Hi,
Have you been able to successfully run the test cases in parallel?
-Dan
On Wednesday, March 13, 2013, David Winogradoff wrote:
> Hey Jason,
>
> I followed the directions on your wiki for installing amber 12,
> running ./patch_amber.py --update-tree until it returns that Amber12
> and AmberTools12 are up to date.
>
> The error messages from the supercomputer I'm using are,
>
>
> "[compute-f17-1.deepthought.umd.edu:21064] [[44012,1],39]
> ORTE_ERROR_LOG: Data unpack would read past end of buffer in file
> runtime/orte_init.c at line 132
> --------------------------------------------------------------------------
> It looks like MPI_INIT failed for some reason; your parallel process is
> likely to abort. There are many reasons that a parallel process can
> fail during MPI_INIT; some of which are due to configuration or environment
> problems. This failure appears to be an internal failure; here's some
> additional information (which may only be relevant to an Open MPI
> developer):
>
> ompi_mpi_init: orte_init failed
> --> Returned "Data unpack would read past end of buffer" (-26)
> instead of "Success" (0)
> --------------------------------------------------------------------------
> *** The MPI_Init() function was called before MPI_INIT was invoked.
> *** This is disallowed by the MPI standard.
> *** Your MPI job will now abort.
> [compute-f17-1.deepthought.umd.edu:21061] Abort before MPI_INIT
> completed successfully; not able to guarantee that all other processes
> were killed!"
>
>
> which I thought were generic openmpi error messages. For your
> reference, the version of openmpi I am using is openmpi16-gnu.
>
> I have also attempted to use the openmpi143-gnu version of openmpi,
> obtaining the error message,
>
>
> "[compute-f16-26.deepthought.umd.edu:30072] mca: base: component_find:
> unable to open
> /cell_root/software/openmpi/1.4.3/gnu46/sys/lib/openmpi/mca_btl_openib:
> perhaps a missing symbol, or compiled for a different version of Open
> MPI? (ignored)
> --------------------------------------------------------------------------
> MPI_ABORT was invoked on rank 14 in communicator MPI_COMM_WORLD
> with errorcode 1.
>
> NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
> You may or may not see output from other processes, depending on
> exactly when Open MPI kills them."
>
>
> My REMD simulation is 50 replicas at different temperatures in NVT
> ensembles attempting exchanges every 5 ps. The template for my mdin
> files is
>
> replica exchange
> &cntrl
> imin=0, irest=1, ntx=5,
> ntt=3, temp0=XXXXX, gamma_ln=2.0, ig=-1,
> nstlim=2500, dt=0.002,
> ntb=1, ntc=2, ntf=2,
> ntr=0, ntp=0,
> ntwr=100000, ntwx=2500, ntpr=2500,
> cut=12,
> numexchg=1000,
> /
> &wt type='END'
> /
>
>
>
> Hopefully this answers your questions, thanks,
> David
>
> On Mon, Mar 11, 2013 at 6:56 PM, Jason Swails <jason.swails.gmail.com<javascript:;>>
> wrote:
> > On Mon, Mar 11, 2013 at 6:34 PM, David Winogradoff <dwino218.gmail.com<javascript:;>
> >wrote:
> >
> >> I am currently working on a project using amber in parallel to conduct
> >> a long REMD run. A few weeks ago, using the same input file that
> >> worked before (modified to start from where the previous simulation
> >> ended), my parallel run failed with a generic openmpi error message.
> >> After some initial hesitation, I decided to delete my installation of
> >> amber 12 and recompile and rebuild. I have run 'make test'
> >> successfully with my new serial and parallel versions of amber 12, but
> >> my REMD simulation still won't run. After many unsuccessful attempts
> >> to run, I decided to look at some of my previous output files. To my
> >> surprise, the older output files had amber 11 at the top. Even though
> >> my new simulations won't run, they do write the header, which says
> >> amber 12.
> >>
> >> I did see that it is possible to have some combination of amber 11 and
> >> amber 12, but I never installed amber 11 and do not have the tar files
> >> to do so. Is there any way to install amber 11 with amber 12 tar files
> >> (accidentally)? My hypothesis is that some requirement, such as my
> >> version of gnu, was out of date during my previous installation,
> >> preventing the complete installation of amber 12. I have tried to
> >> recreate the mistake, since I didn't think I could make files meant
> >> for amber 11 to work with amber 12, without success. The only solution
> >> I can think of right now is to purchase amber 11 and create a hybrid
> >> form of amber 11 and amber 12 that will hopefully run my simulations
> >> once more.
> >>
> >
> > When Amber 12 was released, it actually still said "Amber 11" at the top.
> > That was updated in bugfix.2 for Amber 12. Do you have all bug fixes
> > applied?
> >
> > What error messages are you getting? And what exactly are you doing?
> The
> > better way forward, IMO, is to fix Amber 12... What kind of REMD are you
> > doing?
> >
> > All the best,
> > Jason
> >
> > --
> > Jason M. Swails
> > Quantum Theory Project,
> > University of Florida
> > Ph.D. Candidate
> > 352-392-4032
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org <javascript:;>
> > http://lists.ambermd.org/mailman/listinfo/amber
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org <javascript:;>
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
-------------------------
Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 201
Salt Lake City, UT 84112-5820
http://home.chpc.utah.edu/~cheatham/
(801) 587-9652
(801) 585-9119 (Fax)
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Mar 13 2013 - 12:30:02 PDT
