Yes, the non-mpi version works fine.
The weird thing is that if I run the test that fails using LAM mpi
(7.1.1), but with only 1 process, it works. But as soon as I have more
than 1 process, it fails. This is illustrated below:
1. USING LAM MPI, 1 PROCESS:
---------------------------------
$ mpirun -np 1 ./flux.dac
Using MPI version 1.2, 1 processes
Trying out the fluxes...
Passed Split flux plane split by 7...
Energy starts at 0.0166242
Delta E: 0.00198055
Integrated Flux: 0.00162266
Ratio: 1.22056
Fractional error: 0.0215284
Passed Cavity 1D 6.01 73
Energy starts at 0.0166154
Delta E: -4.49277e-05
Integrated Flux: 2.01486e-05
Ratio: -2.22982
Fractional error: -0.00391662
Passed Cavity 1D 5.0 1
Energy starts at 0.0165876
Delta E: 0.000800872
Integrated Flux: 0.000865302
Ratio: 0.92554
Fractional error: -0.00388424
Passed Cavity 1D 3.85 55
...
1. USING LAM MPI, 2 PROCESS:
---------------------------------
$ mpirun -np 2 ./flux.dac
Using MPI version 1.2, 2 processes
Trying out the fluxes...
Passed Split flux plane split by 7...
Energy starts at 0.0166242
Delta E: 0.00198055
Flux differs by 0.00126477 out of 0.0133789
This gives a fractional error of 0.0945351
meep: Failed Cavity 1D 6.01 73!
Integrated Flux:
-----------------------------------------------------------------------------
One of the processes started by mpirun has exited with a nonzero exit
code. This typically indicates that the process finished in error.
If your process did not finish in error, be sure to include a "return
0" or "exit(0)" in your C code before exiting the application.
PID 23114 failed on node n0 (10.2.0.3) with exit status 1.
-----------------------------------------------------------------------------
Do I need to give any special command line switches to the mpirun
command? From the output above, it does not 'look' like my error is
compilation/linking related. It looks like a runtime error where the
error threshold is exceeded; but I don't know anything about Meep, so
please correct me if I'm wrong here.
Any ideas?
Regards,
Andre
Steven G. Johnson wrote:
On Mon, 28 Aug 2006, Andre Charbonneau wrote:
I managed to compile and install the latest version of Meep (0.10) on
my cluster. But when I try to run 'make check', I get the following
error after a while. (I tried with the 'enable-debuggin' switch but
still get the same results):
Does the non-MPI version of Meep work? If so, it may be a problem with
your MPI implementation. (I tend to recommend LAM or OpenMPI, and have
had bad experiences with MPICH in the past.)
Steven
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
--
Andre Charbonneau
Grid Computing and Applications Specialist
Research Computing Support, IMSB
National Research Council Canada
Ottawa, ON, Canada K1A 0R6
-- Linux uptime: 207 days 0:28 --
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
