MPI_STARTALL(3)
π MPI_STARTALL(3) π
okay bestie, this is a page from the Open MPI manual and it's all about kicking off a whole collection of requests at once. Group-chat energy on, brain on, let's go β¨
πΈ NAME πΈ
MPI_Startall β starts a whole collection of requests, all at once, no cap.
β¨ SYNTAX β¨
π C Syntax
#include <mpi.h>
int MPI_Startall(int count, MPI_Request array_of_requests[])
π Fortran Syntax
USE MPI
! or the older form: INCLUDE 'mpif.h'
MPI_STARTALL(COUNT, ARRAY_OF_REQUESTS, IERROR)
INTEGER COUNT, ARRAY_OF_REQUESTS(*), IERROR
π¦ Fortran 2008 Syntax
USE mpi_f08
MPI_Startall(count, array_of_requests, ierror)
INTEGER, INTENT(IN) :: count
TYPE(MPI_Request), INTENT(INOUT) :: array_of_requests(count)
INTEGER, OPTIONAL, INTENT(OUT) :: ierror
π₯ INPUT PARAMETER π₯
countβ the list length, and it's aninteger. Yes bestie, it's just how many we're dealing with πββοΈ
π INPUT/OUTPUT PARAMETER π
array_of_requestsβ an array of requests. Literally an array of handle, that's the whole vibe.
π€ OUTPUT PARAMETER π€
ierrorβ Fortran only: error status, and yes it's aninteger.
π DESCRIPTION π
So here's the tea β: MPI_Startall starts all the communications associated with the requests sitting in array_of_requests.
A call to MPI_Startall(count, array_of_requests) has the exact same effect as making a bunch of calls to MPI_Start (&array_of_requests[i]), executed for i=0 ,β¦, count-1, in some arbitrary order. Like, don't come for the ordering, okay? It's arbitrary. Periodt.
A communication that you started with either MPI_Start or MPI_Startall gets completed by a call to MPI_Wait, MPI_Test, or one of the derived besties MPI_Waitany, MPI_Testany, MPI_Waitall, MPI_Testall, MPI_Waitsome, MPI_Testsome. (These are all described in Section 3.7.5 of the MPI Standard β https://www.mpi-forum.org/docs/ β under βMultiple Completionsβ, iykyk π
.) Once such a call successfully completes it, the request becomes inactive. It is not deallocated though β nope β and it can absolutely be activated anew by another MPI_Start or MPI_Startall call. It's giving main character energy, coming back for round two π.
A persistent request gets deallocated by a call to MPI_Request_free (peep Section 3.7.3 of the MPI Standard β https://www.mpi-forum.org/docs/ β βCommunication Completionβ π«Ά). That MPI_Request_free call can happen at any point in the program after the persistent request was created. However, the request will only actually be deallocated after it becomes inactive.
π¨ bestie, listen: active receive requests should not be freed. Otherwise it will not be possible to check that the receive has completed. Not the move. π
It's preferable, in general, to free requests when they're inactive. If you follow this rule, then the persistent communication request functions will be invoked in a sequence of the form,
Create (Start Complete)* Free
where * indicates zero or more repetitions. And if the same communication object is used in several concurrent threads, then it is the user's responsibility to coordinate calls so that the correct sequence is obeyed. That one's on you, babes π.
A send operation initiated with MPI_Start can be matched with any receive operation, and likewise a receive operation initiated with MPI_Start can receive messages generated by any send operation. Slay, matching is flexible β.
π ERRORS π
Almost all MPI routines return an error value; C routines hand it back as the return result of the function, and Fortran routines put it in the last argument.
Before that error value is returned, the current MPI error handler associated with the communication object (e.g., communicator, window, file) gets called. If no communication object is associated with the MPI call, then the call is considered attached to MPI_COMM_SELF and will call the associated MPI error handler. When MPI_COMM_SELF is not initialized (i.e., before MPI_Init/MPI_Init_thread, after MPI_Finalize, or when using the Sessions Model exclusively), the error raises the initial error handler. The initial error handler can be changed by calling MPI_Comm_set_errhandler on MPI_COMM_SELF when using the World model, or the mpi_initial_errhandler CLI argument to mpiexec or info key to MPI_Comm_spawn/MPI_Comm_spawn_multiple. If no other appropriate error handler has been set, then the MPI_ERRORS_RETURN error handler is called for MPI I/O functions and the MPI_ERRORS_ABORT error handler is called for all other MPI functions.
Open MPI includes three predefined error handlers that can be used:
MPI_ERRORS_ARE_FATALβ causes the program to abort all connected MPI processes. π± dramatic, but make it finalized.MPI_ERRORS_ABORTβ an error handler that can be invoked on a communicator, window, file, or session. When called on a communicator, it acts as ifMPI_Abortwas called on that communicator. If called on a window or file, it acts as ifMPI_Abortwas called on a communicator containing the group of processes in the corresponding window or file. If called on a session, it aborts only the local process.MPI_ERRORS_RETURNβ returns an error code to the application.
MPI applications can also implement their own error handlers by calling:
MPI_Comm_create_errhandlerthenMPI_Comm_set_errhandlerMPI_File_create_errhandlerthenMPI_File_set_errhandlerMPI_Session_create_errhandlerthenMPI_Session_set_errhandleror atMPI_Session_initMPI_Win_create_errhandlerthenMPI_Win_set_errhandler
β οΈ Note that MPI does not guarantee that an MPI program can continue past an error. It's giving no promises, fr fr.
See the MPI man page for a full list of MPI error codes.
See the Error Handling section of the MPI-3.1 standard for more information.
See also:
MPI_Bsend_initMPI_Rsend_initMPI_Send_initMPI_Ssend_initMPI_Recv_initMPI_StartMPI_Request_free
π Copyright π
2003-2026, The Open MPI Community
and that's the whole page, bestie!! π written in Sep 10, 2026, and it's page MPI_STARTALL(3) from Open MPI β go start all your requests, you absolute icon πβ¨π
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