[mpich-discuss] (no subject)

[jayesh at mcs.anl.gov]

Hi,
 Are you using MPICH2 on Unix or Windows ?
 Assuming that you are using MPICH2 on a Windows machine, Can you try your code explicitly with the sock and nemesis channels (mpiexec -n 2 -localroot -channel sock mympipgm.exe; mpiexec -n 2 -localroot -channel nemesis mympipgm.exe)?
 Can you send us a test program that shows the problem ?

Regards,
Jayesh

----- Original Message -----
From: "Huasheng Liao" <liao at egr.msu.edu>
To: mpich-discuss at mcs.anl.gov
Sent: Sunday, September 26, 2010 4:05:20 PM GMT -06:00 US/Canada Central
Subject: [mpich-discuss] (no subject)

Hi, there,
I have been using MPICH2-1.0.5p since I satrted my first MPI program. I 
happened to update the old version to MPICH2-1.1.2, and ran the old
program (GUI, -localroot was used) and found that it dosen't work well
with MPICH2-1.1.2 :

In my program, MPI_SPAWN_MULTIPLE routine is involved to spawn jobs to
several child machines, then a few Send and Recv calls are involved to
communicate with master machine. When I spawn only one child machine,
works fine,but will be stuck without any error when spawn more than one
child machine.

I was wondering whether I have missed any new setting in MPICH2-1.2.1p1 ?

BTW, it works fine when I degraded my code to purely fortran code without
GUI.



Huasheng Liao



_______________________________________________
mpich-discuss mailing list
mpich-discuss at mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss