[gmx-users] gromacs parallel doesn't work

Arneh Babakhani ababakha at mccammon.ucsd.edu
Wed Jun 14 08:13:27 CEST 2006


Hello, I'm having a similar problem, although my OS sees all 4 of my 
processors just fine.

When I run mpirun -c 4 mdrun . . .

all 4 of my processors are running, but only at 50% each.  Furthermore, 
in present working directory, I see several backed up files:

[ababakha at chemcca50 TEMP]$ ls
AfterResMD1.gro      #mdout.mdp.1#   ResMD1.job      ResMD1.trr
#AfterResMD1.gro.1#  NoBadWater.top  ResMD1.log      #ResMD1.trr.1#
#AfterResMD1.gro.2#  posre.itp       #ResMD1.log.1#  #ResMD1.trr.2#
dmpc.itp             ResMD1.edr      #ResMD1.log.2#  #ResMD1.trr.3#
lipid.itp            #ResMD1.edr.1#  #ResMD1.log.3#
lipid_posre.itp      #ResMD1.edr.2#  ResMD1.mdp
mdout.mdp            #ResMD1.edr.3#  ResMD1.tpr

It's almost as if I'm running 4 identical jobs (one on each processor) , 
instead of one job in parallel using 4 processors.

Any thoughts?

Thanks,

Arneh






Beniamino Sciacca wrote:
> OK!!!
> now it works.
> I installed a kernel-image smp (for my kernel 2.6)
> and now with "-c 2" it works!!!
> I've two processes running at nearly 100%, and the simulation time is 
> reduced at 50%!
> But now i don't understand why before (one week ago without smp it 
> worked....) mah....
> I thank all you very much
>
> regards
> Beniamino
>
> 2006/6/4, Yang Ye <leafyoung81-group at yahoo.com 
> <mailto:leafyoung81-group at yahoo.com>>:
>
>     So, the kernel you used from Debian could be a non-SMP one. Change
>     another in the boot screen if there is.
>     \
>     Yang Ye
>
>     David van der Spoel wrote:
>     > Beniamino Sciacca wrote:
>     >> with "-c 2" I've two processes, each one running at nearly 50%.
>     >> with "-c 4" I've four processes, each one running at nearly 25%
>     >> with "-c 1" I've only one process wich runs at nearly 99%
>     >> I tried also with mdrun_Nmpi (compiled without enabling mpi),
>     and it
>     >> runs at nearly 99%.
>     >> Is it possible that the software sees only one core?
>     >
>     > I think your OS only sees one processor.
>     > if Linux try:
>     > cat /proc/cpuinfo
>     >
>     >
>     >>
>     >> Beniamino
>     >>
>     >> Yang Ye ha scritto:
>     >>
>     >>> Use top to check, you shall have two processes running at nearly
>     >>> 100% with "-c 2"
>     >>>
>     >>> Yang Ye
>     >>>
>     >>> Beniamino Sciacca wrote:
>     >>>
>     >>>> I've just tried "mpirun -c 2 mdrun..." (34 h/ns).
>     >>>> It works, but the problem is the same... the simulation time....
>     >>>> In fact I also tried "mpirun -c 1 mdrun" (32 h/ns)
>     and  "mpirun -c
>     >>>> 4 mdrun" (34 h/ns), but the simulation time remain the same!
>     >>>>
>     >>>>
>     >>>> Yang Ye ha scritto:
>     >>>>
>     >>>>> Beniamino Sciacca wrote:
>     >>>>>
>     >>>>>> I never used mprun -np 2, and I had not problem. Before I
>     didn't
>     >>>>>> use "mprun -np 2", now I don't use "mprun -np 2" but simulation
>     >>>>>> time now is twice. It has changed from a day to the other.
>     >>>>>> However I've just tried, like you suggested,  to use " -np
>     2" (in
>     >>>>>> grompp and in mdrun) but I obtain an error:
>     >>>>>> -------------------------------------------
>     >>>>>> "Program mdrun_d, VERSION 3.3.1
>     >>>>>> Source code file: init.c, line: 69
>     >>>>>>
>     >>>>>> Fatal error:
>     >>>>>> run input file DNA8_mdEq.tpr was made for 2 nodes,
>     >>>>>>              while mdrun_d expected it to be for 1 nodes.
>     >>>>>> "
>     >>>>>> -------------------------------------------
>     >>>>>> In fact I've only one node.
>     >>>>>
>     >>>>> Try David's mpirun -c 2 mdrun
>     >>>>>
>     >>>>>>
>     >>>>>> Laptop should be running in top speed, I never modified the
>     >>>>>> configuration of energy profile.
>     >>>>>> Beniamino
>     >>>>>>
>     >>>>>>
>     >>>>>> 2006/6/3, Yang Ye <leafyoung81-group at yahoo.com
>     <mailto:leafyoung81-group at yahoo.com>
>     >>>>>> <mailto: leafyoung81-group at yahoo.com
>     <mailto:leafyoung81-group at yahoo.com>>>:
>     >>>>>>
>     >>>>>>     First, make sure that you used mdrun -np 2.
>     >>>>>>     Second, check your energy profile. Is your laptop
>     running in top
>     >>>>>>     speed?
>     >>>>>>     or does it have sleep time?
>     >>>>>>
>     >>>>>>     Yang Ye
>     >>>>>>
>     >>>>>>     Beniamino Sciacca wrote:
>     >>>>>>     > I don't understand what you talking about...
>     >>>>>>     > I didn't write any hostfile.... I'm not in a cluster. I
>     >>>>>> only use my
>     >>>>>>     > notebook with the two cores.
>     >>>>>>     > In fact I type "lamboot" without -v.......
>     >>>>>>     >
>     >>>>>>     > Beniamino
>     >>>>>>     >
>     >>>>>>     > Mark Abraham ha scritto:
>     >>>>>>     >> Did your default LAM hostfile configuration (or whatever
>     >>>>>> it is)
>     >>>>>>     >> change from 2 processors to one?
>     >>>>>>     >>
>     >>>>>>     >> Mark
>     >>>>>>     >> _______________________________________________
>     >>>>>>     >> gmx-users mailing list     gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     >>>>>>     <mailto:gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>>
>     >>>>>>     >> http://www.gromacs.org/mailman/listinfo/gmx-users
>     >>>>>>     >> Please don't post (un)subscribe requests to the
>     list. Use
>     >>>>>> the www
