[FLASH-USERS] HELP: configure FLASH to run on large scale cluster

[Scott Feister]

Hi Yubo,

Also, your output log file would be helpful -- towards the end, it gives a
breakdown of where time is being spent.

Best,

Scott


Scott Feister, Ph.D.
Postdoctoral Researcher, Flash Center for Computational Science
University of Chicago, Department of Astronomy and Astrophysics

On Tue, Oct 4, 2016 at 9:45 AM, Scott Feister <sfeister at gmail.com> wrote:

> Hi Yubo,
>
> Can you also send your script for your job submission? E.g. a PBS script,
> something you type into the terminal, etc.
>
> Scott
>
>
> Scott Feister, Ph.D.
> Postdoctoral Researcher, Flash Center for Computational Science
> University of Chicago, Department of Astronomy and Astrophysics
>
> On Tue, Oct 4, 2016 at 9:41 AM, Yubo Qin <qybo123 at gmail.com> wrote:
>
>> Hi All,
>>
>> I am trying to run FLASH code on relative large scale, i.e., 512 cores.
>> But did not find an appropriate input configuration.
>> I pick up a Cellular simulation problem from the built-in FLASH test suit
>> (~PATHTOFLASH/flash/source/Simulation/SimulationMain/Cellular)and was
>> running it from 4 cores to 256 cores, as shown in attached figure, and
>> found out that this test input configuration might suitable for running on
>> around 64 cores, however, I want to run it on more around 512 cores.
>> Thought I know change AMR refinement level may help, but its execution time
>> will goes to extremely high. Also, I guess merely change AMR refinement
>> level is not enough.
>>
>> Any idea of how to appropriate configure this input parameter to make
>> this simulation problem suitable for running on large scale cluster?
>>
>> I attached my current input configuration file (flash.par) as well.
>>
>>
>> Many thanks in advance!
>>
>> Regards,
>> Yubo
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20161004/f68f3303/attachment.htm>