[Wien] All cpus are not running for a single calculation
Gavin Abo
gsabo at crimson.ua.edu
Thu Dec 12 07:49:34 CET 2019
a) Don't confuse a thread and a CPU (e.g. with multiple processor
cores):
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg18964.html
b) I suggest you search the mailing list archive, there is probably many
good examples that have been posted in the past. For example, one of
many that came up in a search:
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg18967.html
c) I see "export OMP_NUM_THREADS=10" in your .bashrc. I don't see a
mention of what CPU you are using. Each CPU would need to support more
than 1 thread. If each CPU only supports 1 thread, you should not use
more than OMP_NUM_THREADS=8 for the 8 CPU that you have. I guess you
are not using a .machines file as you did not provide one. If you are
using a .machines file, the OMP_NUM_THREADS in your .bashrc might not be
important. As I recall, if you have omp_global in your .machine it will
override what is in .bashrc unless you have removed that line (e.g., if
the omp_global line is placed there by default by w2web [1] from the
template).
[1]
https://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg10767.html
On 12/11/2019 11:04 PM, shamik chakrabarti wrote:
> Dear Sir,
>
> I am sending the .bashrc file herewith this mail. for
> your consideration. I have used the same configuration for a
> workstation having 8 cpus & all the cpus are running fine there.
>
> Looking forward to hearing from you.
>
> with regards,
>
> On Wed, 11 Dec 2019 at 20:37, Peter Blaha
> <pblaha at theochem.tuwien.ac.at <mailto:pblaha at theochem.tuwien.ac.at>>
> wrote:
>
> How have you set OMP_NUM_THREAD and how did you run the calculations.
>
> In WIEN2k_19 the proper way is to set omp_global:2 (or 4,..) in
> .machines
>
> In addition: How did you compile wien2k ? Our default options in
> siteconfig set "openmp" and you get multi-threaded executables. If
> you
> use your own switches, you may have created a single-thread
> program and
> will never see more than 100%.
>
>
>
> On 12/11/19 1:46 PM, shamik chakrabarti wrote:
> > Dear Sir,
> >
> > We are running simulation in a work station having 16
> cpu. We
> > have tried to run all the cpu by setting OMP_NUM_THREAD = 2,4,8,16.
> > However, it is seen that only one cpu is running ( showing 100 %
> while
> > typing top command in the terminal) for a single calculation.
> If we run
> > 4 calculations together each task is running in a single cpu
> (total 4
> > cpu, showing 100% 100% 100% 100%) . If we want to run all the
> cpu for a
> > single calculation (showing 1600% for a single calculation) what
> should
> > we have to do?
> >
> > Any response in this regard will be helpful for us. Thanks in
> advance.
> >
> > with regards,
> >
> > --
> > Dr. Shamik Chakrabarti
> >
> >
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> <mailto:Wien at 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: blaha at theochem.tuwien.ac.at
> <mailto:blaha at theochem.tuwien.ac.at> WIEN2k: http://www.wien2k.at
> WWW: http://www.imc.tuwien.ac.at/TC_Blaha
> --------------------------------------------------------------------------
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> <mailto:Wien at 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
>
>
>
> --
> Dr. Shamik Chakrabarti
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20191211/3f1b8aba/attachment.html>
More information about the Wien
mailing list