[gmx-users] Load imbalance

Justin A. Lemkul jalemkul at vt.edu
Fri Jan 22 13:35:32 CET 2010



Carla Jamous wrote:
> Hi everyone,
> 
> Lately I've been using gromacs version 3.3.3. Yesterday, I started 
> simulations in a cluster where gromacs version is 4.0.3.

Why start with 4.0.3?  You should be using the latest stable version (4.0.7).

> After a run, I got this error message:
>  Average load imbalance: 0.5 %
>  Part of the total run time spent waiting due to load imbalance: 0.3 %
>  Steps where the load balancing was limited by -rdd, -rcon and/or -dds: 
> X 0 % Y 0 %
>  Average PME mesh/force load: 0.898
>  Part of the total run time spent waiting due to PP/PME imbalance: 2.7 %
> 
> I don't have a clue what this means & how can I fix it?

There's nothing that needs to be fixed, per se.  "Load imbalance" has to do with 
how well divided the system in terms of PP/PME balance (discussed in the Gromacs 
manual - read about domain decomposition).  The very small percent imbalance you 
got means you probably achieved decent performance given however this PP/PME 
imbalance.  There are many threads in the list archive about how to make things 
faster and what this all means.  The manual should be the first port of call, 
after which you should have a look in the archives.  If you want to try to 
optimize something about your system, we will need a lot more information - the 
DD grid, .mdp file, how many nodes are being assigned for PME (and whether or 
not you're doing it manually), etc.

-Justin

> 
> Please can anyone help?
> 
> Thank you
> Carla
> 

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

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

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



More information about the gromacs.org_gmx-users mailing list