Re: AMBER: Amber9 under SGI Prism (Altix) with Intel 8.0 error in pbpgb test

From: David McGiven <david.mcgiven.fusemail.com>
Date: Wed, 21 Jun 2006 17:54:44 +0200

Hi Roberto!

That is strange because I applied the bugfix.all patch! I just double
checked it now.

I now Installed the following compilers :

l_cc_p_9.0.021.tar.gz
l_fc_p_9.0.021.tar.gz

I recompiled amber 9 and now the pbpgb test (and all the other
ones as well) succeed!!

So then I suppose there was a problem with my Intel 8.1 compilers
(srange though...).

In the parallel version, the pbpgb test succeeds as well, but there are
some errors like these :

==============================================================
cd qmmm2/1NLN_periodic_lnk_atoms; ./Run.1NLN_MD_ntb2
sander.MPI(19313): unaligned access to 0x600000000008249c,
ip=0x20000000000f0d81 sander.MPI(19313): unaligned access to
0x600000000008249c, ip=0x20000000000f0d81 diffing
mdout.1NLN_MD_ntb2.save with mdout.1NLN_MD_ntb2 PASSED
==============================================================

The test succeeds. Should I then worry about that ?

==============================================================

  Unit 8 Error on OPEN: spcfw_pimd.top
MPI: On host incal, Program /opt/soft/amber9/exe/sander.PIMD.MPI, Rank
0, Process 20324 called MPI_Abort(<communicator>, 1)

MPI: --------stack traceback-------
sh: line 1: idb: command not found

MPI: -----stack traceback ends-----
MPI: MPI_COMM_WORLD rank 0 has terminated without calling MPI_Finalize()
MPI: aborting job
diffing spcfw_pimd.out.save with spcfw_pimd.out
possible FAILURE: check spcfw_pimd.out.dif
==============================================================

==============================================================
cd bintraj; ./Run.bintraj

sander and ptraj: test sander netCDF output and ptraj netCDF input
MPI: On host incal, Program /opt/soft/amber9/exe/sander.MPI, Rank 0,
Process 23221 called MPI_Abort(<communicator>, 1) MPI: --------stack
traceback------- sh: line 1: idb: command not found

MPI: -----stack traceback ends-----
MPI: MPI_COMM_WORLD rank 0 has terminated without calling MPI_Finalize()
MPI: aborting job
  ./Run.bintraj: Program error
make[1]: [test.sander.BASIC] Error 1 (ignored)
make[1]: Leaving directory `/opt/soft/amber9/test'
export TESTsander=/opt/soft/amber9/exe/sander.MPI; cd 4096wat;
./Run.column_fft diffing mdout.column_fft.save with mdout.column_fft
PASSED
==============================================================


And here this (at least) two strange errors. Any clues ?

Almost all other the tests run OK

Best Regards,
David McGiven

On Wed, 21 Jun 2006 10:10:44 -0400
Roberto Gomperts <roberto.sgi.com> wrote:

> Hi David,
>
> This problem (not altix specific) has been solved with bugfix #2.
> Please
> make sure you patch your source (preferably) with bugfix.all to
> resolve
> the latest known problems.
> Please let me know if there are any issues after you apply the patch.
>
> Thanks,
> Roberto
>
>
> David McGiven wrote:
> > Dear Amber users,
> >
> > We recently bought amber 9 and I compiled it in one of our SGI
> > machines.
> >
> > The Machine is an SGI Prism (Basically an Altix with Graphics)
> > running
> > SLES9 with SGI ProPack 4SP3 for Linux, Build 403r1-0512182217
> >
> > I compiled the amber9 source tarball with Intel 8.0 compilers :
> >
> > Those are the versions I used :
> >
> > l_cc_pc_8.1.032.tar.gz
> > l_fc_pc_8.1.028.tar.gz
> >
> > I ran ./configure -nopar sgi_altix
> > make serial
> >
> > The compilation was ok and then I started to run the tests.
> > Everything
> > was going fine till it reached the pbpgb test. It crashes with the
> > following error :
> >
> > ==============================================================
> > cd alp; ./Run.alp
> > diffing mdout.alp.save with mdout.alp
> > PASSED
> > ==============================================================
> > cd pb_pgb; ./Run.pbpgb
> > getRegFromUnwindContext: Can't get Gr0 from UnwindContext, using 0
> > forrtl: severe (174): SIGSEGV, segmentation fault occurred
> > Image PC Routine Line
> > Source sander 40000000000AC821 Unknown
> > Unknown Unknown sander 4000000000095300 Unknown
> >
> > Unknown Unknown sander 400000000008FA30 Unknown
> >
> > Unknown Unknown sander 40000000005EEBA0 Unknown
> > Unknown Unknown sander 40000000001CB390 Unknown
> > Unknown Unknown sander 4000000000177EF0
> > Unknown Unknown Unknown sander
> > 4000000000176730 Unknown Unknown Unknown sander
> >
> > 4000000000003E90 Unknown Unknown Unknown libc.so.6.1
> >
> > 2000000001AB9C50 Unknown Unknown Unknown sander
> > 4000000000003840 Unknown Unknown Unknown
> > ./Run.pbpgb: Program error make: *** [test.sander.BASIC] Error 1
> >
> >
> > At the same time, in another folder, I was compiling it with mpi
> > support
> >
> > I ran ./configure -mpi sgi_altix
> > make parallel
> >
> > Again, the compilation was ok and then I started to run the tests.
> > Everything was going fine till it reached the pbpgb test again!. It
> > crashes with the following error :
> >
> > ==============================================================
> > cd alp; ./Run.alp
> > diffing mdout.alp.save with mdout.alp
> > PASSED
> > ==============================================================
> > cd pb_pgb; ./Run.pbpgb
> > MPI: On host incal, Program /opt/soft/amber9-mpi-mkl/exe/sander.MPI,
> > Rank 0, Process 16286 received signal SIGSEGV(11)
> >
> >
> > MPI: --------stack traceback-------
> > sh: line 1: idb: command not found
> >
> > MPI: -----stack traceback ends-----
> > MPI: On host incal, Program /opt/soft/amber9-mpi-mkl/exe/sander.MPI,
> > Rank 0, Process 16286: Dumping core on signal SIGSEGV(11) into
> > directory
> > /opt/soft/amber9-mpi-mkl/test/pb_pgb MPI: MPI_COMM_WORLD rank 0 has
> > terminated without calling MPI_Finalize() MPI: aborting job
> > MPI: Received signal 11
> >
> > ./Run.pbpgb: Program error
> > make[1]: *** [test.sander.BASIC] Error 1
> > make[1]: Leaving directory `/opt/soft/amber9-mpi-mkl/test'
> > make: *** [test.sander.BASIC.MPI] Error 2
> >
> >
> > Best Regards,
> > David McGiven
> > -------------------------------------------------------------------
> > ----
> > The AMBER Mail Reflector
> > To post, send mail to amber.scripps.edu
> > To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
> >
> >
>
> --
> Roberto Gomperts VNET: 483-8851
> One Cabot Road, Suite 290 Phone: (978) 562-8851
> Hudson, MA 01749 Fax: (978) 562-7450
>
>
> ---------------------------------------------------------------------
> --
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Jun 25 2006 - 06:07:07 PDT
Custom Search