[Wien] Question about parallel mode calculation
Jorissen Kevin
Kevin.Jorissen at ua.ac.be
Thu Apr 15 16:54:49 CEST 2004
Hi, I cant solve your problems right away. Just a few thoughts :
* k-point parall and MPI (fine grain) parall are two completely different things. you can use k-point p. without MPI !! Since MPI usually takes more effort than k-point to get things working, I advise you to focus on k-point p first.
* if you're working on a large machine that you yourself don't control, you should probably be in touch with your system administrator for stuff like MPI, passwordless access to machines, ...
* lapw1para says it's going to do just one parallel job. That means : no k-point parall !!
Probably, your .machines-file is wrong. Does it contain only one line? It should contain more!!
good luck,
Kevin.
-----Original Message-----
From: Dong YuHui [mailto:dongyh at mail.ihep.ac.cn]
Sent: Thu 4/15/2004 4:22 AM
To: wien at zeus.theochem.tuwien.ac.at
Cc:
Subject: [Wien] Question about parallel mode calculation
Dear Wien users,
I want to launch the parallel calculation in a PC farm. However I can not
get enough information about how to do it.
I run the k-point parallelization and fail. The dayfile reported:
Calculating ZnO_para in /home/wien2k/w2work/ZnO_para
on bsrf-serv.ihep.ac.cn
start (Thu Apr 15 08:57:31 CST 2004) with lapw0 (20/20 to go)
> lapw0 -p (08:57:31) starting parallel lapw0 at Thu Apr 15 08:57:31
CST 2004
--------
running lapw0 in single mode
3.760u 0.110s 0:03.86 100.2% 0+0k 0+0io 2052pf+0w
> lapw1 -c -p (08:57:35) starting parallel lapw1 at Thu Apr 15
08:57:35 CST 2004
-> starting parallel LAPW1 jobs at Thu Apr 15 08:57:35 CST 2004
running LAPW1 in parallel mode (using .machines)
1 number_of_parallel_jobs
** LAPW1 crashed!
0.100u 0.140s 0:03.20 7.5% 0+0k 0+0io 10315pf+0w
> stop error
STDOUT reported:
FORTRAN STOP LAPW0 END
cat: No match.
testpara past, testpara1 failed.
I also checked the file lapw1para_lapw, and found
if ( $?WIEN_MPIRUN ) then
set mpirun = "$WIEN_MPIRUN"
else
set mpirun='mpirun -np _NP_ _EXEC_'
endif
I do not know how to define WIEN_MPIRUN. It seems essential for
k-point parallelization. I also install mpich-1.2.5 and set
setenv WIEN_MPIRUN /home/wien2k/mpi/mpich-1.2.5/bin/mpirun
before I set WIEN_MPIRUN, the program reported "/root/bin/mpirun:
Permission denied", now no this message but testpara1 still fails.
Anybody have any suggestions or ideas about how to do? Many thanks.
Yu-Hui Dong
_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 6286 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20040415/d574b86b/attachment.bin
More information about the Wien
mailing list