SHMEM_LONGLONG_XOR_TO_ALL(3)

📦
version 5.0.10 · 4 versions
, in ubuntu stonking

💖 SHMEM_LONGLONG_XOR_TO_ALL 💖

okay bestie, sit down, hydrate, because we're about to talk about SHMEM_LONGLONG_XOR_TO_ALL(3) from Open MPI and honestly? it's giving group project but make it parallel. ✨

This one page is the home base for a whole little friend group of routines, all doing the same bitwise XOR thing:

and what do they all do? they perform a bitwise XOR operation on symmetric arrays over the active set of PEs. periodt. 💅

✨ SYNOPSIS ✨

C or C++ 🫶

#include <mpp/shmem.h>

void shmem_int_xor_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_xor_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_xor_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_xor_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_COMP4_XOR_TO_ALL(target, source, nreduce,
& PE_start, logPE_stride, PE_size, pWrk, pSync)

CALL SHMEM_INT4_XOR_TO_ALL(target, source, nreduce,
& PE_start, logPE_stride, PE_size, pWrk, pSync)

CALL SHMEM_INT8_XOR_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 literally just an associative binary operation run across a set of values — no cap. want the full roster of other SHMEM reduction routines? go peek at intro_shmem(3). 💜

and like every other SHMEM collective routine, each of these assumes that only PEs in the active set call the routine. if a PE that is not in the active set calls a SHMEM collective routine, you get undefined behavior. don't do that. that's not the vibe. 💀

the nreduce argument decides how many separate reductions you're doing. the source array on all PEs in the active set hands over one element per reduction. the results land in the target array on all PEs in the active set. and the active set itself? defined by the PE_start, logPE_stride, PE_size triplet, bestie. that's the trio. 🎀

source and target can be the same array — iconic, we love an in-place moment — but they may not be overlapping arrays. 🚫

okay now the arguments, one by one:

  • target — a symmetric array of length nreduce elements that receives the results of the reduction operations. the data type of target varies depending on which version of the reduction routine you're calling and which language you're in. calling from C/C++? check the SYNOPSIS section for the deets. calling from Fortran? the target data types are:
    • shmem_comp8_xor_to_all:** Complex, with an element size equal to two 8-byte real values
    • shmem_comp4_xor_to_all:** Complex, with an element size equal to two 4-byte real values
    • shmem_int8_xor_to_all:** Integer, with an element size of 8 bytes
    • shmem_int4_xor_to_all:** Integer, with an element size of 4 bytes
    • shmem_real8_xor_to_all:** Real, with an element size of 8 bytes
    • shmem_real4_xor_to_all:** Real, with an element size of 4 bytes
  • source — a symmetric array, of length nreduce elements, holding one element for each separate reduction operation. source must have the same data type as target. matching fits, periodt. ✨
  • nreduce — the number of elements in the target and source arrays. must be of type integer. if you're in Fortran, it has to be a default integer value. no exceptions bestie.
  • PE_start — the lowest virtual PE number of the active set of PEs. must be of type integer. Fortran? default integer, again.
  • logPE_stride — the log (base 2) of the stride between consecutive virtual PE numbers in the active set. must be of type integer. Fortran folks: default integer. yes, log base 2, don't come at me.
  • PE_size — the number of PEs in the active set. must be of type integer. in Fortran it must be a default integer value.
  • pWrk — a symmetric work array. pWrk must have the same data type as target. in C/C++, it contains max(nreduce/2 + 1, _SHMEM_REDUCE_MIN_WRKDATA_SIZE) elements. in Fortran, it contains max(nreduce/2 + 1, SHMEM_REDUCE_MIN_WRKDATA_SIZE) elements. note the underscore situation, it matters. 👑
  • 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. if you're using Fortran, it must be a default integer value. and every single 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. do not skip this step, I am begging. 🙈

the values of nreduce, PE_start, logPE_stride, and PE_size must all be equal on all PEs in the active set. same target and source arrays, and same pWrk and pSync work arrays, passed to every PE in the active set. everyone matching, everyone coordinating. it's giving choreography. 🧚

before any PE calls a reduction routine, you've gotta make sure these conditions exist (synchronization via a barrier or some other method is often needed to make it happen, no shortcuts):

  • 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.

and once a reduction routine returns, here's what's true for the local PE:

  • the target array is updated. 💖
  • the values in the pSync array are restored to their original values. ✨
fr fr, read those two lists twice. collectives are not a "send it and hope" situation.

🌈 NOTES 🌈

the terms collective, symmetric, and cache aligned are defined over in intro_shmem(3). go read it, it's the 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. cute, right?

you must ensure that the pSync array is not being updated on any PE in the active set while any of the PEs are participating in the processing of a SHMEM reduction routine. be careful of 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 that 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, you can only be sure of that by doing some type of synchronization. however — special case incoming — if you're calling reduction routines with the same active set, you can allocate two pSync and pWrk arrays and alternate between them on successive calls. two buffers, tag team, no blocking. 💅

🎀 EXAMPLES 🎀

Example 1

this Fortran example statically initializes the pSync array and computes the exclusive OR of variable FOO across all even PEs. we love an even-numbered exclusive set. 👑

INCLUDE "mpp/shmem.fh"

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

IF ( MOD(MY_PE(),2) .EQ. 0) THEN
  CALL SHMEM_REAL8_XOR_TO_ALL(FOOXOR, FOO, NR, 0, 1, N$PES/2,
  & PWRK, PSYNC)
  PRINT *, 'Result on PE ', MY_PE(), ' is ', FOOXOR
ENDIF

Example 2

now consider this C/C++ call:

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

that call is more efficient, but semantically equivalent to, this whole combo:

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

notice how two sets of pWrk and pSync arrays get used alternately — because no synchronization is done between those calls. alternating buffers: the actual main character move. 🦋

🫶 SEE ALSO 🫶

2003-2026, The Open MPI Community

and that's the whole page bestie, you ate it. 💖 this was shmem_longlong_xor_to_all(3) from Open MPI, last touched Sep 10, 2026 — go forth, initialize your pSync, and slay. 🎀✨💅

📦 where this page comes from