[gmx-users] only one cpu "works" in my linux cluster

[Alan Dodd]

Yeah.
Check the machinefile tells mpich what you think it does.  It looks pretty certain to be an issue with mpich rather than gromacs.  Unfortunately, I don't use mpich any more because I find LAM better.  Standard procedure, turn up the verbosity on everything, check the outputs, re-read the relevant manuals.

----- Original Message ----
From: liu xin <zgxjlx at gmail.com>
To: Discussion list for GROMACS users <gmx-users at gromacs.org>
Sent: Tuesday, October 9, 2007 1:38:50 PM
Subject: Re: [gmx-users] only one cpu "works" in my linux cluster


Thanks Alan
in fact I've already added the option -machinefile, here's the script:

cat $PBS_NODEFILE >/home/liuxin/mpd.hosts
/home/liuxin/mpich2/bin/mpdboot -n 6 -f  mpd.hosts
/home/liuxin/mpich2/bin/mpiexec -machinefile $PBS_NODEFILE -np 12  /home/liuxin/
programs/gromacs33mpi/bin/mdrun -v -s 12np.tpr -o -c -e -g  -np 12
 
I am a little busy today, sorry for replying so late...
 
On 10/8/07, Alan Dodd <anoddlad at yahoo.com> wrote: 
Check the MPICH manuals for how to specify the nodes to run on.  From memory, the option -machinefile lets you do this. 



----- Original Message ----
From: liu xin < zgxjlx at gmail.com>
To: Discussion list for GROMACS users <gmx-users at gromacs.org>
Sent: Monday, October 8, 2007 1:51:12 PM 
Subject: [gmx-users] only one cpu "works" in my linux cluster


Dear GMXers
 
this is how I've done it so far:
grompp -f -c -p -o 12np.tpr -np 12
qsub -l node=6 12np.sh (/home/me/mpich2/bin/mpirun -np 12 mdrun -s 12np.tpr -np 12 )
then it seems my mdrun works fine, but when I ssh to each node to check the cpu efficiency with "top", I find that there's only one cpu works with 12 processes! The other nodes are completely idle.
 
the cluster have 56 Intel Xeon duo 3.0G CPUs, below is my system info:
Linux login 2.4.21-32.EL #1 SMP Fri Apr 15 21:02:58 EDT 2005 x86_64 x86_64 x86_64 GNU/Linux
our administrator have installed mpich1.2.7 in the default DIR (/usr/local), but I have some problems to do mpi-mdrun with mpich1.2.7, so I installed mpich2 in my personal dir.
I've searched the list, but cant get any solution. Does it have something to do with the kernel or there's some conflict between mpich1 and 2, or something else?
 
This is the first time I build up a Linux Cluster on my own, ANY suggestions are appreciated !
 
Xin Liu





Fussy? Opinionated? Impossible to please? Perfect. Join Yahoo!'s user panel and lay it on us. 

_______________________________________________
gmx-users mailing list     gmx-users at gromacs.org
http://www.gromacs.org/mailman/listinfo/gmx-users
Please search the archive at http://www.gromacs.org/search before posting!
Please don't post (un)subscribe requests to the list. Use the 
www interface or send it to gmx-users-request at gromacs.org.
Can't post? Read http://www.gromacs.org/mailing_lists/users.php


      ____________________________________________________________________________________
Check out the hottest 2008 models today at Yahoo! Autos.
http://autos.yahoo.com/new_cars.html
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20071009/e417b3d5/attachment.html>