| MPI_Comm_join(3) | LAM/MPI | MPI_Comm_join(3) |
MPI_Comm_join - Connect two MPI processed joined by a socket
#include <mpi.h> int MPI_Comm_join(int fd, MPI_Comm *newcomm)
This function only works between two LAM/MPI processes that are connected by a socket. They either must have the same endian orientation, or not have used the homogeneous flag to mpirun (1) (-O). Both processes must be in a single LAM universe -- they must share LAM daemons that are already connected to each other. That is, they were either initially lamboot(1)ed together, or a lamgrow(1) command was given to grow an initial LAM universe such that the resulting set includes the two hosts in question.
The socket may not be used by the calling application until MPI_Comm_join has completed. When MPI_Comm_join completes, fd may be used by the calling application. Per MPI-2, the socket is only used for initial handshaking -- it will not be used as the communications channel between the newly connected processes.
The following code shows an abbreviated example using MPI_Comm_join since no example is provided in the MPI-2 standard. Note that only one rank in each MPI application should call this function since each socket can only have two endpoints (i.e., an endpoint in each process). Upon successful completion, the intercommunicator that is returned will have two members -- the local process and the remote process.
MPI_Comm
do_join(char* server_name, int port) {
int sfd, fd = -1;
unsigned char addr[4];
MPI_Comm intercomm;
if (server_name != NULL) {
if (getinetaddr(server_name, &addr) == 0)
fd = open_socket_to_server(addr, port);
} else {
if ((sfd = open_listening_socket(&port)) >= 0) {
fd = accept_client_socket(sfd, -1);
close(sfd);
}
}
if (fd < 0)
MPI_Abort(MPI_COMM_WORLD, 0);
MPI_Comm_join(fd, &intercomm);
return intercomm;
}
All MPI routines in Fortran (except for MPI_WTIME and MPI_WTICK ) have an additional argument ierr at the end of the argument list. ierr is an integer and has the same meaning as the return value of the routine in C. In Fortran, MPI routines are subroutines, and are invoked with the call statement.
All MPI objects (e.g., MPI_Datatype , MPI_Comm ) are of type INTEGER in Fortran.
The IMPI standard only supports MPI-1 functions. Hence, this function is currently not designed to operate within an IMPI job.
If an error occurs in an MPI function, the current MPI error handler is called to handle it. By default, this error handler aborts the MPI job. The error handler may be changed with MPI_Errhandler_set ; the predefined error handler MPI_ERRORS_RETURN may be used to cause error values to be returned (in C and Fortran; this error handler is less useful in with the C++ MPI bindings. The predefined error handler MPI::ERRORS_THROW_EXCEPTIONS should be used in C++ if the error value needs to be recovered). Note that MPI does not guarantee that an MPI program can continue past an error.
All MPI routines (except MPI_Wtime and MPI_Wtick ) return an error value; C routines as the value of the function and Fortran routines in the last argument. The C++ bindings for MPI do not return error values; instead, error values are communicated by throwing exceptions of type MPI::Exception (but not by default). Exceptions are only thrown if the error value is not MPI::SUCCESS .
Note that if the MPI::ERRORS_RETURN handler is set in C++, while MPI functions will return upon an error, there will be no way to recover what the actual error value was.
For more information, please see the official MPI Forum web site, which contains the text of both the MPI-1 and MPI-2 standards. These documents contain detailed information about each MPI function (most of which is not duplicated in these man pages).
http://www.mpi-forum.org/
join.c
| 6/24/2006 | LAM/MPI 7.1.4 |