[gmx-users] Problem with the mdrun_openmpi on cluster

[Mark Abraham]

Hi,


On Tue, Mar 15, 2016 at 11:57 AM James Starlight <jmsstarlight at gmail.com>
wrote:

> just performed some benchmarks with full atomic system- short md of
> the water soluble protein still using mpiexec -np 46 mdrun_openmpi of
> the GMX 4.5 and there were no such errors with DD so it seems that the
> problem indeed in MARTINI atoms representation.
>

Not really. The bonded interactions have a longer physical range in a CG
model, and that limits the current implementation of domain decomposition.


> BTW how I could quickly check some info about performance of the
> simulation7 what logs should I expect7 If somebody has already done it
>

Depends on your simulation and hardware, so nobody has anything that is
obviously comparable.


> I will be very thankful for some usefull combination of shell commands
> which will extract performance information from sim log.
>

Start with tail -n 50 md.log ;-)

Mark

Thanks in advance!!
>
> J.
>
> 2016-03-14 18:27 GMT+01:00 Justin Lemkul <jalemkul at vt.edu>:
> >
> >
> > On 3/14/16 1:26 PM, James Starlight wrote:
> >>
> >> For that system I have not defined virtual sites.
> >
> >
> > That disagrees with the error message, which explicitly complains about
> > vsites.
> >
> >> BTW the same simulation on local desctop using 2 cores from core2 duo
> runs
> >> OK =)
> >>
> >
> > Because you're not invoking DD there.
> >
> >> so one of the solution probably is to try to use more recent gmx 5.0
> >> to see what will happenes
> >>
> >
> > Good idea.
> >
> > -Justin
> >
> >
> >> 2016-03-14 18:22 GMT+01:00 Justin Lemkul <jalemkul at vt.edu>:
> >>>
> >>>
> >>>
> >>> On 3/14/16 1:19 PM, James Starlight wrote:
> >>>>
> >>>>
> >>>> I tried to increase size on the system providding much bigger bilayer
> >>>> in the system
> >>>>
> >>>> for this task I obtained another error also relevant to DD
> >>>>
> >>>> Program g_mdrun_openmpi, VERSION 4.5.7
> >>>> Source code file:
> >>>> /builddir/build/BUILD/gromacs-4.5.7/src/mdlib/domdec_con.c, line: 693
> >>>>
> >>>> Fatal error:
> >>>> DD cell 0 2 1 could only obtain 0 of the 1 atoms that are connected
> >>>> via vsites from the neighboring cells. This probably means your vsite
> >>>> lengths are too long compared to the domain decomposition cell size.
> >>>> Decrease the number of domain decomposition grid cells.
> >>>> For more information and tips for troubleshooting, please check the
> >>>> GROMACS
> >>>> website at http://www.gromacs.org/Documentation/Errors
> >>>> -------------------------------------------------------
> >>>>
> >>>> "It's So Fast It's Slow" (F. Black)
> >>>>
> >>>> Error on node 9, will try to stop all the nodes
> >>>> Halting parallel program g_mdrun_openmpi on CPU 9 out of 64
> >>>>
> >>>>
> >>>> BTW I checked the bottom of the syste,.gro file and found the next
> >>>> sizes which are seems too small for my syste, consisted for several
> >>>> hundreds of lipid, arent it7
> >>>>
> >>>>     15.00000  15.00000  15.00000   0.00000   0.00000   0.00000
>  0.00000
> >>>>     0.00000   0.00000
> >>>>
> >>>
> >>> No, that seems fine.  But if your box is set up wrong, that's your
> fault
> >>> from the command below :)
> >>>
> >>>>
> >>>> for my case that gro file was produced automatically using MARTINI
> >>>> method
> >>>>
> >>>> ./insane.py -f test.pdb -o system.gro -p system.top -pbc cubic -box
> >>>> 15,15,15 -l DPPC:4 -l DOPC:3 -l CHOL:3 -salt 0.15 -center -sol W
> >>>>
> >>>>
> >>>> Will be very thankful for any help!!
> >>>>
> >>>
> >>> So you've got a system that is a CG model, with virtual sites?  That's
> >>> going
> >>> to create all kinds of havoc.  Please do try Googling your error,
> because
> >>> this difficulty has come up before specifically in the case of CG
> >>> systems,
> >>> which have longer-than-normal bonded interactions and requires some
> mdrun
> >>> tuning.
> >>>
> >>> -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
> >>>
> >>> ==================================================
> >>>
> >>> --
> >>> Gromacs Users mailing list
> >>>
> >>> * Please search the archive at
> >>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> >>> posting!
> >>>
> >>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >>>
> >>> * For (un)subscribe requests visit
> >>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> >>> send a
> >>> mail to gmx-users-request at gromacs.org.
> >
> >
> > --
> > ==================================================
> >
> > 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
> >
> > ==================================================
> > --
> > Gromacs Users mailing list
> >
> > * Please search the archive at
> > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
> >
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
> > * For (un)subscribe requests visit
> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a
> > mail to gmx-users-request at gromacs.org.
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>