[Pw_forum] PW is frozen

[TRINITE Virginie]

Dear all

I find a very strange behavior of pw.x:
                It run endlessly without output anything and with no error, and I have  finally to kill it
This behavior appears by using 6 processors for Iron calculation, everything is fine otherwise (and I made number of other calculations with no error with this version).
I use espresso-5.0 with ifort 12.1.3 and openmpi-1.4.3i
I find a very old post on this type of problem: http://www.democritos.it/pipermail/pw_forum/2004-December/001733.html.
It seems related to vc-relax so I try scf calculation but it was the same
I try to run directly on the node without the queue system and it was the same: no output on the screen for 6 processors and 1 2 3 4 5 processors is fine.
I try to clean all the files, in case pw.x try to read old wavefunctions, without solving the problem.
In the old post Paolo Giannozzi explain that can be due to MPI or MKL library. But since 2004, I think that the libraries has a lot evolved?

Did anybody have this kind of behavior?

Here is the input file:

&control
    calculation   = 'scf'
    verbosity     ='high'
    outdir        = '/tmp/'
    prefix        = 'feFM'
    tstress       =.true.
    tprnfor       =.true.
    forc_conv_thr = 1e-4
 /
 &system
    ibrav         = 2
    celldm(1)     = 6.524
    nat           = 1
    ntyp          = 1
    ecutwfc       = 120
    ecutrho       = 1200
    starting_magnetization(1)=1.0
    occupations   = 'smearing'
    smearing      = 'mp'
    degauss       = 0.01
    nspin         =2
 /
 &electrons
    conv_thr      = 1e-8
    mixing_beta   = 0.7
    mixing_mode   ='plain'
 /
ATOMIC_SPECIES
  Fe   55.847    Fe.pw91-spn-rrkjus.UPF
ATOMIC_POSITIONS alat
  Fe    0.0  0.0  0.0
K_POINTS {automatic}
  8 8 8  1 1 1

Dr Virginie Trinite
---------------------------------------
Modeling Infra-Red Lasers and Detectors
III-V Lab THALES Research&  Technology, France

Campus Polytechnique
1, avenue Augustin Fresnel 91767 Palaiseau cedex France
Phone : +33 (0)1.69.41.58.20


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20130329/9561ba08/attachment.html>