[Wien] parallel calc. without NFS

[Kevin Jorissen]

Only vectorfiles and helpfiles are written to the scratch-directories.  The
other files, like case.energy_? etc., are written to the case directory,
where your struct-file etc. is.
The help-files are not needed for anything.  But you will need the
vector-files for some things :
qtl, for example.  If you want to use lapw2 -qtl, eg., you will have to copy
the vector files back to the case directory (I added a few lines in
lapw2para that use the .processes-file to do exactly that).
Also, lapw2 itself needs the vector-files in each iteration, which is what
Stefaan and I discussed earlier in this conversation - lapw2_1 must be
executed on the same machine as lapw1_1 etc. (well, technically you could
start copying vector-files there, too, but I reckon that would get a little
complicated).

Kevin.


----- Original Message -----
From: "Steven Homolya" <Steven.Homolya at spme.monash.edu.au>
To: <wien at zeus.theochem.tuwien.ac.at>
Sent: Tuesday, July 15, 2003 5:02 AM
Subject: Re: [Wien] parallel calc. without NFS


> On Mon, 14 Jul 2003, Peter Blaha wrote:
>
> > >
> > > That is what I'm doing at present. The problem is that $PBS_JOBFS
(jobspace
> > > path) is different on every node, so I can't use more than one node (4
cpus)
> > > for one calculation.
> >
> > Yes you can. All the "sed",... is done only on the "first" node, and all
> > files are completely local to this node.
> >
>
> What about files named *_1, *_2, *_3 etc. in the case directory? I assumed
> that they were generated by processes 1,2,3 etc. Is this correct? If so
> *_[1-4] will only exist on node 1, *_[5-8] only on node 2, etc...
>
> --
> Steven Homolya
> School of Physics and Materials Engineering
> Monash University, VIC 3800
> Australia
> Tel: +61 3 9905 3694
> Fax: +61 3 9905 3637
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>