[gmx-users] nstlist and performances

[Mark Abraham]

On Mon, Nov 18, 2013 at 3:58 PM, Riccardo Concu <ric.concu at gmail.com> wrote:

> Dear all,
> I'm running a simulation of a quite big system and is running very low
> 2ns/day. I need to boost-up the simulation speed but i don't know how. I
> tried to use nstlist=10 but the system crash due to too many lincs
> warning. Previously I minimized the system, then annealed and now i need
> to run the production. All the statistics of the minimization step and
> the anneling step seems good. Below I have attached the .mdp file.
> Thank you,
> Riccardo
>
> integrator               = md
> tinit                    = 0
> dt                       = 0.002
> nsteps                   = 2000000;4ns
> comm-mode                = Linear
> nstcomm                  = 10
> comm-grps                = System
> nstxout                  = 25000
> nstvout                  = 25000
> nstfout                  = 25000
> nstlog                   = 25000
> nstenergy                = 25000
> nstxtcout                = 25000
> xtc-precision            = 25000
> xtc-grps                 = System
>

Writing output is not free - choose to write the kinds of data you actually
want.


> energygrps               = NAP NA WAT MET SI3 I LI+
>

This can be costly - do this in an mdrun -rerun, on only the frames you
want, if you actually need it.


> nstlist                  = 5
> ns_type                  = grid
> pbc                      = xyz
> periodic_molecules       = no
> rlist                    = 0.9
> coulombtype              = PME
> rcoulomb                 = 0.9
> vdw-type                 = Cut-off
> rvdw                     = 0.9
> fourierspacing           = 0.12
> fourier_nx               = 0
> fourier_ny               = 0
> fourier_nz               = 0
> pme_order                = 6
>

Why? Nobody does this AFAIK


> ewald_rtol               = 1e-5
> optimize_fft             = yes
> Tcoupl                   = v-rescale
> tc-grps                  = System
> tau_t                    =  0.1
> ref_t                    = 298
> Pcoupl                   = Parrinello-Rahman
> Pcoupltype               = Isotropic
> tau_p                    = 1
> compressibility          = 4.5e-5
> ref_p                    = 1.01325
> gen_vel                  = yes
>

I'd follow the advice here
http://www.gromacs.org/Documentation/How-tos/Steps_to_Perform_a_Simulation
about
equilibrating, which should relax the effective nstlist constraint, because
you're not doing big volume changes with a marginally stable barostat.

Then I'd do my measuring of performance over a production run, and look at
the reports at the end of the .log file to see what's taking time that
might be inappropriate.

Mark


> gen_temp                 = 298
> gen_seed                 = 173529
> constraints              = all-bonds
> constraint-algorithm     = Lincs
> lincs-order              = 4
> lincs-iter               = 1
>
>
> --
> gromacs.org_gmx-users mailing list
> gromacs.org_gmx-users at maillist.sys.kth.se
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>