Abort (core dumped)

Mistry__Mital · ‎02-15-2011

Dear All,

I am trying to run my code but after 2 loop iteration it gives "Abort (core dumped) " error.
I have copied output and the core file output.

Thank you,

Mital

****************************************************************************************
mcmistry/EHD> ./xx.out DONE1

DONE Am2

DONE CALCULATING

0

0.00000000000000

0

1.207496986890922E+016

Abort (core dumped)

mcmistry/EHD> gdb app core.25743

GNU gdb 6.6

GDB is free software, covered by the GNU General Public License, and you are

welcome to change it and/or distribute copies of it under certain conditions.

Type "show copying" to see the conditions.

There is absolutely no warranty for GDB. Type "show warranty" for details.

This GDB was configured as "ia64-suse-linux"...

app: No such file or directory.

(no debugging symbols found)

Using host libthread_db library "/lib/libthread_db.so.1".

Core was generated by `./xx.out'.

Program terminated with signal 6, Aborted.

#0 0xa000000000010620 in __kernel_syscall_via_break ()

(gdb)
************************************************************************************************

Ron_Green · ‎02-15-2011

I can't compile this code, it's missing subroutines dgetrf and dgetrs.

Also, you did not say what compiler version you are using, nor what compiler options you used to compile.

Read this article: http://software.intel.com/en-us/articles/determining-root-cause-of-sigsegv-or-sigbus-errors/

Compile with -g -traceback.

Increase the stack limit with ulimit -s unlimited and run again, as mentioned in the above article. If you are not using OpenMP or -parallel, try using -heap-arrays

ron

Mistry__Mital · ‎02-15-2011

Thanks Ron,

Compiler: Intel-10.0.1.014

Compiler Option:

"gfortran -w -I../../include NCF.f90 -x none -L/home/mcmistry/EHD/libmkl_lapack.a -L/home/mcmistry/EHD/libmkl_solver_lp64.a /home/mcmistry/EHD/libmkl_intel_lp64.a -Wl,--start-group /home/mcmistry/EHD/libmkl_intel_thread.a /home/mcmistry/EHD/libmkl_core.a -Wl,--end-group /home/mcmistry/EHD/libguide.a -lpthread -lm -o xx.out
mcmistry/EHD> ./xx.out"

Few options are not necessary but I haven't changed as I am using MKL_Pardiso in my other program.

I also tried with limit stakesize unlimited. Let me try with -heap-arrays

Regards,
Mital

mecej4 · ‎02-15-2011

"I also tried with limit stakesize unlimited. Let me try with -heap-arrays"

That amounts to groping in the dark. You have a program that reads multiple input files. Most programs of that nature take execution paths that are data-dependent. Without knowing the contents of those files, it is futile to hunt for what caused the program to crash.

Mistry__Mital · ‎02-16-2011

Hi, I checked my output from DGETRF and DGETRS. That I am using to solve = , where A and B are known and m solving for X. All are square matrices of size (920,920) and all are full matrices. The result of is not correct (checked with MATLAB result). As I am solving same with Matlab. I am transfering my code from Matlab to FORTRAN. In matlab, I m using "" operator.

Is linking the problem?

As I am not able to give a path...

TimP · ‎02-16-2011

Please consult the link advisor on the MKL forum if you are seeing MKL link problems. You would normally need at least the intel_lp64 library inside the start-group .... end-group directives. It seems you must have an MKL recent enough to include libiomp5.a, which would be preferable to libguide.

Mistry__Mital · ‎02-16-2011

Thanks All,

Things are working fine.... I used DGESV instead of DGETRF and DGETRS!

Mital