SHMEM_SHORT_OR_TO_ALL(3)

📦
version 5.0.10 · 7 versions
, in ubuntu stonking

💅 shmem_short_or_to_all(3) — Open MPI said gay rights (and bitwise OR) 💖

okay bestie sit DOWN because we're doing collective reductions today and honestly? it's giving teamwork. this is a page from Open MPI, and it's about a whole girl-group of routines that all do the same iconic thing: bitwise OR across a set of PEs. ✨

Here's the whole squad:

…and what do they do? They perform a bitwise OR function reduction across a set of processing elements (PEs). periodt. 🫶

✨ SYNOPSIS ✨

🎀 C or C++

#include <mpp/shmem.h>

void shmem_int_or_to_all(int *target, const int *source,
  int nreduce, int PE_start, int logPE_stride, int PE_size,
  int *pWrk, long *pSync)

void shmem_long_or_to_all(long *target, const long *source,
  int nreduce, int PE_start, int logPE_stride, int PE_size,
  long *pWrk, long *pSync)

void shmem_longlong_or_to_all(long long *target,
  const long long *source, int nreduce, int PE_start, int logPE_stride,
  int PE_size, long long *pWrk, long *pSync)

void shmem_short_or_to_all(short *target, const short *source,
  int nreduce, int PE_start, int logPE_stride, int PE_size,
  short *pWrk, long *pSync)

🌸 Fortran

INCLUDE "mpp/shmem.fh"

INTEGER pSync(SHMEM_REDUCE_SYNC_SIZE)
INTEGER nreduce, PE_start, logPE_stride, PE_size

CALL SHMEM_INT4_OR_TO_ALL(target, source, nreduce, PE_start,
& logPE_stride, PE_size, pWrk, pSync)

CALL SHMEM_INT8_OR_TO_ALL(target, source, nreduce, PE_start,
& logPE_stride, PE_size, pWrk, pSync)

💜 DESCRIPTION 💜

So. The shared memory (SHMEM) reduction routines compute one or more reductions across symmetric arrays on multiple virtual PEs. A reduction is just an associative binary operation performed across a set of values — like a group project but the group actually works. For a list of other SHMEM reduction routines, go peek at intro_shmem(3).

⚠️ like every SHMEM collective routine, this assumes ONLY the PEs in the active set call it. If a PE that is NOT in the active set calls a SHMEM collective routine, you get undefined behavior. not a vibe. 💀

The nreduce argument decides how many separate reductions we're doing. The source array on every PE in the active set hands over one element per reduction, and the results land in the target array on every PE in the active set. The active set is defined by the PE_start, logPE_stride, PE_size triplet. 👑

Also: source and target may be the same array, but they may not be overlapping arrays. no cap.

The arguments are as follows:

  • target — a symmetric array of length nreduce elements that receives the results of the reduction operations. The data type of target varies with which version of the reduction routine you're calling. From C/C++, check the SYNOPSIS for the data type. From Fortran, the target data types are:
    • shmem_int8_or_to_all — Integer, with an element size of 8 bytes.
    • shmem_int4_or_to_all — Integer, with an element size of 4 bytes.
  • source — a symmetric array of length nreduce elements, containing one element for each separate reduction operation. source must have the same data type as target.
  • nreduce — the number of elements in the target and source arrays. nreduce must be of type integer. If you're in Fortran, it must be a default integer value.
  • PE_start — the lowest virtual PE number of the active set of PEs. Must be of type integer. Fortran girlies: default integer value.
  • logPE_stride — the log (base 2) of the stride between consecutive virtual PE numbers in the active set. Must be of type integer. Fortran: default integer.
  • PE_size — the number of PEs in the active set. Must be of type integer. Fortran: default integer.
  • pWrk — a symmetric work array. It must have the same data type as target. In C/C++ it holds max(nreduce/2 + 1, _SHMEM_REDUCE_MIN_WRKDATA_SIZE) elements. In Fortran it holds max(nreduce/2 + 1, SHMEM_REDUCE_MIN_WRKDATA_SIZE) elements.
  • pSync — a symmetric work array. In C/C++, pSync is of type long and size _SHMEM_REDUCE_SYNC_SIZE. In Fortran, pSync is of type integer and size SHMEM_REDUCE_SYNC_SIZE, and if you're in Fortran it must be a default integer value. Every element of this array must be initialized with the value _SHMEM_SYNC_VALUE (C/C++) or SHMEM_SYNC_VALUE (Fortran) before any of the PEs in the active set enter the reduction routine. 🚨

