[gmx-users] Restart simulation from checkpoint file with fewer nodes

Mark Abraham mark.j.abraham at gmail.com
Sun May 15 12:06:54 CEST 2016 

Hi,

Yes, that's one way to work around the problem. In some places, a module
subsystem can be used to take care of the selection automatically, but you
don't want to set one up for just you to use.

Mark

On Sun, May 15, 2016 at 11:48 AM <jkrieger at mrc-lmb.cam.ac.uk> wrote:

> Thanks Mark,
>
> My sysadmins have let me install my own GROMACS versions and have not
> informed me of any such mechanism. Would you suggest I qrsh into a node of
> each type and build an mdrun-only version on each? I'd then select a
> particular node type for a submit script with the relevant mdrun.
>
> Many thanks
> James
>
> > Hi,
> >
> > On Sat, May 14, 2016 at 1:09 PM <jkrieger at mrc-lmb.cam.ac.uk> wrote:
> >
> >> In case it's relevant/interesting to anyone, here are the details on our
> >> cluster nodes:
> >>
> >> nodes                   #       model           # cores         cpu
> >> model
> >>   RAM   node_type
> >> fmb01 - fmb33           33      IBM HS21XM      8               3 GHz
> >> Xeon
> >> E5450
> >>  16GB   hs21
> >> fmb34 - fmb42           9       IBM HS22        8               2.4 GHz
> >> Xeon E5530
> >> 16GB    hs22
> >> fmb43 - fmb88           45      Dell PE M610    8               2.4 GHz
> >> Xeon E5530
> >>      16GB       m610
> >> fmb88 - fmb90           3       Dell PE M610+   12              3.4 GHz
> >> Xeon X5690
> >>       48GB      m610+
> >> fmb91 - fmb202          112     Dell PE M620    24 (HT)         2.9 GHz
> >> Xeon E5-2667
> >>        64GB     m620
> >> fmb203 - fmb279         77      Dell PE M620    24 (HT)         3.5 GHz
> >> Xeon E5-2643 v2         64GB
> >>         m620+
> >> fmb280 - fmb359         80      Dell PE M630    24 (HT)         3.4 GHz
> >> Xeon E5-2643 v3         64GB
> >>         m630
> >>
> >> I could only run GROMACS 4.6.2 on the last three node types and I
> >> believe
> >> the same is true for 5.0.4
> >>
> >
> > Sure. GROMACS is designed to target whichever hardware was selected at
> > configure time, which your sysadmins for such a heterogeneous cluster
> > should have documented somewhere. They should also be making available to
> > you a mechanism to target your jobs to nodes where they can run programs
> > that use the hardware efficiently, or providing GROMACS installations
> that
> > work regardless of which node you are actually on. You might like to
> > respectfully remind them of the things we say at
> >
> http://manual.gromacs.org/documentation/5.1.2/install-guide/index.html#portability-aspects
> > (These thoughts are common to earlier versions also.)
> >
> > Mark
> >
> >
> > Best wishes
> >> James
> >>
> >> > I have found that only some kinds of nodes on our cluster work for
> >> gromacs
> >> > 4.6 (the ones we call m620, m620+ and m630 but not others - I can
> >> check
> >> > the details tomorrow). I haven't tested it again now I'm using 5.0 so
> >> > don't know if that's still an issue but if it is it could explain why
> >> your
> >> > restart failed even and the initial run didn't.
> >> >
> >> >> thanks a lot for your fast response.
> >> >>
> >> >> I have tried it, and it failed. I ask in this forum just to make
> >> sure.
> >> >> However, there was something in my cluster that probably make it
> >> failed.
> >> >> I'll handle it first and then retry to restart again.
> >> >>
> >> >> Regards,
> >> >>
> >> >> Husen
> >> >>
> >> >> On Sat, May 14, 2016 at 7:58 AM, Justin Lemkul <jalemkul at vt.edu>
> >> wrote:
> >> >>
> >> >>>
> >> >>>
> >> >>> On 5/13/16 8:53 PM, Husen R wrote:
> >> >>>
> >> >>>> Dear all
> >> >>>>
> >> >>>> Does simulation able to be restarted from checkpoint file with
> >> fewer
> >> >>>> nodes ?
> >> >>>> let's say, at the first time, I run simulation with 3 nodes. At
> >> >>>> running
> >> >>>> time, one of those nodes is crashed and the simulation is
> >> terminated.
> >> >>>>
> >> >>>> I want to restart that simulation immadiately based on checkpoint
> >> file
> >> >>>> with
> >> >>>> the remaining 2 nodes. does gromacs support such case ?
> >> >>>> I need help.
> >> >>>>
> >> >>>
> >> >>> Have you tried it?  It should work.  You will probably get a note
> >> about
> >> >>> the continuation not being exact due to a change in the number of
> >> >>> cores,
> >> >>> but the run should proceed fine.
> >> >>>
> >> >>> -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.
> >> >>>
> >> >> --
> >> >> 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.
> >> >
> >>
> >>
> >> --
> >> 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.
> >
>
>
> --
> 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.
>

More information about the gromacs.org_gmx-users mailing list