Re: [AMBER] Amber11 parallel testing error

From: Zhu, James <James.Zhu.ARS.USDA.GOV>
Date: Wed, 2 Feb 2011 15:53:46 -0500

See the following

1) What bugfixes (if any) have been applied to your Amber11 source?

Both bugfix.all files for Amber11 and AmberTools1.4 were downloaded to Amber11 folder and applied according the instructions


2) What compilers (and version) were used to compile?

Openmpi-1.4.1

3) What parallel library was used (e.g. mpich1, mpich2, openmpi etc)

Openmpi-1.4.1/lib (? I do not know where to look for)

and what version, along with what compilers were used to compile the
parallel library.

(I got the following from the output of ./configure -mpi gnu)
Obtaining the gnu suite version:
      gcc -v
The version is 4.1.2

Testing the gcc compiler:
     gcc -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -o testp testp.c
OK

Validating the C++ compiler version:
     g++ -v
The version is 4.1.2
OK

Testing the gfortran compiler:
     gfortran -O0 -o testp testp.f
OK

Testing mixed C/Fortran compilation:
     gcc -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -c -o testp.c.o testp.c
     gfortran -O0 -c -o testp.f.o testp.f
     gcc -o testp testp.c.o testp.f.o -lgfortran
OK


4) Are there any error messages in the output files produced by the
test (neb_gb_partial_XX.out files)?

It is a long file and I do not know where to check.

Also, can you repeat just the NEB tests with 8 threads (type 'make
test.neb') and see if the error still occurs?

I ran it but it is still running after 15 min. How long does this take?

I also got this warning when I ran ./configure -mpi gnu and I checked that the files as shown in the following.

[jzhu.jzhu-desktop src]$ ./configure -mpi gnu
Your AMBERHOME environment variable should be set to /home/jzhu/bin/amber11
 
Warning: the X11 libraries are not in the usual location !
    To search for them try the command: locate libXt
    On new Fedora OS's install the libXt-devel libXext-devel
    libX11-devel libICE-devel libSM-devel packages.
    On old Fedora OS's install the xorg-x11-devel package.
    On RedHat OS's install the XFree86-devel package.
    On Ubuntu OS's install the xorg-dev package.
    For the moment Amber will be configured not to build XLEaP.

[jzhu.jzhu-desktop src]$ locate libXt
/usr/lib/libXt.so.6
/usr/lib/libXt.so.6.0.0
/usr/lib/libXtst.so.6
/usr/lib/libXtst.so.6.1.0
/usr/lib64/libXt.so.6
/usr/lib64/libXt.so.6.0.0
/usr/lib64/libXtst.so.6
/usr/lib64/libXtst.so.6.1.0
/usr/share/doc/libXt-1.0.2
/usr/share/doc/libXtst-1.0.1
/usr/share/doc/libXt-1.0.2/AUTHORS
/usr/share/doc/libXt-1.0.2/COPYING
/usr/share/doc/libXt-1.0.2/ChangeLog
/usr/share/doc/libXt-1.0.2/INSTALL
/usr/share/doc/libXt-1.0.2/NEWS
/usr/share/doc/libXt-1.0.2/README
/usr/share/doc/libXtst-1.0.1/AUTHORS
/usr/share/doc/libXtst-1.0.1/COPYING
/usr/share/doc/libXtst-1.0.1/ChangeLog
/usr/share/doc/libXtst-1.0.1/INSTALL
/usr/share/doc/libXtst-1.0.1/README


-----Original Message-----
From: Daniel Roe [mailto:daniel.r.roe.gmail.com]
Sent: Wednesday, February 02, 2011 1:04 PM
To: AMBER Mailing List
Subject: Re: [AMBER] Amber11 parallel testing error

Hi,

If possible, could you provide some more information? In particular:

1) What bugfixes (if any) have been applied to your Amber11 source?
2) What compilers (and version) were used to compile?
3) What parallel library was used (e.g. mpich1, mpich2, openmpi etc)
and what version, along with what compilers were used to compile the
parallel library.
4) Are there any error messages in the output files produced by the
test (neb_gb_partial_XX.out files)?

Also, can you repeat just the NEB tests with 8 threads (type 'make
test.neb') and see if the error still occurs?

Thanks!

-Dan

On Tue, Feb 1, 2011 at 11:36 AM, Zhu, James <James.Zhu.ars.usda.gov> wrote:
> I ran test.parallel with -np 2 option and the test was fine.  When I
> changed to -np 4 using the following:
>
>
>
> export DO_PARALLEL='mpirun -np 4'
>
> make test.parallel
>
>
>
> It stopped at the step shown in the following.  I let it ran over the
> weekend but it did not move on to the next test.  I hit the return key
> and the computer output Broken pipe.  What is wrong with the parallel
> version?
>
>
>
> James
>
>
>
> ==============================================================
>
> make[2]: Leaving directory `/home/jzhu/bin/amber11/test'
>
> cd neb-testcases/neb_gb_partial && ./Run.neb_gb_partial
>
>
>
> Running multisander version of sander Amber11
>
>    Total processors =     4
>
>    Number of groups =     4
>
>
>
> Broken pipe
>
> program  error.
>
> make[1]: [test.neb] Error 1 (ignored)
>
> cd neb-testcases/neb_gb_full && ./Run.neb_gb_full
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>

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

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Feb 02 2011 - 13:00:06 PST
Custom Search