konstantin_kudin at yahoo.com
Fri Dec 3 18:36:41 CET 2004
I am not sure if this is the same thing, but anyway ...
The version 2.0 at some point also got frozen in a similar way during
the step of reading wavefunction files from 'file', if those existed
only on some nodes but not the others.
For example, if there were 8 nodes running (2 4-way boxes), and one of
them had *.wfc1, *.wfc2, *.wfc3, *.wfc4 on the local temporary space,
and the other node did NOT have *.wfc5-*.wfc8, then the code would get
frozen while trying to read the wavefunctions.
Perhaps, somehow the nodes get moved around and the *.wfcN files do
not correspond to the earlier order?
--- Sergey Lisenkov <proffess at yandex.ru> wrote:
> Dear PWscf authors anf users,
> After yersteday discussions about using local disks for temporary
> files I use them. But I got very strange problem. If I use more than
> 2 processors (1 nodes) after second step (see output) the code is
> frozen, but does not stop:
> Total force = 0.014233 Total SCF correction =
> Entering Dynamics: iteration = 2
> time = 0.01162 pico-seconds
> ATOMIC_POSITIONS (alat)
> Al 0.500000000 0.500000000 -2.118003560
> Al 0.000000000 0.000000000 -1.414499206
> Al 0.500000000 0.500000000 -0.706281386
> Al 0.000000000 0.000000000 0.000000000
> Al 0.500000000 0.500000000 0.706281386
> Al 0.000000000 0.000000000 1.414499206
> Al 0.500000000 0.500000000 2.118003560
> kinetic energy (Ekin) = 0.00014120 Ry
> temperature = 2.12234919 K
> Ekin + Etot (const) = -29.21994544 Ry
> Linear momentum : 0.0000000000 0.0000000000
> NEW-OLD atomic charge density approx. for the potential
> If I use NFS directory - no any problems. So, I wonder is it my
> problem (permissions,etc) or other?
> Pw_forum mailing list
> Pw_forum at pwscf.org
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
More information about the users