[gmx-users] Re: parallel simulation

[Kukol, Andreas]

You need to contact your cluster administrator for instructions of how to submit jobs to the cluster. Usually you have to create some kind of shell-script that specifies various parameters of your job and then submit it to a queue system.

Below you submitted the job most likely to the 'head-node', the node you login first to work the cluster.

Andreas

>-----Original Message-----
>From: gmx-users-bounces at gromacs.org [mailto:gmx-users-
>bounces at gromacs.org] On Behalf Of pratibha kapoor
>Sent: 07 October 2013 14:46
>To: gmx-users at gromacs.org
>Subject: [gmx-users] Re: parallel simulation
>
>To add : I am running simulations on institute cluster with 8 nodes (2 cores
>each).
>Please suggest me the way so that I can run one simulation on all available
>nodes, cores and threads.
>Thanks in advance.
>
>
>
>On Mon, Oct 7, 2013 at 1:55 PM, pratibha kapoor
><kapoorpratibha7 at gmail.com>wrote:
>
>> I would like to run one simulation in parallel so that it utilises all
>> the available nodes and cores. For that, I have compiled gromacs with
>> mpi enabled and also installed openmpi on my machine.
>> I am using the following command:
>> mpirun -np 4 mdrun_mpi -v -s *.tpr
>>
>> When i use top command, I get:
>>
>> PID  USER   PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
>>
>> 22449 root      20   0  107m  59m 3152 R    25           2.9   0:05.42
>> mdrun_mpi
>> 22450 root      20   0  107m  59m 3152 R    25           2.9   0:05.41
>> mdrun_mpi
>> 22451 root      20   0  107m  59m 3152 R    25           2.9   0:05.41
>> mdrun_mpi
>> 22452 root      20   0  107m  59m 3152 R    25           2.9   0:05.40
>> mdrun_mpi
>>
>> Similarly when i use mpirun -np 2 mdrun_mpi -v -s *.tpr, I get
>>
>> PID  USER   PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
>> 22461 root      20   0  108m  59m 3248 R   50          3.0   5:58.64
>> mdrun_mpi
>> 22462 root      20   0  108m  59m 3248 R   50           3.0   5:58.56
>> mdrun_mpi
>>
>> If I look at %CPU column, it is actually 100/(no. of processes) Why is
>> all the cpu not 100% utilised?
>> Also if I compare my performance, it is significantly hampered.
>> Please suggest me the way so that I can run one simulation on all
>> available nodes, cores and threads.
>> Thanks in advance.
>>
>--
>gmx-users mailing list    gmx-users at gromacs.org
>http://lists.gromacs.org/mailman/listinfo/gmx-users
>* Please search the archive at
>http://www.gromacs.org/Support/Mailing_Lists/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/Support/Mailing_Lists