[Wien] scratch folder in k-point parallel calculation

[Oleg Rubel]

If you are talking about a mixed (MPI & k-parallel) job, there are 
suggestions in FAQs: http://www.wien2k.at/reg_user/faq/pbs.html

I use SGE scheduler and the following parallel_options file:

setenv USE_REMOTE 0
setenv MPI_REMOTE 0
setenv WIEN_GRANULARITY 1
setenv WIEN_MPIRUN "mpiexec -machinefile _HOSTS_ -n _NP_ _EXEC_"

Eventually, it works with the local SCRATCH directory. I can share a 
submission script for the mixed MPI+k parallel job, if needed.

Oleg


On 13-06-27 4:46 PM, Yundi Quan wrote:
> I forgot to add that after lapw2, all the output files should be written
> to a shared folder (not to scratch folder of each node) so that sumpara
> -p could access all the output files. In the version that I'm using
> (11.0), it seems that lapw1 is not executed locally.
>
>
> Yundi
>
> On Thu, Jun 27, 2013 at 1:37 PM, Yundi Quan <yquan at ucdavis.edu
> <mailto:yquan at ucdavis.edu>> wrote:
>
>
>
>     ---------- Forwarded message ----------
>     From: *Stefaan Cottenier* <Stefaan.Cottenier at ugent.be
>     <mailto:Stefaan.Cottenier at ugent.be>>
>     Date: Thu, Jun 27, 2013 at 1:33 PM
>     Subject: Re: [Wien] scratch folder in k-point parallel calculation
>     To: A Mailing list for WIEN2k users <wien at zeus.theochem.tuwien.ac.at
>     <mailto:wien at zeus.theochem.tuwien.ac.at>>
>
>
>
>         I'm working on a cluster with many nodes. Each node has its own
>         /scratch
>         folder which cannot be accessed by other nodes. My own data
>         folder is
>         accessible to all nodes. When the WIEN2k scratch folder is set
>         to './',
>         everything works fine except that all the data write/read are
>         done in my
>         own data folder which slows down the system. When the WIEN2k scratch
>         folder is set to '/scratch', then the scratch files such as
>         case.vector_
>         created on one node would not be visible to the other. This
>         would not be
>         a problem for lapw1 -p and lapw2 -p if each node sticks to certain k
>         points and searches for scratch files related to these k-points. Is
>         there a way of configuring WIEN2k to work in this manner?
>
>
>     The k-point parallelization scheme works exactly in this way (mpi
>     parallelization is different).
>
>     STefaan
>
>
>     _________________________________________________
>     Wien mailing list
>     Wien at zeus.theochem.tuwien.ac.__at
>     <mailto:Wien at zeus.theochem.tuwien.ac.at>
>     http://zeus.theochem.tuwien.__ac.at/mailman/listinfo/wien
>     <http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien>
>     SEARCH the MAILING-LIST at:
>     http://www.mail-archive.com/__wien@zeus.theochem.tuwien.ac.__at/index.html
>     <http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html>
>
>
>
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:  http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
>