[gmx-users] parallel calculation problem in implicit solvent modeling

[Justin Lemkul]

On 3/19/15 9:25 PM, Ming Tang wrote:
> Dear all,
>
> I had a problems with parallel calculation of a implicit solvent problem. My gromacs is compiled with mpi, so technically, I run mdrun_mpi for my simulation. I am calling 12 cpus for my simulation and started 12 treads.
>
> When I put a limited rlist in the simulation, say rlist=4, 12 cpus calculation runs good. However, when I put an infinite one, which is rlist=rvdw=rcoulmb=rgbradii=0, the parallel calculation doesn't move anymore, and doesn't give me error msg, either. The output window is freezed, showing a note:
>
>
> Non-default thread affinity set probably by the OpenMP library,
> disabling internal thread affinity
>
> If I change it to one CPU calculation, the simulation will still run, with rlist=0.
>
> I also tried 'simple' method for ns algorithm, but the mpi compiles parallel calculation does not allow to use it due to the default 'Domain decomposition' algrithm.
>
> Can any mate give me some hints? Thanks in advance.
>

You're probably just triggering a totally untested combination, so the code 
hangs because no one's looked into that combination.  12 CPU (DD?) + infinite 
cutoffs + grid searching...it gets complicated in a hurry.  There is not much to 
be gained by using many cores/threads for implicit solvent; in my experience, it 
doesn't scale very well.

-Justin

> Cheers
> Tony
>
> My running command is:
>
> export OMP_NUM_THREADS=1
> mdrun_mpi -v -ntomp 1 -cpi -cpo -c nvt.pdb
>
> My mdp file is:
>
> define                   = -DPOSRES
> ; RUN CONTROL PARAMETERS
> integrator               = steep
> ; Start time and timestep in ps
> tinit                    = 0
> dt                       = 0.001
> nsteps                   = 50000
> ; For exact run continuation or redoing part of a run
> simulation_part          = 1
> init_step                = 0
> ; mode for center of mass motion removal
> comm-mode                = Linear
> ; number of steps for center of mass motion removal
> nstcomm                  = 1
> ; group(s) for center of mass motion removal
> comm-grps                =
>
> ; LANGEVIN DYNAMICS OPTIONS
> ; Friction coefficient (amu/ps) and random seed
> bd-fric                  = 0
> ld-seed                  = 1993
>
> ; ENERGY MINIMIZATION OPTIONS
> ; Force tolerance and initial step-size
> emtol                    = 5
> emstep                   = 0.1
> ; Max number of iterations in relax_shells
> niter                    = 20
> ; Step size (ps^2) for minimization of flexible constraints
> fcstep                   = 0
> ; Frequency of steepest descents steps when doing CG
> nstcgsteep               = 1000
> nbfgscorr                = 10
>
> ; TEST PARTICLE INSERTION OPTIONS
> rtpi                     = 0.05
>
> ; OUTPUT CONTROL OPTIONS
> ; Output frequency for coords (x), velocities (v) and forces (f)
> nstxout                  = 100
> nstvout                  = 100
> nstfout                  = 100
> ; Output frequency for energies to log file and energy file
> nstlog                   = 100
> nstenergy                = 100
> ; Output frequency and precision for xtc file
> nstxtcout                = 100
> xtc-precision            = 100
>
> ; NEIGHBORSEARCHING PARAMETERS
> ; nblist update frequency
> nstlist                  = 40
> ; ns algorithm (simple or grid)
> ns_type                  = grid
> ; Periodic boundary conditions: xyz, no, xy
> pbc                      = no
> ; nblist cut-off
> cutoff-scheme         = group
> rlist                    = 0
>
> ; OPTIONS FOR ELECTROSTATICS AND VDW
> ; Method for doing electrostatics
> coulombtype              = cut-off
> rcoulomb-switch          = 1
> rcoulomb                 = 0
>
> ; Method for doing Van der Waals
> vdw-type                 = cut-off
> ; cut-off lengths
> rvdw-switch              = 1
> rvdw                     = 0
>
> ; IMPLICIT SOLVENT ALGORITHM
> implicit-solvent         = GBSA
> gb-algorithm             = still
> rgbradii                 = 0
> sa_surface_tension     = 2.05016
>

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

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

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