cp2k stops execuation without being terminated: memory leaks

[Sangkha Borah]

Hi everyone,

       I need some help regarding cp2k installation in parallel on a 
supercomputer  with specifications given in the link: 
https://www.iitg.ernet.in/param-ishan . 
I am trying to install the popt version with the following arch file

# 
------------------------------------------------------------------------------------------------------------------------------------------------------
# Arch file targeting Linux 64-bit using the Intel tool chain
ROOT := $(dir $(word $(words $(MAKEFILE_LIST)), $(MAKEFILE_LIST)))
ACC := 0
SMP := 0

include $(ROOT)Linux-x86-64-intel-mic.psmp
INTEL_MKL =/cm/shared/apps/intel/compilers_and_libraries/2016.2.181/mkl
INTEL_INC = $(INTEL_MKL)/include/fftw
INTEL_LIB = $(INTEL_MKL)/lib/intel64
CC = mpiicc
FC = mpiifort
LD = mpiifort
AR = /usr/bin/ar -r
DFLAGS =-D__FFTW3 -D__FFTSG -D__FFTMKL -D__parallel -D__BLACS -D__SCALAPACK 
-D__MKL 
CPPFLAGS = -C -traditional $(DFLAGS) -I$(INTEL_INC)
#FCFLAGS = $(DFLAGS) -I$(INTEL_INC) -O2 -xHost -heap-arrays 64 -fpp -free 
–unroll -pc 64 -convert big_endian
FCFLAGS = $(DFLAGS) -I$(INTEL_INC) -O2 -xHost -fpp -free -pc 64 -convert 
big_endian
LDFLAGS = $(FCFLAGS) -static-intel

MKL_LIB=/cm/shared/apps/intel/compilers_and_libraries/2016.2.181/mkl/lib/intel64
LIBS = -L$(INTEL_LIB) -lmkl_intel_lp64   -lmkl_sequential  -lmkl_core 
 -lmkl_scalapack_lp64  -lmkl_blacs_intelmpi_lp64
OBJECTS_ARCHITECTURE = machine_intel.o
optimize_ri_basis.o: mp2_optimize_ri_basis.F
        $(FC) -c $(FCFLAGS) $<
# 
------------------------------------------------------------------------------------------------------------------------------------------------------
Everything goes fine as far as the installation is concerned, also runs 
fine for about 1 ps with  time step of 0.5 fs for integration, and then 
stops, every time, at the following line:
" Number of independent orbital functions:                                 
  2466" 

Following previous posts and whatsoever I got from internet surfing, I 
tried to modify the arch file, but without any success. For example, (i) I 
installed with a different mkl version (parallel_studio_xe),
(ii) tried combinations of the last few lines of the above scripts, (iii) 
compiled without fftw (iv) changed FCFLAGS, (v) tried installing different 
versions of cp2k, etc.
I guess it must be due to some kinds of memory leaks.  

Note: I have followed, for example, the following web links for making my 
arch file, along with cp2k INSTALL file
1. http://www.hpcadvisorycouncil.com/pdf/CP2K_Best_Practices.pdf 
<http://www.hpcadvisorycouncil.com/pdf/CP2K_Best_Practices.pdf>
2.
 http://hpckp.org/index.php/articles/best-practices/94-how-to-build-cp2k-with-intel-compiler-12-1-openmpi-1-4-3-and-infiniband-support 
<http://hpckp.org/index.php/articles/best-practices/94-how-to-build-cp2k-with-intel-compiler-12-1-openmpi-1-4-3-and-infiniband-support>
3. http://cp2k-www.epcc.ed.ac.uk
4. http://erg.biophys.msu.ru/wordpress/archives/587  
<http://erg.biophys.msu.ru/wordpress/archives/587>

N.B. I have only basic ideas for installation and I found that the above 
settings work for our machine.  I have seen that without the flag " 
-convert big_endian" 
the program does not execute (this is based on link 1 above)

Thanks in advance.

Best regards,
Sangkha Borah,
Research scholar, 
Dept. of Physics, IIT Guwahati, India



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20170921/bd98e417/attachment.htm>