AMBER Archive (2009)

Subject: Re: [AMBER] amber10 parallel compiling error

From: nam kim (namkkim_at_gmail.com)
Date: Tue Jul 21 2009 - 12:06:32 CDT


I tested amber.parallel with a PBS script.
Please help!

Here is the error:
------------------------------------------------------------------------------------------
[compute-1-9.local:09524] MPI_ABORT invoked on rank 0 in communicator
MPI_COMM_WORLD with err
orcode 1
mpirun noticed that job rank 1 with PID 9525 on node compute-1-9.local
exited on signal 15 (T
erminated).
make[1]: *** [test.sander.BASIC] Error 1
make: *** [test.sander.BASIC.MPI] Error 2
[namkim_at_goddard test]$ less testmake2x8proc.o7613
export TESTsander=/share/apps/amber10/exe/sander.MPI; make test.sander.BASIC
make[1]: Entering directory `/share/apps/amber10/test'
cd cytosine && ./Run.cytosine
----------------------------------------------------------------------------------------------

On Mon, Jul 20, 2009 at 10:33 AM, Ross Walker<ross_at_rosswalker.co.uk> wrote:
> Hi Nam,
>
> Can you send the .out file produced in the rdc directory please so I can see
> where it crashes.
>
> All the best
> Ross
>
>> -----Original Message-----
>> From: amber-bounces_at_ambermd.org [mailto:amber-bounces_at_ambermd.org] On
>> Behalf Of nam kim
>> Sent: Monday, July 20, 2009 10:24 AM
>> To: amber_at_ambermd.org
>> Subject: Re: [AMBER] amber10 parallel compiling error
>>
>> Ross,
>>
>> I can compile amber10 without netCDF option in success.
>>  But, when I test it, there is Run.dip error.
>> My cluster use openmpi 1.2.8 with Infiniband connection and PBS.
>> Do you think I should use PBS when testing " setenv DO_PARALLEL
>> 'mpirun -np 4' make test.parallel.MM < /dev/null"?
>>
>>  Here is the error:
>>
>> cd rdc && ./Run.dip
>>  [MPI_ABORT invoked on rank 0 in
>>  communicator MPI_COMM_WORLD with errorcode 1
>>  mpirun noticed that job rank 1 with PID 31846 on node
>>  exited on signal 15 (Terminated).
>>  2 additional processes aborted (not shown)
>>  ./Run.dip:  Program error
>>  make[1]: *** [test.sander.BASIC] Error 1
>> make[1]: Leaving directory `/apps/amber10/test'
>>  make: *** [test.sander.BASIC.MPI] Error 2
>>
>>
>>   Many Thanks
>>  -Nam
>>
>> > On Sat, Jul 18, 2009 at 10:57 AM, vitor.felix<vitor.felix_at_ua.pt>
>> wrote:
>> >>
>> >>
>> >> Sent from my HTC Android phone.
>> >>
>> >> ----- Reply message -----
>> >> From: "nam kim" <namkkim_at_gmail.com>
>> >> Date: Sat, Jul 18, 2009 5:19 PM
>> >> Subject: [AMBER] amber10 parallel compiling error
>> >> To: <amber_at_ambermd.org>
>> >>
>> >> Ross,
>> >>
>> >> I will try when I get back to the office.
>> >>
>> >> Thanks
>> >> -Nam
>> >>
>> >>
>> >> Hi Nam,
>> >>
>> >>> I have successfully compiled amber10 with g95 in serial, but for
>> the
>> >>> parallel mode, I failed with the error message:
>> >>> /share/apps/openmpi/bin/mpif90 -c -O3 -fno-second-underscore
>> >>> -ffree-form -o ncsu-umbrella.o _
>> >>>
>> >>> ncsu-umbrella.f
>> >>>
>> >>> Fatal Error: Reading module netcdf at line 3 column 1: Expected
>> left
>> >>> parenthesis
>> >>>
>> >>> make[1]: *** [ncsu-umbrella.o] Error 1
>> >>>
>> >>> make[1]: Leaving directory `/share/apps/amber10/src/sander'
>> >>>
>> >>> make: *** [parallel] Error 2
>> >>
>> >> I have not seen this error before but a couple of things to try.
>> >>
>> >> 1) As a work around you could compile without NetCDF support:
>> >>
>> >> ../configure_amber -mpich2 -nobintraj g95
>> >>
>> >> 2) If you want to keep the netcdf support try doing a full clean of
>> both
>> >> amber and ambertools before configuring for parallel. E.g.
>> >>
>> >> cd $AMBERHOME/src/
>> >> make -f Makefile_at clean
>> >> make clean
>> >> ./configure_amber -mpich2 g95
>> >> make parallel
>> >>
>> >> Obviously adjusting the options (mpich2) above for the MPI version
>> you are
>> >> using.
>> >>
>> >> Good luck,
>> >> Ross
>> >>
>> >> /\
>> >> \/
>> >> |\oss Walker
>> >>
>> >> | Assistant Research Professor |
>> >> | San Diego Supercomputer Center |
>> >> | Tel: +1 858 822 0854 | EMail:- ross_at_rosswalker.co.uk |
>> >> | http://www.rosswalker.co.uk | PGP Key available on request |
>> >>
>> >> Note: Electronic Mail is not secure, has no guarantee of delivery,
>> may not
>> >> be read every day, and should not be used for urgent or sensitive
>> issues.
>> >>
>> >> _______________________________________________
>> >> AMBER mailing list
>> >> AMBER_at_ambermd.org
>> >> http://lists.ambermd.org/mailman/listinfo/amber
>> >>
>> >
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER_at_ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>
>
> _______________________________________________
> AMBER mailing list
> AMBER_at_ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber