SHMEM_GET128_NBI(3)
π SHMEM_GET128_NBI(3) β ok bestie, sit DOWN π
Hiii welcome to the <SHMEM_GET128_NBI(3)> glow-up page from the Open MPI manual set. This is the nonblocking get family and it's giving β¨ fetch quest energy β¨ β you're grabbing data from somebody else's PE and dropping it into your own, and you don't even wait around for it to land. Iconic behaviour honestly.
π NAME π
Okay so this page covers a whole friend group of routines, not just one:
shmem_getmem_nbi(3)shmem_char_get_nbi(3)shmem_short_get_nbi(3)shmem_int_get_nbi(3)shmem_long_get_nbi(3)shmem_longlong_get_nbi(3)shmem_float_get_nbi(3)shmem_double_get_nbi(3)shmem_longdouble_get_nbi(3)shmem_get8_nbi(3)shmem_get16_nbi(3)shmem_get32_nbi(3)shmem_get64_nbi(3)shmem_get128_nbi(3)
And what do they all do, you ask? The nonblocking get routines provide a method for copying data from a contiguous remote data object on the specified PE to the local data object. Simple as that. π¦
β¨ SYNOPSIS β¨
C or C++
#include <mpp/shmem.h>
void shmem_getmem_nbi(void *dest, const void *source,
size_t nelems, int pe)
void shmem_char_get(char *dest, const char *source,
size_t nelems, int pe)
void shmem_short_get(short *dest, const short *source,
size_t nelems, int pe)
void shmem_int_get(int *dest, const int *source,
size_t nelems, int pe)
void shmem_long_get(long *dest, const long *source,
size_t nelems, int pe)
void shmem_longlong_get(long long *dest, const long long *source,
size_t nelems, int pe)
void shmem_float_get(float *dest, const float *source,
size_t nelems, int pe)
void shmem_double_get(double *dest, const double *source,
size_t nelems, int pe)
void shmem_longdouble_get(long double *dest, const long double *source,
size_t nelems, int pe)
void shmem_get8(void *dest, const void *source,
size_t nelems, int pe)
void shmem_get16(void *dest, const void *source,
size_t nelems, int pe)
void shmem_get32(void *dest, const void *source,
size_t nelems, int pe)
void shmem_get64(void *dest, const void *source,
size_t nelems, int pe)
void shmem_get128(void *dest, const void *source,
size_t nelems, int pe)
no cap, copy those signatures character for character β don't be out here "fixing" them, they're serving exactly as intended π
Fortran
INCLUDE "mpp/shmem.fh"
INTEGER nelems, pe
CALL SHMEM_GETMEM_NBI(dest, source, nelems, pe)
CALL SHMEM_CHARACTER_GET_NBI(dest, source, nelems, pe)
CALL SHMEM_COMPLEX_GET_NBI(dest, source, nelems, pe)
CALL SHMEM_DOUBLE_GET_NBI(dest, source, nelems, pe)
CALL SHMEM_INTEGER_GET_NBI(dest, source, nelems, pe)
CALL SHMEM_LOGICAL_GET_NBI(dest, source, nelems, pe)
CALL SHMEM_REAL_GET_NBI(dest, source, nelems, pe)
CALL SHMEM_GET4_NBI(dest, source, nelems, pe)
CALL SHMEM_GET8_NBI(dest, source, nelems, pe)
CALL SHMEM_GET32_NBI(dest, source, nelems, pe)
CALL SHMEM_GET64_NBI(dest, source, nelems, pe)
CALL SHMEM_GET128_NBI(dest, source, nelems, pe)
πΈ DESCRIPTION πΈ
The get routines provide a method for copying a contiguous symmetric data object from a different PE to a contiguous data object on the local PE. Here's the tea though: the routines return after posting the operation. So no, bestie, you do not have your data yet. The operation is considered complete after a subsequent call to shmem_quiet. At the completion of shmem_quiet, the data has been delivered to the dest array on the local PE. That's the whole moment. π
the arguments, aka the cast of characters π«Ά
destβ Local data object to be updated. This is your lil container, the one catching feelings (and bytes).sourceβ Data object on the PE identified bypethat contains the data to be copied. This data object must be remotely accessible. Yes, that's a real requirement, we don't make the rules.nelemsβ Number of elements in the target and source arrays.lenmust be of type integer. If you are using Fortran, it must be a constant, variable, or array element of default integer type.peβ PE number of the remote PE.pemust be of type integer. If you are using Fortran, it must be a constant, variable, or array element of default integer type.
Fortran girlies listen UP: if you are using Fortran, data types must be of default size. For example, a real variable must be declared asREAL,REAL*4, orREAL(KIND=4). Do not freestyle this. π
π NOTES π
See intro_shmem(3) for a definition of the term remotely accessible. If you're sitting there like "what does remotely accessible even mean" β that page has you. Go read it, it's literally one click away (well, one manpage away). π¦
π₯ EXAMPLES π₯
Okay consider this simple example for Fortran. It's cute, it's classic, it's the vibe check:
PROGRAM REDUCTION
REAL VALUES, SUM
COMMON /C/ VALUES
REAL WORK
CALL START_PES(0) ! ALLOW ANY NUMBER OF PES
VALUES = MY_PE() ! INITIALIZE IT TO SOMETHING
CALL SHMEM_BARRIER_ALL
SUM = 0.0
DO I = 0,NUM_PES()-1
CALL SHMEM_REAL_GET_NBI(WORK, VALUES, 1, I)
CALL SHMEM_QUIET ! wait for delivery
SUM = SUM + WORK
ENDDO
PRINT *, 'PE ', MY_PE(), ' COMPUTED SUM=', SUM
CALL SHMEM_BARRIER_ALL
END
Notice how SHMEM_QUIET is the one doing the actual waiting β SHMEM_REAL_GET_NBI just posts and dips. That's the nonblocking lifestyle, and honestly? Main character energy. π
See also:
π Copyright π
2003-2026, The Open MPI Community
And that's a wrap on <SHMEM_GET128_NBI(3)> from Open MPI π β this lil page was living its best life back on Sep 10, 2026, and the copyright belongs to the Open MPI Community, 2003-2026. Go forth, post your gets, call your shmem_quiet, and slay. πβ¨ππ¦
π¦ where this page comes from
- ubuntu stonking: openmpi-doc 5.0.10-6 Β· other
- π girlified by deepseek-ai/DeepSeek-V4.1-Flash on : a language model's rewrite of the original, under its terms