Dear Mariusz,
It seems to me from the error message that you have compiled mpi with
the shared memory device (ch_shmem ?). I seem to remember that the
default limits on the memory that could be shared between processors was
too low.
In any case, recompiling mpi without the use of shared memory for
interprocessor communication might solve your problem.
Best,
Xavier
Mariusz Krawiec wrote:
Dear Marcos,
Thanks for your suggestions. Changing optimization from O3 to O2 didn't
help but the problem, as you suggested, is probably in mpi or libraries.
At the momemnt I can run parallel siesta to the point where the memory
is checked. Then it crashes. Here is a part of out file:
InitMesh: Mesh cutoff (required, used) = 225.000 244.755 Ry
p0_5146: (60.036989) xx_shmalloc: returning NULL; requested 4976688 bytes
p0_5146: (60.036989) p4_shmalloc returning NULL; request = 4976688 bytes
You can increase the amount of memory by setting the environment variable
P4_GLOBMEMSIZE (in bytes); the current size is 4194304
p0_5146: p4_error: alloc_p4_msg failed: 0
It is quite wired.
All the best,
Mariusz.
----------------------------------------------------------------------
Dr Mariusz Krawiec
Institute of Physics Phone: +48 81 537 6146
M Curie-Sklodowska University Fax: +48 81 537 6191
pl. M. Curie-Sklodowskiej 1 E-mail: [EMAIL PROTECTED]
20-031 Lublin, Poland http://www.fnano.umcs.lublin.pl
---- ----
If you think of MS-DOS as mono, and Windows as stereo, then Linux is
Dolby Pro-Logic Surround Sound with Bass Boost and all the music is
free. - unknown
---- ----
On Wed, 30 May 2007, Marcos Verissimo Alves wrote:
Hi Mariusz,
I would use O2 optimization instead of O3. O3 does not always yield
improvements in performance, and can in some cases lead to numerical
errors. However, from your error messages, I would guess the error could
lie not in siesta, but in the mpi compilation - that is, if mpi is
compiled from scratch. Just a wild guess.
Cheers,
Marcos
Dear Siesta users,
Happy to join the list.
I have a problem with running Siesta in parallel mode on IA64 cluster.
The compilation was sucessful, albeit with some warnings:
fortcom: Warning: Optimization suppressed due to excessive resource
requirements; contact Intel Premier Support
in file siesta.F, line 11, column 0
At the end I got executable file.
For compilation (following arch.make file distributed with Siesta) I
used:
ifort -Vaxlib -c -O3 -tpp2 -W0 -DMPI -DNODAT -DWXML_INIT_FIX
-DHAS_DLAMC3 -DALLOC_SAVE_BUG
When I try to run it on 2 processors (mpirun -np 2) the program stops
and
I get a message:
rm_5015: p4_error: semget failed for setnum: 0
p0_16457: (0.200573) net_recv failed for fd = 5
p0_16457: p4_error: net_recv read, errno = : 104
When I run on one processor only (mpirun -np 1) everything is ok.
Note that siesta is compiled to run in parallel mode.
Thanks for any suggestions.
Mariusz.
----------------------------------------------------------------------
Dr Mariusz Krawiec
Institute of Physics Phone: +48 81 537 6146
M Curie-Sklodowska University Fax: +48 81 537 6191
pl. M. Curie-Sklodowskiej 1 E-mail: [EMAIL PROTECTED]
20-031 Lublin, Poland http://www.fnano.umcs.lublin.pl
---- ----
If you think of MS-DOS as mono, and Windows as stereo, then Linux is
Dolby Pro-Logic Surround Sound with Bass Boost and all the music is
free. - unknown
---- ----
--
Dr. Marcos Verissimo Alves
Post-Doctoral Fellow
Condensed Matter and Statistical Physics Sector
International Centre for Theoretical Physics
Trieste, Italy
--------
I have become so addicted to vi that I try to exit OpenOffice by
typing :wq!
