4.2.7. Blocking Send-Receive

PreviousUpNext
Up: Blocking Send and Receive Operations Next: Datatype Matching and Data Conversion Previous: Passing MPI_STATUS_IGNORE for Status

The send-receive operations combine in one operation the sending of a message to one destination and the receiving of another message, from another MPI process. The two (source and destination) are possibly the same. A send-receive operation is very useful for executing a shift operation across a chain of MPI processes. If blocking sends and receives are used for such a shift, then one needs to order the sends and receives correctly (for example, MPI processes with even rank in the communicator send, then receive, MPI processes with odd rank in the communicator receive first, then send) so as to prevent cyclic dependencies that may lead to deadlock. When a send-receive operation is used, the communication subsystem takes care of these issues. The send-receive operation can be used in conjunction with the procedures described in Chapter Virtual Topologies for MPI Processes in order to perform shifts on various logical topologies. Also, a send-receive operation is useful for implementing remote procedure calls.

A message sent by a send-receive operation can be received by a regular receive operation or probed by a probe operation; a send-receive operation can receive a message sent by a regular send operation.

MPI_SENDRECV(sendbuf, sendcount, sendtype, dest, sendtag, recvbuf, recvcount, recvtype, source, recvtag, comm, status)
IN sendbufinitial address of send buffer (choice)
IN sendcountnumber of elements in send buffer (non-negative integer)
IN sendtypetype of elements in send buffer (handle)
IN destrank of destination (integer)
IN sendtagsend tag (integer)
OUT recvbufinitial address of receive buffer (choice)
IN recvcountnumber of elements in receive buffer (non-negative integer)
IN recvtypetype of elements receive buffer element (handle)
IN sourcerank of source or MPI_ANY_SOURCE (integer)
IN recvtagreceive tag or MPI_ANY_TAG (integer)
IN commcommunicator (handle)
OUT statusstatus object (status)
C binding
int MPI_Sendrecv(const void *sendbuf, int sendcount, MPI_Datatype sendtype, int dest, int sendtag, void *recvbuf, int recvcount, MPI_Datatype recvtype, int source, int recvtag, MPI_Comm comm, MPI_Status *status)
int MPI_Sendrecv_c(const void *sendbuf, MPI_Count sendcount, MPI_Datatype sendtype, int dest, int sendtag, void *recvbuf, MPI_Count recvcount, MPI_Datatype recvtype, int source, int recvtag, MPI_Comm comm, MPI_Status *status)
Fortran 2008 binding
MPI_Sendrecv(sendbuf, sendcount, sendtype, dest, sendtag, recvbuf, recvcount, recvtype, source, recvtag, comm, status, ierror)

TYPE(*), DIMENSION(..), INTENT(IN) :: sendbuf
INTEGER, INTENT(IN) :: sendcount, dest, sendtag, recvcount, source, recvtag
TYPE(MPI_Datatype), INTENT(IN) :: sendtype, recvtype
TYPE(*), DIMENSION(..) :: recvbuf
TYPE(MPI_Comm), INTENT(IN) :: comm
TYPE(MPI_Status) :: status
INTEGER, OPTIONAL, INTENT(OUT) :: ierror
MPI_Sendrecv(sendbuf, sendcount, sendtype, dest, sendtag, recvbuf, recvcount, recvtype, source, recvtag, comm, status, ierror) !(_c)

TYPE(*), DIMENSION(..), INTENT(IN) :: sendbuf
INTEGER(KIND=MPI_COUNT_KIND), INTENT(IN) :: sendcount, recvcount
TYPE(MPI_Datatype), INTENT(IN) :: sendtype, recvtype
INTEGER, INTENT(IN) :: dest, sendtag, source, recvtag
TYPE(*), DIMENSION(..) :: recvbuf
TYPE(MPI_Comm), INTENT(IN) :: comm
TYPE(MPI_Status) :: status
INTEGER, OPTIONAL, INTENT(OUT) :: ierror
Fortran binding
MPI_SENDRECV(SENDBUF, SENDCOUNT, SENDTYPE, DEST, SENDTAG, RECVBUF, RECVCOUNT, RECVTYPE, SOURCE, RECVTAG, COMM, STATUS, IERROR)

<type> SENDBUF(*), RECVBUF(*)
INTEGER SENDCOUNT, SENDTYPE, DEST, SENDTAG, RECVCOUNT, RECVTYPE, SOURCE, RECVTAG, COMM, STATUS(MPI_STATUS_SIZE), IERROR

Execute a blocking send-receive operation. Both send and receive use the same communicator, but possibly different tags. The send buffer and receive buffers must be disjoint, and may have different lengths and datatypes.

The semantics of a send-receive operation is what would be obtained if the caller forked two concurrent threads, one to execute the send, and one to execute the receive, followed by a join of these two threads.

MPI_SENDRECV_REPLACE(buf, count, datatype, dest, sendtag, source, recvtag, comm, status)
INOUT bufinitial address of send and receive buffer (choice)
IN countnumber of elements in send and receive buffer (non-negative integer)
IN datatypetype of elements in send and receive buffer (handle)
IN destrank of destination (integer)
IN sendtagsend message tag (integer)
IN sourcerank of source or MPI_ANY_SOURCE (integer)
IN recvtagreceive message tag or MPI_ANY_TAG (integer)
IN commcommunicator (handle)
OUT statusstatus object (status)
C binding
int MPI_Sendrecv_replace(void *buf, int count, MPI_Datatype datatype, int dest, int sendtag, int source, int recvtag, MPI_Comm comm, MPI_Status *status)
int MPI_Sendrecv_replace_c(void *buf, MPI_Count count, MPI_Datatype datatype, int dest, int sendtag, int source, int recvtag, MPI_Comm comm, MPI_Status *status)
Fortran 2008 binding
MPI_Sendrecv_replace(buf, count, datatype, dest, sendtag, source, recvtag, comm, status, ierror)

TYPE(*), DIMENSION(..) :: buf
INTEGER, INTENT(IN) :: count, dest, sendtag, source, recvtag
TYPE(MPI_Datatype), INTENT(IN) :: datatype
TYPE(MPI_Comm), INTENT(IN) :: comm
TYPE(MPI_Status) :: status
INTEGER, OPTIONAL, INTENT(OUT) :: ierror
MPI_Sendrecv_replace(buf, count, datatype, dest, sendtag, source, recvtag, comm, status, ierror) !(_c)

TYPE(*), DIMENSION(..) :: buf
INTEGER(KIND=MPI_COUNT_KIND), INTENT(IN) :: count
TYPE(MPI_Datatype), INTENT(IN) :: datatype
INTEGER, INTENT(IN) :: dest, sendtag, source, recvtag
TYPE(MPI_Comm), INTENT(IN) :: comm
TYPE(MPI_Status) :: status
INTEGER, OPTIONAL, INTENT(OUT) :: ierror
Fortran binding
MPI_SENDRECV_REPLACE(BUF, COUNT, DATATYPE, DEST, SENDTAG, SOURCE, RECVTAG, COMM, STATUS, IERROR)

<type> BUF(*)
INTEGER COUNT, DATATYPE, DEST, SENDTAG, SOURCE, RECVTAG, COMM, STATUS(MPI_STATUS_SIZE), IERROR

Execute a blocking send and receive. The same buffer is used both for the send and for the receive, so that the message sent is replaced by the message received.


Advice to implementors.

Additional intermediate buffering is needed for the ``replace'' variant. ( End of advice to implementors.)


PreviousUpNext
Up: Blocking Send and Receive Operations Next: Datatype Matching and Data Conversion Previous: Passing MPI_STATUS_IGNORE for Status


Return to MPI-4.1 Standard Index
Return to MPI Forum Home Page

(Unofficial) MPI-4.1 of November 2, 2023
HTML Generated on November 19, 2023