[Pw_forum] Wired and large (4Ry) total energy difference on different machines

[Fenggong Wang]

Dear Prof. Giannozzi:
         Thank you very much.
         I checked the diff of the outputs. The first part (before
calculating) of the output is essentially the same except some trivial
information. The main difference starts from the first iteration. In
output1, it is:
Computed dipole along edir(3) :
        Elec. dipole         -0.7303 Ry au,          -1.8561 Debye
        Ion. dipole          -0.6420 Ry au,        -1.6317 Debye
        Dipole               42.0156 Ry au,        106.7936 Debye
        Dipole field          0.0883 Ry au

        Potential amp.      -19.3274 Ry
        Total length        109.4531 bohr

     Starting wfc are random

     total cpu time spent up to now is        9.8 secs

     per-process dynamical memory:   163.1 Mb

     Self-consistent Calculation

     iteration #  1     ecut=    50.00 Ry     beta=0.10
     Davidson diagonalization with overlap
     ethr =  1.00E-02,  avg # of iterations = 17.1

     negative rho (up, down):  0.195E-03 0.000E+00

     Adding external electric field

     Computed dipole along edir(3) :
        Elec. dipole          3.4941 Ry au,           8.8811 Debye
        Ion. dipole          -0.6420 Ry au,        -1.6317 Debye
        Dipole            -1968.2460 Ry au,      -5002.8089 Debye
        Dipole field         -4.1360 Ry au

        Potential amp.      905.4020 Ry
        Total length        109.4531 bohr


     total cpu time spent up to now is      178.7 secs

while in output2 it is:

    Computed dipole along edir(3) :
        Elec. dipole         -0.7303 Ry au,          -1.8561 Debye
        Ion. dipole          -0.6420 Ry au,        -1.6317 Debye
        Dipole               42.0156 Ry au,        106.7936 Debye
        Dipole field          0.0883 Ry au

        Potential amp.      -19.3274 Ry
        Total length        109.4531 bohr

     Starting wfc are random

     total cpu time spent up to now is        4.5 secs

     per-process dynamical memory:    24.2 Mb

     Self-consistent Calculation

     iteration #  1     ecut=    50.00 Ry     beta=0.10
     Davidson diagonalization with overlap
     ethr =  1.00E-02,  avg # of iterations = 18.5

     negative rho (up, down):  0.196E-03 0.000E+00

     Adding external electric field

     Computed dipole along edir(3) :
        Elec. dipole          3.4862 Ry au,           8.8612 Debye
        Ion. dipole          -0.6420 Ry au,        -1.6317 Debye
        Dipole            -1964.5250 Ry au,      -4993.3512 Debye
        Dipole field         -4.1282 Ry au

        Potential amp.      903.6903 Ry
        Total length        109.4531 bohr


     total cpu time spent up to now is       58.6 secs


  Thank you.

Fenggong
On Mon, Sep 17, 2012 at 4:27 PM, Paolo Giannozzi <giannozz at democritos.it>wrote:

>
> On Sep 17, 2012, at 22:13 , Fenggong Wang wrote:
>
> > Recently we got quite different total energy using exactly
> > the same input files for pwscf calcuation when running
> > on different machines.
>
> make a diff of the outputs, just to be sure that everything
> else is the same.
>
> Small differences between different machines/number of
> processors/compilers/libraries/phases of the moon are
> unavoidable. 4Ry is definitely too much.
>
> P.
> ---
> Paolo Giannozzi, Dept of Chemistry&Physics&Environment,
> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
> Phone +39-0432-558216, fax +39-0432-558222
>
>
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>



-- 
Dr. Fenggong Wang
Postdoctoral Researcher

Department of Chemistry
University of Pennsylvania
231 South 34th Street, BOX 314
Philadelphia, PA 19104-6323
e-mail: wangfenggong at gmail.com; fenggong at sas.upenn.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20120917/edd43b3a/attachment.html>