[Beowulf] RE: Compare and contrast MPI implementations

Bill Bryce bill at platform.com
Fri Dec 16 06:45:34 PST 2005

Hi all I totally agree with Doug.  Modules is a very useful package and
we have used it to manage different mpi versions for our products.  We
used it exactly how Doug described below 'module load intel' etc... Some
SDK vendors such as Absoft provide modules in their package just to
manage all the different compilers and versions of MPI.



-----Original Message-----
From: beowulf-bounces at beowulf.org [mailto:beowulf-bounces at beowulf.org]
On Behalf Of Douglas Eadline
Sent: Thursday, December 15, 2005 4:38 PM
To: David Mathog
Cc: beowulf at beowulf.org
Subject: [Beowulf] RE: Compare and contrast MPI implementations

>> MPI detente is possible. Check out the modules package.
>> http://modules.sourceforge.net/
> Sort of the same problem though.  lamboot (from lam-mpi) assumes
> that hboot is in the path.  There's no time before it invokes
> hboot to run "modules".  If there were a preceding stage in
> this instance just using an export statement with PATH would
> have done the trick.

When I used modules in the past, there were environments set up for
compilers and MPIs.

For instance if I wanted to run LAM/MPI compiled for the Intel Compiler,

module load intel
module load lam/intel

now I have a Intel compiler and LAM environment (compiled for intel).

Now the module configuation files need to be created and
you will need an MPI tree (/opt/mpi/) where you have a bunch of MPI
versions. I have found modules one of the most useful things
on a cluster (from a users point of view).


> Is there some way to coerce lamboot to either:
> 1.  set up environmental variables (preferably via  a script) before
> running its "hboot" command.
> or
> 2.  run something other than "hboot" (ie /usr/common/bin/hboot).
> Thanks,
> David Mathog
> mathog at caltech.edu
> Manager, Sequence Analysis Facility, Biology Division, Caltech

Beowulf mailing list, Beowulf at beowulf.org
To change your subscription (digest mode or unsubscribe) visit

More information about the Beowulf mailing list