Actually, in a not-too-distant future release, there will be an option to mpirun called "--connect" that will allow you to specify that this job is to be connected to a specified earlier job. The run-time environment will then spawn the new job and exchange all required communication information between the two jobs for you. You could therefore accomplish your desired operation by:

> nohup mpirun --np xx app1
(system returns job number to you)
> mpirun --np yy --connect job1 app2
(system starts app2 and connects it to job1)

Should be a little more transparent. No specific coding for making the connection would be required in your application itself.

Ralph


Jean Latour wrote:
Hello,

It seems to me there is only one way to create a communication between
two MPI_COMM_WORLD :  use MPI_Open_Port with a specific
IP + port address, and then MPI_comm_connect / MPI_comm_accept.

In order to ease the port number communication, the use of MPI_publish-name
/ MPI_lookup_name is also possible with the constraint that the "publish"
must be done before the "lookup", and this involves some synchronization
between the processes anyway.

Simple examples can be found in the handbook on MPI : "Using MPI-2"
by William Gropp et al.

Best Regards,
Jean

Ali Eghlima wrote:



Hello,

I have read MPI-2 documents as well as FAQ. I am confused as the best way to establish communication
between two  MPI_COMM_WORLD which has been created by two mpiexec calls on the same node.

mpiexec -conf  config1
     This start 20 processes on 7 nodes

mpiexec -conf  config2
      This start 18 processes on 5 nodes

I do appreciate any comments or pointer to a document or example.

Thanks

Ali,

 

------------------------------------------------------------------------

_______________________________________________
users mailing list
us...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/users



_______________________________________________ users mailing list us...@open-mpi.org http://www.open-mpi.org/mailman/listinfo.cgi/users


Reply via email to