.B *
\fIshmem_real8_prod_to_all\fP(3)
.TP
.B *
\fIshmem_int4_prod_to_all\fP(3)
.TP
.B *
\fIshmem_int8_prod_to_all\fP(3)
.TP
.B *
\fIshmem_int4_or_to_all\fP(3)
.TP
.B *
\fIshmem_int8_or_to_all\fP(3)
.TP
.B *
\fIshmem_int4_xor_to_all\fP(3)
.TP
.B *
\fIshmem_int8_xor_to_all\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Broadcast routines
.PP
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_broadcast32\fP(3)
.TP
.B *
\fIshmem_broadcast64\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Cache management routines
.PP
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_udcflush\fP(3)
.TP
.B *
\fIshmem_udcflush_line\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Byte\-granularity block put routines
.PP
.RS
.TP
.B *
C/C++ and Fortran
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_putmem\fP(3)
.TP
.B *
\fIshmem_getmem\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
Fortran only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_character_put\fP(3)
.TP
.B *
\fIshmem_character_get\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Collect routines
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_collect32\fP(3)
.TP
.B *
\fIshmem_collect64\fP(3)
.TP
.B *
\fIshmem_fcollect32\fP(3)
.TP
.B *
\fIshmem_fcollect64\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Atomic memory fetch\-and\-operate (fetch\-op) routines
.RS
.TP
.B *
C/C++ only:
.RS
.TP
.B *
\fIshmem_double_swap\fP(3)
.TP
.B *
\fIshmem_float_swap\fP(3)
.TP
.B *
\fIshmem_int_cswap\fP(3)
.TP
.B *
\fIshmem_int_fadd\fP(3)
.TP
.B *
\fIshmem_int_finc\fP(3)
.TP
.B *
\fIshmem_int_swap\fP(3)
.TP
.B *
\fIshmem_long_cswap\fP(3)
.TP
.B *
\fIshmem_long_fadd\fP(3)
.TP
.B *
\fIshmem_long_finc\fP(3)
.TP
.B *
\fIshmem_long_swap\fP(3)
.TP
.B *
\fIshmem_longlong_cswap\fP(3)
.TP
.B *
\fIshmem_longlong_fadd\fP(3)
.TP
.B *
\fIshmem_longlong_finc\fP(3)
.TP
.B *
\fIshmem_longlong_swap\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
Fortran only:
.RS
.TP
.B *
\fIshmem_int4_cswap\fP(3)
.TP
.B *
\fIshmem_int4_fadd\fP(3)
.TP
.B *
\fIshmem_int4_finc\fP(3)
.TP
.B *
\fIshmem_int4_swap\fP(3)
.TP
.B *
\fIshmem_int8_swap\fP(3)
.TP
.B *
\fIshmem_real4_swap\fP(3)
.TP
.B *
\fIshmem_real8_swap\fP(3)
.TP
.B *
\fIshmem_int8_cswap\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Atomic memory operation routines
.RS
.TP
.B *
Fortran only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_int4_add\fP(3)
.TP
.B *
\fIshmem_int4_inc\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Remote memory pointer function
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_ptr\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Reduction routines
.RS
.TP
.B *
C/C++ only:
.RS
.TP
.B *
\fIshmem_longdouble_max_to_all\fP(3)
.TP
.B *
\fIshmem_longdouble_min_to_all\fP(3)
.TP
.B *
\fIshmem_longdouble_prod_to_all\fP(3)
.TP
.B *
\fIshmem_longdouble_sum_to_all\fP(3)
.RE
.RS
.PP
.RE
.TP
.B *
Fortran only:
.RS
.PP
.RS
.RE
.TP
.B *
\fIshmem_real16_max_to_all\fP(3)
.TP
.B *
\fIshmem_real16_min_to_all\fP(3)
.TP
.B *
\fIshmem_real16_prod_to_all\fP(3)
.TP
.B *
\fIshmem_real16_sum_to_all\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Accessibility query routines
.RS
.TP
.B *
C/C++ and Fortran:
.RS
.TP
.B *
\fIshmem_pe_accessible\fP(3)
.TP
.B *
\fIshmem_addr_accessible\fP(3)
.RE
.RS
.PP
.RE
.RE
.PP
.RE
.TP
Symmetric Data Objects
.PP
Consistent with the SPMD nature of the OpenSHMEM programming model is the
concept of symmetric data objects. These are arrays or variables that
exist with the same size, type, and relative address on all PEs.
Another term for symmetric data objects is "remotely accessible data objects".
In the interface definitions for OpenSHMEM data transfer routines, one or more of the
parameters are typically required to be symmetric or remotely accessible.
.PP
The following kinds of data objects are symmetric:
.RS
.TP
.B *
Fortran data objects in common blocks or with the SAVE attribute. These data
objects must not be defined in a dynamic shared object (DSO).
.TP
.B *
Non\-stack C and C++ variables. These data objects must not be defined in a DSO.
.TP
.B *
Fortran arrays allocated with \fIshpalloc\fP(3F)
.TP
.B *
C and C++ data allocated by \fIshmalloc\fP(3C)
.RE
.RS
.PP
.RE
.TP
Collective Routines
Some SHMEM routines, for example, \fIshmem_broadcast\fP(3)
and
\fIshmem_float_sum_to_all\fP(3),
are classified as collective routines
because they distribute work across a set of PEs.
They must be called concurrently by all PEs in the active set defined by the PE_start,
logPE_stride, PE_size argument triplet. The following man pages describe the OpenSHMEM
collective routines:
.RS
.TP
.B *
\fIshmem_and\fP(3)
.TP
.B *
\fIshmem_barrier\fP(3)
.TP
.B *
\fIshmem_broadcast\fP(3)
.TP
.B *
\fIshmem_collect\fP(3)
.TP
.B *
\fIshmem_max\fP(3)
.TP
.B *
\fIshmem_min\fP(3)
.TP
.B *
\fIshmem_or\fP(3)
.TP
.B *
\fIshmem_prod\fP(3)
.TP
.B *
\fIshmem_sum\fP(3)
.TP
.B *
\fIshmem_xor\fP(3)
.RE
.RS
.PP
.RE
.PP
.SH USING THE SYMMETRIC WORK ARRAY, PSYNC

