[gmx-users] 2nd question related to load imbalance

[Erik Marklund]

delara aghaie skrev 2011-04-13 11.26:
> Hello
> Previously I used gromacs/3.3 for my simulations of DPPc 
> monolayer+Tip4P2005. I saw that it uses particle decomposition
> Now I want to use versin 4.0.7 because I need tcoupl=v-rescale
> This version does the domain decomposition which causes load imbalance.
> with this version because of load imbalance and loss of performance 
> the simulation needs longer time to run.
> so the question:
> what is the benefit of this domain decomposition. I think it causes 
> the runs to be done slower !!!.
> thanks
> D.aghaie
>
Most of the time it reduces communication of coordinates, which in turn 
can is one of the big bottlenecks in parallel MD. Domain decomposition 
makes sure that atoms which are close to eachother, and hence are likely 
to interact, are passed to the same core. With particle decomposition 
the atoms are statically asigned to the different cores, and the 
decomposition may not reflect how the aroms interact.

Why go for 4.0.7, why not 4.5.4?

-- 
-----------------------------------------------
Erik Marklund, PhD student
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596,    75124 Uppsala, Sweden
phone:    +46 18 471 4537        fax: +46 18 511 755
erikm at xray.bmc.uu.se    http://folding.bmc.uu.se/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20110413/7a210eca/attachment.html>