[QE-users] libxc needed for this functional - QE bulid with libxc

[Husak Michal]

Hi


Bellow is the MPI QE error we get from different compiler and different MPI implementation with any data (including test suite for sure). Please let me know if anybody have any glue 😎. I was trying to trace the issue inside source code with no result ...


M:\qe_project\QE_7_0\LALA_scf\MPI4>set PMIX_MCA_gds=hash

M:\qe_project\QE_7_0\LALA_scf\MPI4>orterun -np 4 pw.exe  0<pwscf.in 1>pwscf.out

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     Error in routine  good_fft_order (1):
  invalid np
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

     stopping ...

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     Error in routine  good_fft_order (1):
  invalid np
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

     stopping ...

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     Error in routine  good_fft_order (1):
  invalid np
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

     stopping ...

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     Error in routine  good_fft_order (1):
  invalid np
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

     stopping ...
--------------------------------------------------------------------------
Sorry!  You were supposed to get help about:
    mpi-abort
But I couldn't open the help file:
    /usr/share/openmpi/help-mpi-api.txt: No such file or directory.  Sorry!



________________________________
From: Paolo Giannozzi <paolo.giannozzi at uniud.it>
Sent: Tuesday, December 19, 2023 12:01 PM
To: Quantum ESPRESSO users Forum
Cc: Husak Michal
Subject: Re: [QE-users] libxc needed for this functional - QE bulid with libxc

On 12/17/23 10:16, Husak Michal via users wrote:

> We had de-facto stopped QE usage, because from the version 7.0 QE support for MPI parallelization is non functional

funny: nobody else has reported that QE hasn't been working on parallel
machines since two years

P.
--
Paolo Giannozzi, DMIF, Univ. Udine, Italy
*** AVAILABLE POST-DOC POSITION:
*** https://physicslab.uniud.it/persone/paolo-giannozzi/advert