Mpiexec and mcnpx

[Vanhaeren Ben]

Hello Pete,

I did edit the source because the mcnpx binary can't handle the arguments mpiexec passes to it. mcnpx also accepts arguments itself and can't differentiate betweem mpi arguments and mcnpx program arguments. So instead of editing the Fortran code that handles the program arguments, I decided to edit the mpiexec code that passes the arguments. The changes I made work because I see that the mcnpx program starts in pbs with the correct arguments. What happens is that the master process gets started, and starts doing its calculations but the slave processes don't start, because they can't contact the master process.

I can see that it tries to do a read() but that returns an EAGAIN error. It than polls to see if the process dies and loops infinitely because the master process returns nothing.

You say that MPI_Init() should try to contact mpiexec or mpirun somewhere. How would this code look ? Maybe I can change this in the mcnpx code that initiates MPI.

I really appreciate your help,
Thanks

Ben
-----Original Message-----
From: Pete Wyckoff [mailto:pw at osc.edu]
Sent: Fri 10/15/2004 10:25 PM
To: Vanhaeren Ben
Cc: mpiexec at osc.edu
Subject: Re: Mpiexec and mcnpx
 
ben.vanhaeren at sckcen.be wrote on Fri, 15 Oct 2004 14:15 +0200:
> I am trying to get mpiexec to work with mcnpx. I have edited the source so 
> than the right arguments are passed to the mcnpx excecutable but there seems 
> to be a problem with read_p4_master_port function . 
> 
> read(mport_fd,&port, sizeof(port)) always returns EAGAIN error. so mpiexec 
> loops infinitely.
> 
> D o you have any ideas on how to debug this ?

You edited the mpiexec source?  Just to run one particular parallel
program?  Maybe I misinterpret your message, but the idea is that
you never have to do that.

That read(mport_fd...) is mpiexec waiting for process #0 to contact it
as it should do inside MPI_Init() somewhere.  You might debug the
process or add printfs to see what it is trying to do.

		-- Pete



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://email.osc.edu/pipermail/mpiexec/attachments/20041016/61777f5d/attachment.htm