[mcstas-users] mcstas 2.5 mpirun problem

[Peter Kjær Willendrup]

Hi Uli,


As a matter of fact not a lot changed in our mpi-implementation:

- There is only the mother-process to do things like pick up parameters or mkdir etc. So the most probable situation is that the binary was in fact not mpi-capable when launched through mpirun.

The easiest way to check is through ./instrument.out

./ILL_IN4.out --help
ILL_IN4 (ILL_IN4.instr) instrument simulation, generated with McStas 2.5 - Dec. 12, 2018 (Dec. 12, 2018)
Usage: ./ILL_IN4.out [options] [parm=value ...]
Options are:
  -s SEED   --seed=SEED      Set random seed (must be != 0)
  -n COUNT  --ncount=COUNT   Set number of neutrons to simulate.
  -d DIR    --dir=DIR        Put all data files in directory DIR.
  -t        --trace          Enable trace of neutrons through instrument.
  -g        --gravitation    Enable gravitation for all trajectories.
  --no-output-files          Do not write any data files.
  -h        --help           Show this help message.
  -i        --info           Detailed instrument information.
  --format=FORMAT            Output data files using FORMAT=MCSTAS

This instrument has been compiled with MPI support.

… etc.

The above line will not be present in case the instrument is not compiled with MPI.


The only thing that comes to mind in terms of changes is that our two mcrun variants differ in how one specifies to compile with mpi, in the sense that

Perl: mcrun.pl -c --mpi    is enough
Python: mcrun -c --mpi=some_integer  (i.e. --mpi needs a. value)
(For the full list of differences, see https://github.com/McStasMcXtrace/McCode/wiki/mcrun-variants---table-overview

From the (python) gui, the easiest is to use “MPI (recompile)” from the run dialogue.

Also remember that if you inadvertently make a change in the .instr code, then a save and exiting mcgui, the automatic recompile (since now the .instr code is newer than the .c code) may do a non-mpi compile for you.

Finally, our recommended “standard” MPI to use is OpenMPI, but MPICH should be just as good.

Hope this helps, otherwise write again - but expect some delay as I now started my summer vacation. ;-)


Cheers,

Peter



On 19 Jul 2019, at 17.18, Wildgruber, Christoph U. <wildgrubercu at ornl.gov<mailto:wildgrubercu at ornl.gov>> wrote:

Hi,

I am having a problen (mpi-) running mcstas 2.5 on centos 7.? with more than 1 node.
It looks like every node tries to create the target directory wether I start from  mcgui or
use the mcrun command in the terminal. The location of the output directory does not
seem to matter. I am using mpich on a newly installed OS
and don’t remember having had that problem before.
The 2.5 manual mentions new options but I did not find anything wrt this problem.
What am I missing?

Thanks,

Uli
_______________________________________________
mcstas-users mailing list
mcstas-users at mcstas.org<mailto:mcstas-users at mcstas.org>
https://mailman2.mcstas.org/mailman/listinfo/mcstas-users


Peter Kjær Willendrup
Forskningsingeniør, Specialkonsulent
Næstformand for DTU Fysik LSU

DTU Physics


[cid:b6339554-7f28-441d-9f31-5944f811397c at win.dtu.dk]


Technical University of Denmark


[cid:e108e480-fcab-46e2-9531-b38165079572 at win.dtu.dk]



Department of Physics
Fysikvej
Building 307
DK-2800 Kongens Lyngby
Direct +45 2125 4612
Mobil +45 2125 4612
Fax +45 4593 2399
pkwi at fysik.dtu.dk<mailto:pkwi at fysik.dtu.dk>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman2.mcstas.org/pipermail/mcstas-users/attachments/20190720/23cec8c2/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.gif
Type: image/gif
Size: 58 bytes
Desc: image001.gif
URL: <http://mailman2.mcstas.org/pipermail/mcstas-users/attachments/20190720/23cec8c2/attachment-0002.gif>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.gif
Type: image/gif
Size: 1055 bytes
Desc: image002.gif
URL: <http://mailman2.mcstas.org/pipermail/mcstas-users/attachments/20190720/23cec8c2/attachment-0003.gif>