[gmx-users] mpich/share/machines file for dual processor machine
Peter C. Lai
sirmoo at cowbert.2y.net
Thu Sep 29 21:54:40 CEST 2005
On Thu, Sep 29, 2005 at 05:06:18PM +0200, Jeroen Mesters wrote:
> Looking at the md.log file, the first line states the following: *Log
> file opened: nodeid 0, nnodes = 1, host = orange, process = 7356*
> Where does mdrun_mpi get the *nnodes=1* information from?
> Does anyone have an *mpich/share/machines* file that shows me how a
> *dual-processor* machine is configured i.e. how do I tell mpich/gromacs
> there are two processors in one box and it should use both to run the job?
[plai at pulau]:~/speptide % more /usr/local/mpich/share/machines.freebsd
# Change this file to contain the machines that you want to use
# to run MPI jobs on. The format is one host name per line, with either
# where n is the number of processors in an SMP. The hostname should
# be the same as the result from the command "hostname"
I invoke both grompp and mpirun using the -np 2 flag. grompp is also invoked
using the -sort flag.
In my linked libs, I noticed that both grompp and mdrun were linked to:
libmd_mpi.so.3 => /usr/local/lib/libmd_mpi.so.3 (0x280c7000)
libgmx_mpi.so.3 => /usr/local/lib/libgmx_mpi.so.3 (0x2810b000)
libsrfftw_mpi.so.2 => /usr/local/lib/libsrfftw_mpi.so.2 (0x28225000)
libsfftw_mpi.so.2 => /usr/local/lib/libsfftw_mpi.so.2 (0x28228000)
libsrfftw.so.2 => /usr/local/lib/libsrfftw.so.2 (0x2822f000)
libsfftw.so.2 => /usr/local/lib/libsfftw.so.2 (0x28254000)
libm.so.3 => /lib/libm.so.3 (0x28843000)
libpthread.so.1 => /usr/lib/libpthread.so.1 (0x2885e000)
libc.so.5 => /lib/libc.so.5 (0x28884000)
(X11 libs not shown)
You do not need mpd if you aren't in a clustered env. (it's for cluster
> Thanx in advance, Jeroen.
> Jeroen Mesters wrote:
> >Machine: *dual* processor dell precision 650 with SuSE linux 9.2
> >Software: *1)* mpich installed, machines file constructed, mpd started
> >*2)* compiled gromacs twice, once with and once without --enable-mpi
> >(using mpich). So, I ended up with two programmes, mdrun_mpi and
> >mdrun. The program versions are different in size which I would
> >expect... Indeed, one program was compiled with cc and the other with
> >mpicc (no errors during compilation).
> >Step 1: grompp -np 2 -f full.mdp -c after_pr_CA.gro -p ph5.top -n
> >index.ndx -o 2cpu.tpr
> >Step 2: mdrun_mpi -np 2 -s 2cpu.tpr >>>> Fatal error: run input file
> >2cpu.tpr was made for 2 nodes, while mdrun_mpi expected it to be for 1
> >Question 1. Are my commands correct i.e. is this the proper why to start?
> >Question 2. Can this work at all or do I really need *2* computers (2
> >Thanx for any hints, Jeroen.
> >Jeroen Raymundus Mesters, Ph.D.
> >Institut fuer Biochemie, Universitaet zu Luebeck
> >Ratzeburger Allee 160, D-23538 Luebeck
> >Tel: +49-451-5004070, Fax: +49-451-5004068
> >E-mail: mesters at biochem.uni-luebeck.de
> >If you can look into the seeds of time and say
> >which grain will grow and which will not - speak then to me (Macbeth)
> >gmx-users mailing list
> >gmx-users at gromacs.org
> >Please don't post (un)subscribe requests to the list. Use the
> >www interface or send it to gmx-users-request at gromacs.org.
> Jeroen Raymundus Mesters, Ph.D.
> Institut fuer Biochemie, Universitaet zu Luebeck
> Ratzeburger Allee 160, D-23538 Luebeck
> Tel: +49-451-5004070, Fax: +49-451-5004068
> E-mail: mesters at biochem.uni-luebeck.de
> If you can look into the seeds of time and say
> which grain will grow and which will not - speak then to me (Macbeth)
> gmx-users mailing list
> gmx-users at gromacs.org
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
Peter C. Lai
Cesium Hyperfine Enterprises
More information about the gromacs.org_gmx-users