[Pw_forum] postproc on linux cluster

[Paolo Giannozzi]

On Monday 11 August 2003 09:09, Goranka Bilalbegovic wrote:

> I am not able to use postprocessing routines on the linux cluster (mpich,
> PBS). I have learnt how to use PPs using mpi on SGI O2K, but on the linux
> cluster it does not work. The error is: from postproc : error #
> wrong no. of arguments

try to remove everything from PP/start_postproc.f90 but the following:

  version = 'POSTPROC-121' (or whatever you like)
  nodenumber = '   '
  call startup (nodenumber, version)
#ifdef __PARA
  call init_pool
#endif

With these changes, pp.x will properly read arguments, provided pw.x 
does. Note that this may work only on recent CVS version.

Originally the postprocessing code used to read data from a file,
whose file name was given as an argument, in parallel execution. 
This is no longer useful nor desirable, so all the crap contained 
in start_postproc.f90 can (should) (will!) be removed.

> On the page 12 of the manual there is a sentence: "The final result is
> collected into a single file, whose format is processor-number-independent,
> if the code is compiled with -D__NEW_PUNCH in CPPFLAGS." Postprocessing
> routines do not work on the linux cluster I use regardless of this
> -D_NEW_PUNCH option. I noticed that with this option the file *.pun does
> not form, but the number of *.wfc* files is again equal to the number of
> processors. I was expecting only one *.wfc to form.

I am afraid that the -D__NEW_PUNCH option and postprocessing
are not yet compatible. Remember that without -D__NEW_PUNCH
you need to run the postprocessing code on the same number of
processors and of pools that were used to generate data files.

Paolo

-- 
Paolo Giannozzi             e-mail:  giannozz at nest.sns.it
Scuola Normale Superiore    Phone:   +39/050509412
Piazza dei Cavalieri 7      Fax:     +39/050509417, 050563513     
I-56126 Pisa, Italy         Office:  Lab. NEST, Via della Faggiola 19