[gmx-users] Mdrun_mip efficiency

[Justin Lemkul]

On 5/21/13 2:58 PM, Андрей Гончар wrote:
> Yes, I'm using all cores. All nodes are connected by Ethernet. Now I see
> that when I change number of processes or number of threads per node I can
> change mdrun performance. But I think it is still too slow. Is there a way
> to speed up mdrun on cluster? Some specific settongs or so? Sorry for my
> questions, but I'm quite novice in running gromacs on cluster.
>

You will never get good performance with Ethernet connectivity; it is far too 
low-latency for high-speed codes like Gromacs.  Most high-performance clusters 
use InfiniBand.

-Justin

>
> 2013/5/21 Justin Lemkul <jalemkul at vt.edu>
>
>>
>>
>> On 5/21/13 1:15 PM, Андрей Гончар wrote:
>>
>>> Hello. I have a cluster (12 nodes, 24 cores on each).  Also I've compiled
>>> mdrun with MPI.
>>> Now when I try to run a MD on this cluster, it takes near 20 hours to
>>> complete. But when I try the same simulation on a PC with 8 cores, it
>>> takes
>>> near 10 hours to complete. I thought that it will take less on cluster but
>>> I see the opposite. What I'm doing wrong?
>>>
>>>
>> Are you trying to use all available cores on the cluster?  No system is
>> infinitely parallelizable, so if you're exceeding the limit of scaling
>> efficiency, you're going to see degradation in performance.  Do you get
>> similar performance using 8 cores of the cluster?  What type of network do
>> you have between the nodes?
>>
>> Note that the end of the .log file will have a breakdown of where your run
>> spent time doing what.  That information can be informative in determining
>> where performance is being lost.
>>
>> -Justin
>>
>> --
>> ==============================**==========
>>
>> Justin A. Lemkul, Ph.D.
>> Research Scientist
>> Department of Biochemistry
>> Virginia Tech
>> Blacksburg, VA
>> jalemkul[at]vt.edu | (540) 231-9080
>> http://www.bevanlab.biochem.**vt.edu/Pages/Personal/justin<http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin>
>>
>> ==============================**==========
>> --
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://lists.gromacs.org/**mailman/listinfo/gmx-users<http://lists.gromacs.org/mailman/listinfo/gmx-users>
>> * Please search the archive at http://www.gromacs.org/**
>> Support/Mailing_Lists/Search<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<http://www.gromacs.org/Support/Mailing_Lists>
>>
>
>
>

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================