[Nek5000-users] Reading rea file on IBM Blue Gene/Q server

[nek5000-users at lists.mcs.anl.gov]

Thank you very much. Using .re2 instead of .rea brought down
the time from 27 sec to 0.8 sec.


On Tuesday 16 May 2017 12:56 PM, nek5000-users at lists.mcs.anl.gov wrote:
> For performance use .re2 and the latest master available on GitHub.
>
> On 16 May 2017, at 11:57, "nek5000-users at lists.mcs.anl.gov 
> <mailto:nek5000-users at lists.mcs.anl.gov>" 
> <nek5000-users at lists.mcs.anl.gov 
> <mailto:nek5000-users at lists.mcs.anl.gov>> wrote:
>
>> Hi Neks,
>>
>>   We are carrying out an optimization study on IBM Blue Gene/Q server
>> of IDRIS using direct-adjoint method. The direct and adjoint equations
>> are solved using Nek5000. We have observed that for moderate to large
>> number of
>> processors  Nek5000 spends considerable time in reading the .rea file.
>> When we reduce the number of processors, the .rea file is read in lesser
>> time but
>> the time taken for solver increases. To give an idea, a 3D case for
>> incompressible flow
>> with 12000 elements spends 22.132 seconds in reading .rea file if we use
>> 400 processors. If we increase the number of processors to 1000, it
>> takes more than
>> 90 seconds to read the same .rea file. Can someone provide us a 
>> solution on
>> how to make Nek read the rea file quickly? Would using .re2 file instead
>> of .rea increase
>> the performance considering the number of elements is not very large.?
>> I would like to mention that we are using PPLIST="MPIIO" in the makenek
>> file.
>>
>> Many thanks.
>>
>>
>>
>> _______________________________________________
>> Nek5000-users mailing list
>> Nek5000-users at lists.mcs.anl.gov <mailto:Nek5000-users at lists.mcs.anl.gov>
>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>
>
>
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20170517/e55c5bb3/attachment.html>