Please read the UG more carefully.
I think parallelization is explained pretty well.
In particular you have to understand the difference between k-parallel
and fine-grane mpi-parallel mode.
The syntax for lapw1/2 parallelization is:
speed:hostname[:cores for mpi] [other_hostnames_for_mpi]
So the first number is a relative speed number. Useful if you want to
combine a fast and a slow computer in k-parallel mode. For you: always 1
second ist hostname, in your case it can be localhost
:8 would mean 8 fold mpi-parallel. It requires that you have installed
properly the mpi-version (this is a bit more difficult) AND that you
have a bigger case which runs reasonable in mpi-mode. If you need 74
k-points, most likely this is NOT the case.
I'd set OMP_NUM_THREAD=2 and then use:
granularity:1
1:localhost
1:localhost
1:localhost
1:localhost
residue: localhost
# only if you have mpi installed !!
lapw0: localhost:8
dstart: localhost:8
This will start 4 jobs with 18k and a residue job with 2 k.
Each will use 2 cores because of OMP_NUM_THREADS when wien2k is
installed properly.
On 03/16/2014 03:12 PM, Juliet parker wrote:
*Hi dear WIEN2k users*
I want to run wien2k in the parallel mode (i.e. run_lapw -p -cc 0.0001)
on only one computer which has 8 CPUs, /i.e./ all 8 CPUsbe used.
I have read user-guide (section 5.5.3), but it was not explained
“Parallelization” on only one computer.
The default setting of “.machines” file for my system is as follows:
1:localhost
1:localhost
granularity:1
extrafine:1
As I have understood from user-guide, I should edit this file as follows
(I have 74 k-points):
-
granularity:1
9:localhost
9:localhost
9:localhost
9:localhost
9:localhost
9:localhost
9:localhost
9:localhost
residue: localhost:2
lapw0: localhost:8
dstart: localhost:8
# 9kpoints * 8cups =72 kpoints from 74 kpoints, so 2 additional k-points
were assigned to “residue:localhost:2”
OR as:
granularity:1
74:localhost:8
lapw0: localhost:8
dstart: localhost:8
Are such “.machine” files true?
Could you please help me with this problem?
Thank you
J.Parker
___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
--
P.Blaha
--
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-165300 FAX: +43-1-58801-165982
Email: bl...@theochem.tuwien.ac.atWWW:
http://info.tuwien.ac.at/theochem/
--
___
Wien mailing list
Wien@zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
SEARCH the MAILING-LIST at:
http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html