Troubles with mpich installation

[Rusty Lusk]

From: paolo.zini at ipcf.cnr.it
Subject: Re: Troubles with mpich installation
Date: Mon, 23 May 2005 18:00:39 +0200

> 
> > >I have troubles with the mpich installation on one opteron cluster.
> >
> > I suggest that you try MPICH2 (www.mcs.anl.gov/mpi/mpich2) instead of
> > MPICH1.  The ch_p4 device in MPICH1 performs eager sending to relatively
> > large message sizes and this can sometimes run into problems when the
> > messages that are being sent eagerly are larger than the socket
> > buffer.  MPICH2 handles this case properly, and we're unlikely to change
> > MPICH1 to fix this.
> >
> > Bill
> 
> Following your suggestion I have downloaded mpich2.
> With portland 6.0 it dont compile.
> It returns one undefinded that appears to be a debug reference, but I have
> disabled the debug...

Can you tell us what this was, so that we can be sure it is eliminated
in the next release?

> Using gcc + intel fortran it compiles without problems, but behaves
> differently from the expectations: refuses to start more than a process on
> each node, ignoring also the multi-cpu specifications on mpd.hosts file.

Can you post (or send to mpich2-maint at mcs.anl.gov) a more detailed
description of what you have done (mpdboot or other method, tests with
mpdtrace, contents of mpd.hosts, etc.)?  Thanks. 

> I have also some test failure...
> 

Please send the details of this to mpich2-maint at mcs.anl.gov

> I don't have the logs available, but I can regenerate it if you wants...
> 

We need to see the configure and make logs to understand compilation
problems.  I would suggest sending them to mpich2-maint rather than
posting them to mpich-discuss.

> Any way my priority i sto have mpich running with Portland suite...
> 

It works for us so we need to understand the details of why it doesn't
work for you.

Regards,
Rusty Lusk