Multiple pSync arrays are often needed if a particular PE calls as OpenSHMEM collective
routine twice without intervening barrier synchronization. Problems would occur if some PEs
in the active set for call 2 arrive at call 2 before processing of call 1 is complete by all PEs in
the call 1 active set. You can use \fIshmem_barrier\fP(3)
or \fIshmem_barrier_all\fP(3)
to perform a barrier synchronization between consecutive calls to OpenSHMEM collective
routines.
.PP
There are two special cases:
.RE
.TP
.B *
The \fIshmem_barrier\fP(3) routine allows the same pSync array to be used on
consecutive calls as long as the active PE set does not change.
.TP
.B *
If the same collective routine is called multiple times with the same active set, the
calls may alternate between two pSync arrays. The SHMEM routines guarantee that a
first call is completely finished by all PEs by the time processing of a third call begins
on any PE.
.PP
Because the SHMEM routines restore pSync to its original contents, multiple calls that
use the same pSync array do not require that pSync be reinitialized after the first call.
.PP
.SH SHMEM ENVIRONMENT VARIABLES

This section lists the significant SHMEM environment variables.
.TP
.B *
\fBSMA_VERSION\fP print the library version at start\-up.
.TP
.B *
\fBSMA_INFO\fP print helpful text about all these environment variables.
.TP
.B *
\fBSMA_SYMMETRIC_SIZE\fP number of bytes to allocate for the symmetric heap.
.TP
.B *
\fBSMA_DEBUG\fP enable debugging messages.
.PP
The first call to SHMEM must be \fIstart_pes\fP(3)\&.
This routines initialize the SHMEM runtime.
.PP
Calling any other SHMEM routines beforehand has undefined behavior. Multiple calls
to this routine is not allowed.
.PP
.SH COMPILING AND RUNNING OPENSHMEM PROGRAMS

The OpenSHMEM specification is silent regarding how OpenSHMEM programs are compiled,
linked and run. This section shows some examples of how wrapper programs could be utilized
to compile and launch applications. The commands are styled after wrapper programs
found in many MPI implementations.
.PP
The following sample command line demonstrates running an OpenSHMEM Program using a wrapper script (\fBoshrun\fP
in this case):
.PP
.TP
.B *
C/C++:
.Vb
oshcc c_program.c
.Ve
.TP
.B *
FORTRAN:
.Vb
oshfort fortran_program.f
.Ve
.PP
The following sample command line demonstrates running an OpenSHMEM Program assuming that the library provides a wrapper script for such purpose
(named \fBoshrun\fP
for this example):
.PP
.Vb
oshrun \-np 32 ./a.out
.Ve
.PP
.SH EXAMPLES

\fBExample 1\fP:
The following Fortran OpenSHMEM program directs all PEs to sum
simultaneously the numbers in the VALUES variable across all PEs:
.Vb
PROGRAM REDUCTION
  REAL VALUES, SUM
  COMMON /C/ VALUES
  REAL WORK

  CALL START_PES(0)
  VALUES = MY_PE()
  CALL SHMEM_BARRIER_ALL ! Synchronize all PEs
  SUM = 0.0
  DO I = 0, NUM_PES()\-1
    CALL SHMEM_REAL_GET(WORK, VALUES, 1, I) ! Get next value
    SUM = SUM + WORK                ! Sum it
  ENDDO
  PRINT *, 'PE ', MY_PE(), ' COMPUTED SUM=', SUM
  CALL SHMEM_BARRIER_ALL
END
.Ve
\fBExample 2\fP:
The following C OpenSHMEM program transfers an array of 10 longs from
PE 0 to PE 1:
.Vb
#include <mpp/shmem.h>

main() {
  long source[10] = { 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 };
  static long target[10];

  shmem_init();
  if (shmem_my_pe() == 0) {
    /* put 10 elements into target on PE 1 */
    shmem_long_put(target, source, 10, 1);
  }
  shmem_barrier_all(); /* sync sender and receiver */
  if (shmem_my_pe() == 1)
    printf("target[0] on PE %d is %d\\n", shmem_my_pe(), target[0]);
}
.Ve
.SH SEE ALSO

The following man pages also contain information on OpenSHMEM routines. See the
specific man pages for implementation information.
.PP
\fIshmem_add\fP(3),
\fIshmem_and\fP(3),
\fIshmem_barrier\fP(3),
\fIshmem_barrier_all\fP(3),
\fIshmem_broadcast\fP(3),
\fIshmem_cache\fP(3),
\fIshmem_collect\fP(3),
\fIshmem_cswap\fP(3),
\fIshmem_fadd\fP(3),
\fIshmem_fence\fP(3),
\fIshmem_finc\fP(3),
\fIshmem_get\fP(3),
\fIshmem_iget\fP(3),
\fIshmem_inc\fP(3),
\fIshmem_iput\fP(3),
\fIshmem_lock\fP(3),
\fIshmem_max\fP(3),
\fIshmem_min\fP(3),
\fIshmem_my_pe\fP(3),
\fIshmem_or\fP(3),
\fIshmem_prod\fP(3),
\fIshmem_put\fP(3),
\fIshmem_quiet\fP(3),
\fIshmem_short_g\fP(3),
\fIshmem_short_p\fP(3),
\fIshmem_sum\fP(3),
\fIshmem_swap\fP(3),
\fIshmem_wait\fP(3),
\fIshmem_xor\fP(3),
\fIshmem_pe_accessible\fP(3),
\fIshmem_addr_accessible\fP(3),
\fIshmem_init\fP(3),
\fIshmem_malloc\fP(3C),
\fIshmem_my_pe\fP(3I),
\fIshmem_n_pes\fP(3I)