as_run crashes on plate example when re-ran

Hi,

Moved on to another one of Koen's fine tutorials.
This ran perfect on Salome 5.13 & Code_Aster 10.1 ( 64 bits ubuntu ).
The strange thing though is that when I try to re-run the example, but with 4 procs instead of 1, it fails with the following error ( see .export below... ).
Runs perfectly though with a single processor. Still, I'm a little surprised that it uses 900mb to compute. Or is aster trying to use as much memory to speed things up?

Thanks!

-jelle


www.caelinux.org/wiki/index.php/Contrib:...eometry_of_the_plate:


-- CODE_ASTER -- VERSION DE DEVELOPPEMENT STABILISEE 10.01 --

VERSION SEQUENTIELLE

COPYRIGHT EDF-R&D 1991 - 2009

EXECUTION DU : JE-11-FEVR-2010 04:44:36

PLATE-FORME : UBUNTU

NB MAX PROC OpenMP: 1

SYSTEME : LINUX

CPU : X86_64


!

---

!
! <EXCEPTION> <FACTOR_21> !
! !
! !
! Matrice non factorisable ! !
! Si le solveur linéaire est LDLT ou MULT_FRONT on sait en plus que: !
! - pivot est vraiment nul à la ligne 46570 !
! - pour le noeud N15297 et la composante DY !
! !
! !
! !
! !
!

---

!

<S> ERREUR UTILISATEUR RECUPEREE PAR LE SUPERVISEUR
<I> <FIN> FERMETURE DE LA BASE "GLOBALE" EFFECTUEE.
<I> <FIN> ARRET NORMAL DANS "FIN" PAR APPEL A "JEFINI".
<I> FERMETURE DES BASES EFFECTUEE


============== .export file ==============

P profastk jelleferinga@ubuntu.jelle.org:/home/jelleferinga/Desktop/salome_tutorials/staticshell/solid2t.astk
P serveur ubuntu
P noeud ubuntu
P username jelleferinga
P mclient ubuntu.jelle.org
P uclient jelleferinga
P version STA10.1
P lang en
P debug nodebug
P mode interactif
P ncpus 4
P mpi_nbcpu 1
P mpi_nbnoeud 1
P classe
P depart
P distrib
P flashdir
P exectool
P nomjob solid2t
P origine ASTK 1.8.0
P display ubuntu.jelle.org:0.0
A ORBInitRef NameService=corbaname::amdaltrax:2810
A args
A memjeveux 250.0
P mem_aster 100.0
A tpmax 30000
P memjob 2048000
P tpsjob 500
P nbmaxnook 5
P cpresok RESNOOK
P facmtps 1
P corefilesize unlimited
F mail /home/jelleferinga/Desktop/salome_tutorials/staticshell/quad2t.med D 20
F comm /home/jelleferinga/Desktop/salome_tutorials/staticshell/quad2tb.comm D 1
F mess /home/jelleferinga/Desktop/salome_tutorials/staticshell/quad2t.mess R 6
F resu /home/jelleferinga/Desktop/salome_tutorials/staticshell/quad2t.resu.med R 8
F erre /home/jelleferinga/Desktop/salome_tutorials/staticshell/quad2t.erre R 9
R base /home/jelleferinga/Desktop/salome_tutorials/staticshell/quad2t.base R 0
F mmed /home/jelleferinga/Desktop/salome_tutorials/staticshell/res3tq.med R 21
F mast /home/jelleferinga/Desktop/salome_tutorials/staticshell/table.txt R 26
P consbtc oui
P soumbtc oui
P actions make_etude<br /><br />Post edited by: jelle feringa, at: 2010/02/11 15:14

I see! Many thanks for the pointer.
If only apt-get code_aster_mpi ;'(

Scratch that… the thing that surprised me when running the analyses with 4 cores is that it _did_ run, while using 4 cpu's. So, that leaves me thinking that 10.1 does support multiple cpu's, right? Sorry, should have realized that before…

-jelle<br /><br />Post edited by: jelle feringa, at: 2010/02/13 16:41