[QE-users] pw.x prints the first few lines and segmentation fault

Paolo Giannozzi p.giannozzi at gmail.com
Sun Mar 24 20:53:18 CET 2019


Hi Lorenzo, the patch will be made available but will not go into the
development branch: there is nothing wrong with the code, it's a problem of
some old-ish compilers.

Paolo

On Sun, Mar 24, 2019 at 3:26 PM Lorenzo Paulatto <paulatz at gmail.com> wrote:

> Hello Pietro,
> Is this patch already in the git developer branch? Because I am getting a
> similar segfault on Bluegene with xlf, and I would like to test if the
> patch fixes it. Cheers
>
> --
> Lorenzo Paulatto
> Written on a virtual keyboard with real fingers
>
> On Fri, 22 Mar 2019, 17:16 Pietro Davide Delugas, <pdelugas at sissa.it>
> wrote:
>
>> Hello
>> could you try the attached  patch  ?
>>
>> copy the file in the q-e topdir and type
>>
>> patch -p1 --merge < patch_old_intel
>>
>> in case you wanted to undo the patch
>>
>> patch -R -p1 --merge < patch_old_intel
>>
>> please let me know  if it works
>> regards  Pietro
>>
>> On 03/22/2019 04:56 PM, Paolo Giannozzi wrote:
>>
>> Some versions of the intel compiler miscompile the recently modified XML
>> code. A patch will be released soon. Your make.inc is perfectly fine
>>
>> Paolo
>>
>>
>> On Fri, Mar 22, 2019 at 4:42 PM Hari Paudyal <hpaudya1 at binghamton.edu>
>> wrote:
>>
>>> Dear experts,
>>>
>>> After the successful installation of qe-v6.4, pw.x gives segmentation
>>> fault after when it prints the first few lines as;
>>>
>>>      Parallel version (MPI), running on     1 processors
>>>
>>>      MPI processes distributed on     1 nodes
>>>      Waiting for input...
>>>      Reading input from standard input
>>>
>>>      Current dimensions of program PWSCF are:
>>>      Max number of different atomic species (ntypx) = 10
>>>      Max number of k-points (npk) =  40000
>>>      Max angular momentum in pseudopotentials (lmaxx) =  3
>>> forrtl: severe (174): SIGSEGV, segmentation fault occurred
>>>
>>> I had no problem compiling/running all the previous versions (even some
>>> of the development versions I have installed on 15Oct2018, 20Jan2019).
>>> Since I am not an expert, I do not know what has been changed in the
>>> process of installation in this new released version v6.4, but I am pretty
>>> sure that nothing has been changed in my cluster regarding
>>> compilers/libraries. For info, I am using intel/composer_xe_2013.1.117.
>>>
>>> Further, If I look more carefully to the make.inc files generated after
>>> ./configure between qe-v6.3 and qe-v6.4, I found the following differences;
>>>
>>> # TOPDIR = /home/hari/qe-6.3
>>> DFLAGS         =  -D__FFTW -D__MPI
>>> LIBS           = $(CUDA_LIBS) $(SCALAPACK_LIBS) $(LAPACK_LIBS)
>>> $(FOX_LIB) $(FFT_LIBS) $(BLAS_LIBS) $(MPI_LIBS) $(MASS_LIBS) $(HDF5_LIB)
>>> $(LD_LIBS)
>>> # Install directory - not currently used
>>>
>>> # TOPDIR = /home/hari/qe-6.4
>>> DFLAGS         =  -D__DFTI -D__MPI
>>> LIBXC_LIBS     =
>>> QELIBS         = $(CUDA_LIBS) $(SCALAPACK_LIBS) $(LAPACK_LIBS)
>>> $(FOX_LIB) $(FFT_LIBS) $(BLAS_LIBS) $(MPI_LIBS) $(MASS_LIBS) $(HDF5_LIB)
>>> $(LIBXC_LIBS) $(LD_LIBS)
>>> # Install directory - "make install" copies *.x executables there
>>>
>>> If I have understood correctly, the problem is due to the MPI library.
>>> Also, the pw_user_guide says "MPI libraries need to be properly configured"
>>> (pw_user_guide-v6.4_page_20). But, why the previous versions work fine and
>>> the problem occurred only on this new released version v6.4? How this can
>>> be solved?
>>>
>>> I will be happy to see any comments/suggestions from experts.
>>>
>>> Thank you,
>>> Hari Paudyal
>>> SUNY-Binghamton
>>>
>>>
>>> _______________________________________________
>>> users mailing list
>>> users at lists.quantum-espresso.org
>>> https://lists.quantum-espresso.org/mailman/listinfo/users
>>
>>
>>
>> --
>> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
>> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
>> Phone +39-0432-558216, fax +39-0432-558222
>>
>>
>>
>> _______________________________________________
>> users mailing listusers at lists.quantum-espresso.orghttps://lists.quantum-espresso.org/mailman/listinfo/users
>>
>>
>> _______________________________________________
>> users mailing list
>> users at lists.quantum-espresso.org
>> https://lists.quantum-espresso.org/mailman/listinfo/users
>
> _______________________________________________
> users mailing list
> users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users



-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20190324/a563e1ba/attachment.html>


More information about the users mailing list