[Pw_forum] a strange behavior for QE-v4 compared with QE-v3

Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu
Wed Nov 26 09:29:42 CET 2008


this looks a lot like you are using two different versions
of MKL for the two compiles and thus are another "victim"
of the automatic threading of MKL v10. you should not
have a process with >100% CPU with a regular compile.
now if you have 8 cores and 8 MPI tasks and each of them
threads across 8 cores, you have a) a severe overload of
the scheduler and b) a big mess and all kinds of bad
performance issues.

try setting OMP_NUM_THREADS=1 and check if that
changes the behavior.

cheers,
   axel.



On 11/25/08, lan haiping <lanhaiping at gmail.com> wrote:
> Dear Paolo,
> thanks for explaination.
> But this behavior  is really strange i have ever come to.
> i have tried to rebuild the mpich and the QE, and nothing is
> gained.   Another strange problem is  that the work i execute  parallelly
> with QE-v4  sometimes come to idle without execuation or output
> information.
>
> is there any other suggestion to figure what the problem with it ?
>
> thanks
>
> On Wed, Nov 26, 2008 at 1:55 AM, Paolo Giannozzi
> <giannozz at democritos.it>wrote:
>
>> lan haiping wrote:
>>
>> > A strange  behavior is found for QE v4 when i carry out parallel
>> > executation [...] But this situation doesnot happen when i try
>> > to run the same work with QE-v3.
>>
>> it's hard to say anything meaningful based on the output of "top"
>> and little more. There are several parallelization levles in QE,
>> as explained here:
>> http://www.quantum-espresso.org/wiki/index.php/Running_on_parallel_machines
>> I don't remember what happened between v.3 and v.4, but whatever
>> happened, it is unlikely to make parallelism less effective.
>>
>> Paolo
>> --
>> Paolo Giannozzi, Democritos and University of Udine, Italy
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://www.democritos.it/mailman/listinfo/pw_forum
>>
>
>
>
> --
> Hai-Ping Lan
> Department of Electronics ,
> Peking University , Bejing, 100871
> lanhaiping at gmail.com, hplan at pku.edu.cn
>


-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
  Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.


More information about the Pw_forum mailing list