Hi Nicolai and all,
did you compile MPB and Meep from sources?
I have compiled meep, harminv and mpb with a MPI support last February
and they worked fine on a 64bit system with 8 cores.
The system was running CentOS release 6.2, kernel
2.6.32-220.13.el6.x86_64. I used GCC version 4.4.6.
the main trick was to find a right compiler and compile HDF libraries
with --enable-parallel key. I have had a number of C and fortran
compilers installed on my computer, some of them did not work with MPI
at all.
hope it helps
pavel
On 25/06/2012 09:26, Walter, Nicolai wrote:
Hi everyone,
I have a weird problem using meep-mpi/meep on a 32-core,64gb RAM, 64bit system.
It seems that after I started meep-mpi e.g. with 8 processes, one or sometimes
two processes will leap behind in run time.
When meep should be finished, all processes go to sleep except those two. They
will run indefinitely and must be terminated by hand.
If I check the meep output, sometimes everything looks fine (last line: elapsed
run time=....) but sometimes it stops right after harminv!
This problem occurs just occasionally, sometimes everything works just as it
should.
I have experienced the same problem with meep, so i think it's not the MPI.
I already checked for hardware issues and found none,
i also ran the very same files on an 8-core 32bit system and they worked just
fine!
So maybe it's some problem with my harminv installation on 64bit?
If anyone has experienced similar issues or has any hints i would be very
thankful!
Best regards,
Nicolai
_______________________________________________
meep-discuss mailing list
meep-discuss@ab-initio.mit.edu
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
_______________________________________________
meep-discuss mailing list
meep-discuss@ab-initio.mit.edu
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss