Thanks for all valuable advices.
I recompiled the code with diferent libraries and now, the code is
working fine.
As a reference, I put my library information before/after change.
---------> old one
BLAS_LIBS=/opt/apps/intel10/gotoblas/1.30/libgoto_ilp64_mp.a
LAPACK_LIBS=$(TACC_MKL_LIB)/libmkl_lapack.a
BLACS_LIBS=$(TACC_MKL_LIB)/libmkl_blacs_intelmpi20_ilp64.a
SCALAPACK_LIBS=$(TACC_MKL_LIB)/libmkl_scalapack_ilp64.a
------> replaced library
LIBS= -L$(TACC_MKL_LIB) -lmkl -llapack -lguide
$(TACC_SCALAPACK_LIB)/libscalapack.a
$(TACC_SCALAPACK_LIB)/blacsF77init_MPI-LINUX-0.a
$(TACC_SCALAPACK_LIB)/blacs_MPI-LINUX-0.a
$(TACC_GOTOBLAS_LIB)/libgoto_ilp64_mp.a
Thanks all!!
Kyoung
Quoting [email protected]:
Dear Marcos,
Thanks for your valuable advices.
As you mentioned, I checked the fdf file, but all input parameters are
the same in the file in the example included in the siesta package.
Also,currently, the compiler I'm using is mpif90 for the parallel run.
I attached the arch.make file created for my working system.
Thanks for any advice in advance !!
Kyoung
Quoting Marcos Veríssimo Alves <[email protected]>:
Hi Kyoung,
On Thu, Dec 3, 2009 at 11:47 PM, <[email protected]> wrote:
Hello, all Siesta users!!
I'm a new user of siesta. Recently, I just compiled siesta-3.0-b without
any problem.
Congratulations... most people have lots of trouble arriving to this step.
:)
But, I got some problems when I ran small test jobs.
The error message is following (for 4 node parallel run)
forrtl: severe (174): SIGSEGV, segmentation fault occurred
forrtl: severe (174): SIGSEGV, segmentation fault occurred
forrtl: severe (174): SIGSEGV, segmentation fault occurred
forrtl: severe (174): SIGSEGV, segmentation fault occurred
This error was the same when I tried with single node.
The test job is pretty small file with just 2 atoms.
Since compilation was done successfully, I'm not sure why this problem
occurs.
Well, as you will notice in computational work, producing an executable is
one thing. Producing an executable that actually works is a whole other
story... Many different factors could be leading to this, and the
information you are providing is too little. Often, the source of the
problem is in the compiler you are using, or in your compilation options. My
suggestion is that you provide as much information as you can on the
problem. This would include typically:
1) Compiler (with version number: ifort 9.0, 9.1, 10.0, 10.1, 11.0, 11.1?),
mathematical libraries?
2) arch.make so we can see what compilation options you are choosing for
your build of siesta
3) fdf file to allow for checking if there's anything weird in your input.
And it's always advisable to check the list archives for possible problems
about the compiler version, math libraries, and the particular system you
are working on - most probably someone has had the same problems as you are
having now.
Cheers,
Marcos
