[gmx-users] Error during mdrun

[Justin Lemkul]

On 2/23/15 9:32 AM, soumadwip ghosh wrote:
> Dear all,
>               I started an MDrun of Barstar protein in TIP3P water and 8M
> urea, neutralized by the required number of counterions. Until the mdrun,
> there was no issue with either energy minimization and equilibration steps.
> But whenever I want to start MD run, a fatal error shows up saying...
>
>
> Reading file mdrun_100ns.tpr, VERSION 4.5.6 (single precision)
> Starting 80 threads
>
> Will use 70 particle-particle and 10 PME only nodes
> This is a guess, check the performance at the end of the log file
>
> -------------------------------------------------------
> Program mdrun, VERSION 4.5.6
> Source code file: domdec.c, line: 6436
>
> Fatal error:
> There is no domain decomposition for 70 nodes that is compatible with the
> given box and a minimum cell size of 0.952625 nm
> Change the number of nodes or mdrun option -rcon or -dds or your LINCS
> settings
>
> I am sure I am not doing parallel runs on any other processor. It is
> written in the GROMACS error analysis that' If you didn't think you were
> running a parallel calculation, be aware that from 4.5, GROMACS uses
> thread-based
> parallelism <http://www.gromacs.org/Documentation/Terminology/Threading> by
> default. To prevent this, you can either give mdrun
> <http://www.gromacs.org/Documentation/Gromacs_Utilities/mdrun> the "-nt 1"
> command line option, or build GROMACS so that it will not use threads.
> Otherwise, you might be using an MPI-enabled GROMACS and not be aware of
> the fact.'
>
> Therefore, according to the suggestion I have used the option -nt 1 in the
> md run command line. The problem is the simulation is getting slower than
> most of the other calculations I 've performed with the same system. Can
> anyone tell me why is it so? I 've never done parallel computation to
> accelerate the simulations. Also, with the option -nt 1 do I ensure correct
> sampling of the system? I hope I'm not asking for much..
>

You need to figure out what hardware you actually have (e.g. number of cores, 
whether or not the processors support hyperthreading, etc), how GROMACS was 
compiled (if you didn't do it yourself), what your OMP_NUM_THREADS environment 
variable is set to, etc.  Normally, GROMACS finds and uses all available 
processors; you didn't show us your original command that tried to use 80 
threads, so we don't know if you set something wrong or what's going on.

-Justin

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

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
http://mackerell.umaryland.edu/~jalemkul

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