>     >>>>>>     >> interface or send it to
>     gmx-users-request at gromacs.org <mailto:gmx-users-request at gromacs.org>
>     >>>>>>     <mailto:gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>>.
>     >>>>>>     >> Can't post? Read
>     >>>>>> http://www.gromacs.org/mailing_lists/users.php
>     >>>>>>     < http://www.gromacs.org/mailing_lists/users.php>
>     >>>>>>     >>
>     >>>>>>     > _______________________________________________
>     >>>>>>     > gmx-users mailing list     gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     >>>>>>     <mailto:gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>>
>     >>>>>>     > http://www.gromacs.org/mailman/listinfo/gmx-users
>     >>>>>>     > Please don't post (un)subscribe requests to the list. Use
>     >>>>>> the www
>     >>>>>>     > interface or send it to gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>
>     >>>>>>     <mailto:gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>>.
>     >>>>>>     > Can't post? Read
>     >>>>>> http://www.gromacs.org/mailing_lists/users.php
>     >>>>>>     >
>     >>>>>>
>     >>>>>>     --
>     >>>>>>     /Regards,/
>     >>>>>>     Yang Ye
>     >>>>>>     /Computational Biology Lab
>     >>>>>>     School of Biological Sciences
>     >>>>>>     Nanyang Technological University
>     >>>>>>     Singapore
>     >>>>>>     Tel: 6316-2884
>     >>>>>>     /
>     >>>>>>
>     >>>>>>     _______________________________________________
>     >>>>>>     gmx-users mailing list    gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     >>>>>>     <mailto:gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>>
>     >>>>>>     http://www.gromacs.org/mailman/listinfo/gmx-users
>     >>>>>>     Please don't post (un)subscribe requests to the list.
>     Use the
>     >>>>>>     www interface or send it to
>     gmx-users-request at gromacs.org <mailto:gmx-users-request at gromacs.org>
>     >>>>>>     <mailto: gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>>.
>     >>>>>>     Can't post? Read
>     http://www.gromacs.org/mailing_lists/users.php
>     >>>>>>
>     >>>>>>
>     >>>>>>
>     ------------------------------------------------------------------------
>     >>>>>>
>     >>>>>>
>     >>>>>> _______________________________________________
>     >>>>>> gmx-users mailing list    gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     >>>>>> http://www.gromacs.org/mailman/listinfo/gmx-users
>     <http://www.gromacs.org/mailman/listinfo/gmx-users>
>     >>>>>> Please don't post (un)subscribe requests to the list. Use
>     the www
>     >>>>>> interface or send it to gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>.
>     >>>>>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>     >>>>>
>     >>>>>
>     >>>>>
>     >>>>>
>     >>>>> _______________________________________________
>     >>>>> gmx-users mailing list    gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     >>>>> http://www.gromacs.org/mailman/listinfo/gmx-users
>     >>>>> Please don't post (un)subscribe requests to the list. Use
>     the www
>     >>>>> interface or send it to gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>.
>     >>>>> Can't post? Read
>     http://www.gromacs.org/mailing_lists/users.php
>     <http://www.gromacs.org/mailing_lists/users.php>
>     >>>>>
>     >>>> _______________________________________________
>     >>>> gmx-users mailing list    gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     >>>> http://www.gromacs.org/mailman/listinfo/gmx-users
>     >>>> Please don't post (un)subscribe requests to the list. Use the www
>     >>>> interface or send it to gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>.
>     >>>> Can't post? Read
>     http://www.gromacs.org/mailing_lists/users.php
>     <http://www.gromacs.org/mailing_lists/users.php>
>     >>>>
>     >>>
>     >>>
>     >>>
>     >>> _______________________________________________
>     >>> gmx-users mailing list    gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     >>> http://www.gromacs.org/mailman/listinfo/gmx-users
>     >>> Please don't post (un)subscribe requests to the list. Use the www
>     >>> interface or send it to gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>.
>     >>> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>     >>>
>     >> _______________________________________________
>     >> gmx-users mailing list    gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     >> http://www.gromacs.org/mailman/listinfo/gmx-users
>     >> Please don't post (un)subscribe requests to the list. Use the www
>     >> interface or send it to gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>.
>     >> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>     >
>     >
>
>
>
>     _______________________________________________
>     gmx-users mailing list    gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     http://www.gromacs.org/mailman/listinfo/gmx-users
>     Please don't post (un)subscribe requests to the list. Use the
>     www interface or send it to gmx-users-request at gromacs.org
>     <mailto:gmx-users-request at gromacs.org>.
>     Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>     <http://www.gromacs.org/mailing_lists/users.php>
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> 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



More information about the gromacs.org_gmx-users mailing list