[DFTB-Plus-User] dftb+.mpi as spawned process

[Alex A. Schmidt]

Dear Jacek,

Yes, sure! That is an important point, no doubt. My code deals with that by
running one serial dftb+ for each rank requested to do that, each one
executing
from its own temporary subdirectory like /tmp/rank_001, /tmp/rank_002,...
with the proper dftb input files. dftb+.mpi is to be called (by
mpi_comm_spawn)
in the same way...

Alex


2014-12-29 19:02 GMT-02:00 Jacek Jakowski <jjakowski at gmail.com>:

>
> Alex,
>
>  ....I think,  you want to be careful about your  temporary IO files,
> especially when your spawned dftb+ processes  reside in the IO directory.
> This may be a problem when your  different  dftb+  instances read  and
> write to/from the same files.  You can avoid it by  letting each  process
> to run in its own "private" subdirectory.
>
> jacek
>
> On Mon, Dec 29, 2014 at 3:10 PM, Alex A. Schmidt <aas at ufsm.br> wrote:
>
>> Dear Bálint,
>>
>> Thanks for your reply.
>>
>> 2014-12-29 13:09 GMT-02:00 Bálint Aradi <aradi at uni-bremen.de>:
>>
>>>
>>> I've added that to the code in our MPI-branch.
>>> >
>>> >     if (mympi%master.and.mpi_comm_parent.eq.mpi_comm_null) then
>>> >       stdout = output_unit
>>> >     else
>>> >       stdout = 1
>>> >       open(stdout, file="/dev/null", action="write")
>>> >     end if
>>>
>>> This solution, I do no like as a general solution, as for me it is not
>>> trivial at all, that all IO to stdout should be suppressed when DFTB+ is
>>> spawned by other process via mpi_comm_spawn(). One could eventually
>>> think about a flag in the input file (or a command line argument for
>>> DFTB+), which would do that.
>>>
>>
>> I certainly agree with you regarding the above solution to suppress
>> stdout.
>> Its advantage is to be quick and easy to implement and would do the trick
>> in my case...
>>
>> I think a command line argument like "-o outfile" is indeed a much better
>> choice to manage stdout as options like this can be easily passed to spawn
>> processes. It is also aesthetically similar to the regular i/o
>> redirection "> outfile"
>> option and you would not have to alter the current input standard of
>> dftb+, even
>> though I have a feeling that adding a flag to the input file would be
>> much easier
>> to implement.
>>
>> Best regards and a great 2015,
>>
>> Alex
>>
>> 2014-12-29 13:09 GMT-02:00 Bálint Aradi <aradi at uni-bremen.de>:
>>
>>> Dear Alex,
>>>
>>> thank you for your suggestions.
>>>
>>> > As far as I know, all that is necessary for dftb+.mpi to become "spawn
>>> > ready" would be to
>>> > add something like
>>> >
>>> >    integer :: mpi_comm_parent
>>> >
>>> >    call mpi_comm_get_parent(mpi_comm_parent,error0)
>>> >
>>> >    if (mpi_comm_parent.ne.mpi_comm_null) then
>>> >        call mpi_comm_disconnect(mpi_comm_parent,error0)
>>> >    endif
>>> >
>>> > right before "call mpi_finalize(error0)". Please, note that testing the
>>> > value of mpi_comm_parent
>>> > as above make the call to mpi_comm_disconnect safe in case dftb+.mpi is
>>> > called from "mpirun".
>>>
>>> I've added that to the code in our MPI-branch.
>>> >
>>> >     if (mympi%master.and.mpi_comm_parent.eq.mpi_comm_null) then
>>> >       stdout = output_unit
>>> >     else
>>> >       stdout = 1
>>> >       open(stdout, file="/dev/null", action="write")
>>> >     end if
>>>
>>> This solution, I do no like as a general solution, as for me it is not
>>> trivial at all, that all IO to stdout should be suppressed when DFTB+ is
>>> spawned by other process via mpi_comm_spawn(). One could eventually
>>> think about a flag in the input file (or a command line argument for
>>> DFTB+), which would do that.
>>>
>>>   Best regards and a happy new year,
>>>
>>>   Bálint
>>>
>>> --
>>> Dr. Bálint Aradi
>>> Bremen Center for Computational Materials Science, University of Bremen
>>> http://www.bccms.uni-bremen.de/cms/people/b_aradi/
>>>
>>>
>>>
>>> _______________________________________________
>>> DFTB-Plus-User mailing list
>>> DFTB-Plus-User at mailman.zfn.uni-bremen.de
>>> https://mailman.zfn.uni-bremen.de/cgi-bin/mailman/listinfo/dftb-plus-user
>>>
>>>
>>
>> _______________________________________________
>> DFTB-Plus-User mailing list
>> DFTB-Plus-User at mailman.zfn.uni-bremen.de
>> https://mailman.zfn.uni-bremen.de/cgi-bin/mailman/listinfo/dftb-plus-user
>>
>>
>
> _______________________________________________
> DFTB-Plus-User mailing list
> DFTB-Plus-User at mailman.zfn.uni-bremen.de
> https://mailman.zfn.uni-bremen.de/cgi-bin/mailman/listinfo/dftb-plus-user
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20141229/f5491d15/attachment.html>