MPI_REQUEST_FREE(3)

📦
version 5.0.11 · 17 versions
, in arch rolling

💅 MPI_REQUEST_FREE 💅

okay bestie, so this is the Open MPI manual page for MPI_Request_free, section 3 of the manual — the one that's like "hey, I'm done with this request object, take it, I don't need it anymore" but WITHOUT waiting around for the actual communication to finish. main character energy, zero patience. 💨

🌸 NAME 🌸

MPI_Request_free — frees a communication request object. that's it. that's the whole job. freeing. deallocating. releasing it into the void. 🦋

✨ SYNTAX ✨

🎀 C Syntax

#include <mpi.h>

int MPI_Request_free(MPI_Request *request)

🎀 Fortran Syntax

USE MPI
! or the older form: INCLUDE 'mpif.h'
MPI_REQUEST_FREE(REQUEST, IERROR)
     INTEGER REQUEST, IERROR

🎀 Fortran 2008 Syntax

USE mpi_f08
MPI_Request_free(request, ierror)
     TYPE(MPI_Request), INTENT(INOUT) :: request
     INTEGER, OPTIONAL, INTENT(OUT) :: ierror

🫶 INPUT/OUTPUT PARAMETER 🫶

  • request — the communication request (handle). it's input/output because it literally gets set to MPI_REQUEST_NULL on the way out. she leaves and takes her identity with her. 💅

👑 DESCRIPTION 👑

okay so the whole point: this lets a request object get deallocated without waiting for the associated communication to complete. no waiting. no blocking. no sitting there staring at the terminal. ✨

MPI_Request_free marks the request object for deallocation and sets request to MPI_REQUEST_NULL. any ongoing communication that's associated with the request is allowed to complete — like, it gets to finish its little journey, no cap. but the request itself (the object) will only be deallocated after its completion. it's giving "you may leave the group chat but the messages stay". 😌

💀 NOTES 💀

okay this is the part where you pay attention, bestie, because there are real consequences here. 🚨

once a request is freed by a call to MPI_Request_free, it is not possible to check for the successful completion of the associated communication with calls to MPI_Wait or MPI_Test. gone. can't do it. and also — if an error occurs subsequently during the communication, an error code cannot be returned to the user — such an error must be treated as fatal. like, fatal fatal. 💀

so questions arise (yes, actual questions, this is literally in the manual): how does one know when the operations have completed when using MPI_Request_free? well, depending on the program logic, there may be other ways in which the program knows that certain operations have completed, and this is exactly what makes usage of MPI_Request_free practical. for example, an active send request could be freed when the logic of the program is such that the receiver sends a reply to the message sent — the arrival of the reply informs the sender that the send has completed and the send buffer can be reused. iconic indirect communication. 🎀

but an active receive request should never be freed, because the receiver will have no way to verify that the receive has completed and the receive buffer can be reused. that's a one-way ticket to sadness. 🙈

Example:

CALL MPI_COMM_RANK(MPI_COMM_WORLD, rank)
IF(rank == 0) THEN
    DO i=1, n
      CALL MPI_ISEND(outval, 1, MPI_REAL, 1, 0, req, ierr)
      CALL MPI_REQUEST_FREE(req, ierr)
      CALL MPI_IRECV(inval, 1, MPI_REAL, 1, 0, req, ierr)
      CALL MPI_WAIT(req, status, ierr)
    END DO
ELSE IF (rank == 1) THEN
    CALL MPI_IRECV(inval, 1, MPI_REAL, 0, 0, req, ierr)
    CALL MPI_WAIT(req, status)
    DO I=1, n-1
       CALL MPI_ISEND(outval, 1, MPI_REAL, 0, 0, req, ierr)
       CALL MPI_REQUEST_FREE(req, ierr)
       CALL MPI_IRECV(inval, 1, MPI_REAL, 0, 0, req, ierr)
       CALL MPI_WAIT(req, status, ierr)
    END DO
    CALL MPI_ISEND(outval, 1, MPI_REAL, 0, 0, req, ierr)
    CALL MPI_WAIT(req, status)
END IF

so this routine is normally used to free persistent requests created with either MPI_Recv_init or MPI_Send_init and friends. but plot twist — it can be used to free a request created with MPI_Irecv or MPI_Isend and friends; in that case the use can not use the test/wait routines on the request. (yes, that's worded exactly like that in the original and we're keeping it fr fr.) 🧚

also it is permitted to free an active request. however, once freed, you can not use the request in a wait or test routine (e.g., MPI_Wait). 🔥

🌈 ERRORS 🌈

almost all MPI routines return an error value — C routines as the return result of the function, and Fortran routines in the last argument. classic. 💫

before the error value is returned, the current MPI error handler associated with the communication object (e.g., communicator, window, file) is 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_FATAL — causes the program to abort all connected MPI processes. all of them. the whole friend group. 😭
  • MPI_ERRORS_ABORT — an error handler that can be invoked on a communicator, window, file, or session. when called on a communicator, it acts as if MPI_Abort was called on that communicator. if called on a window or file, acts as if MPI_Abort was called on a communicator containing the group of processes in the corresponding window or file. if called on a session, aborts only the local process. very "your drama is your drama, not the group's". 💅
  • MPI_ERRORS_RETURN — returns an error code to the application. the polite one. 🎀

MPI applications can also implement their own error handlers by calling:

  • MPI_Comm_create_errhandler then MPI_Comm_set_errhandler
  • MPI_File_create_errhandler then MPI_File_set_errhandler
  • MPI_Session_create_errhandler then MPI_Session_set_errhandler or at MPI_Session_init
  • MPI_Win_create_errhandler then MPI_Win_set_errhandler

note that MPI does not guarantee that an MPI program can continue past an error. no promises. none. 🙃

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 💜

the whole squad, go say hi:

  • MPI_Isend
  • MPI_Irecv
  • MPI_Issend
  • MPI_Ibsend
  • MPI_Irsend
  • MPI_Recv_init
  • MPI_Send_init
  • MPI_Ssend_init
  • MPI_Rsend_init
  • MPI_Test
  • MPI_Wait
  • MPI_Waitall
  • MPI_Waitany
  • MPI_Waitsome
  • MPI_Testall
  • MPI_Testany
  • MPI_Testsome

2003-2026, The Open MPI Community — that's the source/version line, and the date stamp is Sep 11, 2026, on the MPI_REQUEST_FREE(3) page. so basically this whole vibe has been serving since 2003 and it's still going. slay, Open MPI Community, slay. 👑✨🦋

okay bye bestie, go free your requests responsibly, mwah 💖🎀

📦 where this page comes from