[gmx-users] Domain decomposition error

Justin Lemkul jalemkul at vt.edu
Thu Oct 29 19:15:01 CET 2015

On 10/29/15 4:56 AM, badamkhatan togoldor wrote:
> Dear GMX Users, I am simulating a free energy of a protein chain_A in water
> by parallel. Then i got domain decomposition error in mdrun. Will use 15
> particle-particle and 9 PME only ranksThis is a guess, check the performance
> at the end of the log file
> -------------------------------------------------------Program mdrun_mpi,
> VERSION 5.1.1-dev-20150819-f10f108Source code file:
> /tmp/asillanp/gromacs/src/gromacs/domdec/domdec.cpp, line: 6969 Fatal
> error:There is no domain decomposition for 15 ranks that is compatible with
> the given box and a minimum cell size of 5.68559 nmChange the number of ranks
> or mdrun option -rddLook in the log file for details on the domain
> decomposition
> Then i look through the .log file, there was 24 rank . So how can i change
> this ranks? What's wrong in here? Or something wrong in my .mdp file ?  Or
> wrong construction on my script in parallel ? I am using just 2 nodes with 24
> cpu. Then i don't think my system is too small (one protein chain, solvent is
> around 8000 molecules and few ions). Initializing Domain Decomposition on 24
> ranksDynamic load balancing: offWill sort the charge groups at every domain
> (re)decompositionInitial maximum inter charge-group distances:    two-body
> bonded interactions: 5.169 nm, LJC Pairs NB, atoms 81 558  multi-body bonded

Given the two-body interaction length, your .mdp file probably specifies 
couple-intramol = no, which generates explicit pairs and exclusions for 
intramolecular interactions, thus driving up the minimum size of a DD cell 
considerably.  So your system is incompatible with more than a few DD cells.

The better question is why you're trying to decouple an entire protein; that is 
extremely impractical and unlikely to be useful.


> interactions: 0.404 nm, Ryckaert-Bell., atoms 521 529Minimum cell size due to
> bonded interactions: 5.686 nmMaximum distance for 13 constraints, at 120 deg.
> angles, all-trans: 0.218 nmEstimated maximum distance required for P-LINCS:
> 0.218 nmGuess for relative PME load: 0.38Will use 15 particle-particle and 9
> PME only ranksThis is a guess, check the performance at the end of the log
> fileUsing 9 separate PME ranks, as guessed by mdrunOptimizing the DD grid for
> 15 cells with a minimum initial size of 5.686 nmThe maximum allowed number of
> cells is: X 1 Y 1 Z 0 Can anybody help this issue? Tnx Khatnaa


Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441


More information about the gromacs.org_gmx-users mailing list