The values of the arguments nreduce, PE_start, logPE_stride, and PE_size must be equal on all PEs in the active set. And the same target and source arrays, plus the same pWrk and pSync work arrays, must be passed to all PEs in the active set. y'all have to match. it's giving matching outfits. 💅

Before any PE calls a reduction routine, you must make sure these conditions exist (synchronization via a barrier or some other method is often needed to make it happen):

  • The pWrk and pSync arrays on all PEs in the active set are not still in use from a prior call to a collective SHMEM routine.
  • The target array on all PEs in the active set is ready to accept the results of the reduction.

Upon return from a reduction routine, the following are true:

  • The target array is updated. ✅
  • The values in the pSync array are restored to the original values. ✅

🦋 NOTES 🦋

The terms collective, symmetric, and cache aligned are defined in intro_shmem(3) — go read it, it's lore. All SHMEM reduction routines reset the values in pSync before they return, so a particular pSync buffer only needs to be initialized the first time it's used. iconic behavior honestly.

You must ensure the pSync array is not being updated on any PE in the active set while any of the PEs participate in processing of a SHMEM reduction routine. Be careful to avoid these situations:

  • If the pSync array is initialized at run time, some type of synchronization is needed to ensure all PEs in the working set have initialized pSync before any of them enter a SHMEM routine called with the pSync synchronization array.
  • A pSync or pWrk array can be reused in a subsequent reduction routine call only if none of the PEs in the active set are still processing a prior reduction routine call that used the same pSync or pWrk arrays.

In general, that can be assured only by doing some type of synchronization. However — special case alert 🚨 — when reduction routines are called with the same active set, you can allocate two pSync and pWrk arrays and alternate between them on successive calls. two outfits, alternate days. it's giving capsule wardrobe. ✨

🌈 EXAMPLES 🌈

Example 1: This Fortran example statically initializes the pSync array and finds the logical OR of the integer variable FOO across all even PEs. bestie the MY_PE() check is the whole plot twist.

INCLUDE "mpp/shmem.fh"

INTEGER PSYNC(SHMEM_REDUCE_SYNC_SIZE)
DATA PSYNC /SHMEM_REDUCE_SYNC_SIZE*SHMEM_SYNC_VALUE/
PARAMETER (NR=1)
REAL PWRK(MAX(NR/2+1,SHMEM_REDUCE_MIN_WRKDATA_SIZE))
INTEGER FOO, FOOOR
COMMON /COM/ FOO, FOOOR, PWRK
INTRINSIC MY_PE

IF ( MOD(MY_PE(),2) .EQ. 0) THEN
  CALL SHMEM_INT8_OR_TO_ALL(FOOOR, FOO, NR, 0, 1, N$PES/2,
  & PWRK, PSYNC)
  PRINT *,'Result on PE ',MY_PE(),' is ',FOOOR
ENDIF

Example 2: Consider the following C/C++ call:

shmem_int_or_to_all( target, source, 3, 0, 0, 8, pwrk, psync );

That preceding call is more efficient, but semantically equivalent to, the combination of the following calls:

shmem_int_or_to_all(&(target[0]), &(source[0]), 1, 0, 0, 8,
  pwrk1, psync1);
shmem_int_or_to_all(&(target[1]), &(source[1]), 1, 0, 0, 8,
  pwrk2, psync2);
shmem_int_or_to_all(&(target[2]), &(source[2]), 1, 0, 0, 8,
  pwrk1, psync1);

Note that two sets of pWrk and pSync arrays are used alternately because no synchronization is done between calls. see? alternating. told you. 🧚

👀 See also:

2003-2026, The Open MPI Community

and that's the whole page babes 💅 go forth and reduce, synchronize your pSync like your life depends on it (it kinda does), and remember: same arrays on every PE or it's over. this was SHMEM_SHORT_OR_TO_ALL(3) from Open MPI, dated Sep 10, 2026 — slay, stay symmetric, and hydrate. 🎀🦋✨

📦 where this page comes from