[GE users] lam/mpi sge and msc nastran integration
reuti at staff.uni-marburg.de
Tue Oct 21 22:24:31 BST 2008
[ The following text is in the "ISO-8859-1" character set. ]
[ Your display is set for the "ISO-8859-10" character set. ]
[ Some special characters may be displayed incorrectly. ]
Am 21.10.2008 um 23:07 schrieb Pedro G:
> Thank you both of you, I will try your possibilities.
> Firstly I will comment the lines Reuti suggested and if it doesn't
> work I will try to migrate to hp-MPI.
> I have a question about it... ?HP-MPI is only for HP hardware or it
> can be installed and work in any linux pc?
no, you can get it for several operating systems, including Linux 32-
and 64-bit. It's just one MPI implementation created and sold by HP.
We use it also here with some parallel apps. Usually it's a runtime
license for the enduser, hence you can execute programs built with
it, but you can't compile new programs with it, as the necessary
license key is built into the application only.
> Thank you very much.
> On Tue, Oct 21, 2008 at 6:14 PM, Olesen, Mark
> <Mark.Olesen at emcontechnologies.com> wrote:
> > The problem might be: which version of LAM/MPI was used to compile
> > Nastran? Only since 7.1.1 it's SGE aware. Maybe MSC used an older
> > version.
> According to your original post, you're using nastran2007 on Linux.
> From the MSC release notes, HP MPI 2.2.5 is the default mpi.
> I would strongly recommend that you use this.
> Despite Reuti's valiant efforts, getting LAM/MPI working nicely is
> incredibly annoying - even if you don't have multiple types of
> that use slightly different LAM/MPI installations, and even if your
> LAM/MPI version is ostensibly GridEngine-aware.
> We have both HP-MPICH (STAR-CD) and openmpi (OpenFOAM) in use and both
> work fine with the GridEngine. LAM/MPI was never anything but annoying
> to work with, and I'm glad not to touch it anymore.
> For HP-MPICH, our pe looks like this:
> pe_name mpich
> slots 999
> user_lists NONE
> xuser_lists NONE
> start_proc_args /opt/n1ge6/mpi/startmpi.sh -catch_rsh $pe_hostfile
> stop_proc_args /opt/n1ge6/mpi/stopmpi.sh
> allocation_rule $fill_up
> control_slaves TRUE
> job_is_first_task FALSE
> urgency_slots min
> I prefer not to rely on the rsh wrapper that linked in by the -
> mechanism will actually get seen first in the PATH. The STAR-CD
> for example, have their own rsh wrapper that is also supposed to get
> seen first.
> To be certain that qrsh is actually used by the MPICH transport, we
> export these explicitly in the job-scripts:
> # hp-mpi
> MPI_REMSH=$SGE_ROOT/mpi/rsh; export MPI_REMSH
> # mpich
> P4_RSHCOMMAND=$SGE_ROOT/mpi/rsh; export P4_RSHCOMMAND
> And in the GridEngine configuration we have
> rsh_daemon /usr/sbin/sshd -i
> rsh_command /usr/bin/ssh
> rlogin_command /usr/bin/ssh
> Even if you don't care about correct accounting, the tight integration
> via qrsh is important if you want to avoid leaving zombie processes
> you use qdel to kill parallel jobs.
> This e-mail message and any attachments may contain
> legally privileged, confidential or proprietary Information,
> or information otherwise protected by law of EMCON
> Technologies, its affiliates, or third parties. This notice
> serves as marking of its "Confidential" status as defined
> in any confidentiality agreements concerning the sender
> and recipient. If you are not the intended recipient(s),
> or the employee or agent responsible for delivery of this
> message to the intended recipient(s), you are hereby
> notified that any dissemination, distribution or copying
> of this e-mail message is strictly prohibited.
> If you have received this message in error, please
> immediately notify the sender and delete this e-mail
> message from your computer.
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net
More information about the gridengine-users