[molpro-user] Running molpro2008.1 in parallel doesn't work

[zhendong zhao]

Hi Thierry,

A quick question. Can you run molpro on a single slave node directly?
If Yes, can you run molpro on a single slave node through master node? 
If the second answer is No, you nodes may have communication problems. 

ZZ

On Wed, 18 Mar 2009 15:30:38 +0100 Thierry Leininger
<Thierry.Leininger at irsamc.ups-tlse.fr> wrote:

> Dear all,
> 
> I just installed the Molpro 2008.1 binary version 
> 
> Version 2008.1 for architecture Linux/em64t, standard code, mpp
> (Patchlevel 5)
> 
> on two different Linux clusters but I can't have it run on several
> nodes (either with the -N option or using pbs qsub command).
> 
> I can run it parallel on a single node even with several cpus.
> When I launch it from the master, tasks are actually sent to the
> nodes, but then they just sit there and do not progress (cpu time
> remains 0:00).
> 
> Version 2006 was running without any problem.
> 
> Below is what I get after I submit the job using 
> 
> 	molpro -N leininge:node10:1,leininge:node11:1 H2O.in
> 
> or qsub.
> 
> I really can't figure out what is going on.
> I tried various things around the 
>  workaround: parallel M2008 stalls on startup under SuSE   submitted
> by Gershom (Jan)Martin to the mailing October 2008 but nothing helped.
> 
> Thank you in advance.
> 
> 	Thierry
> 
> ==================================================
> tmp 
> = /home/leininge/pdir//home/leininge/bin/molprop_2008_1_Linux_x86_64_i8.exe.p
>  Creating: host=node10, user=leininge,
>            file=/home/leininge/bin/molprop_2008_1_Linux_x86_64_i8.exe, 
> port=50165
> /home/leininge/bin/molprop_2008_1_Linux_x86_64_i8.exe, len=53
>    -master, len=7
> s10-theorie.lcar.ups-tlse.fr, len=28
>      50165, len=5
>          2, len=1
>          2, len=1
>          0, len=1
>          0, len=1
>  Creating: host=node11, user=leininge,
>            file=/home/leininge/bin/molprop_2008_1_Linux_x86_64_i8.exe, 
> port=50166
> /home/leininge/bin/molprop_2008_1_Linux_x86_64_i8.exe, len=53
>    -master, len=7
> s10-theorie.lcar.ups-tlse.fr, len=28
>      50166, len=5
>          2, len=1
>          2, len=1
>          1, len=1
>          1, len=1
> ARMCI configured for 2 cluster nodes. Network protocol is 'TCP/IP
> Sockets'.
> 
>  Primary working directories    : /tmp/leininge
>  Secondary working directories  : /tmp/leininge
>  Wavefunction directory         : /home/leininge/wfu/
>  Main file repository           : /tmp/leininge/
> 
>  cpu       : P4 2327.549 MHz
>  FC        : /opt/intel/fce/10.1.008/bin/ifort
>  FCVERSION : 10.1
>  
> BLASLIB   : -L/opt/intel/mkl/9.1/lib/em64t -lmkl_em64t -lguide
> -lpthread -openmp id        : leininge
> 
>  MPP nodes         nproc
>  node10.alineos.net    1
>  node11.alineos.net    1
>  ga_uses_ma=false, calling ma_init with nominal heap.
>  GA-space will be limited to   8.0 MW (determined by -G option)
> 
>  MPP tuning parameters: Latency=     0 Microseconds,   Broadcast
> speed=    0 MB/sec
>  default implementation of scratch files=ga
> 
> 
>