[Pw_forum] large system parallel execution problem

[German Samolyuk]

Dear Paolo,

Thank you.

About -npool 8:
I have problems with memory for this task. As I understand k-points
parallelisation does not distribute memory
but will take processors from r,k-parallelisation, these two
distribute memory, and effectively it will increase memory I'll need
to run the task.
Did I understand it wrong?

German




On Thu, Feb 9, 2012 at 10:32 AM, Paolo Giannozzi <giannozz at democritos.it> wrote:
> On Wed, 2012-02-08 at 17:58 -0500, German Samolyuk wrote:
>
>> The program runs electronic structure self consistency and calculate forces
>> but stops at first structure optimization step. Do you know possible reason.
>
> since "OOM killer terminated this process", you should discover who OOM
> killer is (hint: Out Of Memory?)
>
> For a better parallelization, you should run with -npool 8
>
> P.
> --
> Paolo Giannozzi, IOM-Democritos and University of Udine, Italy
>
>