[Clas12_software] CLARA scaling test - COAT 5a.1.3

[Sebastián Mancilla]

I use the distributed YAML file:
https://github.com/JeffersonLab/clas12-offline-software/blob/master/etc/services/reconstruction.yaml,
that is copied by the installer as the default services.yml file.

I don't set any configuration parameters (I have no idea what to set
anyway).

If you give me a configuration, I can run the test on both versions again.



2018-03-15 12:28 GMT-03:00 Francois-Xavier Girod <fxgirod at jlab.org>:

> Dear Sebastian
>
> Changes in performances could also come in principle from the CCDB
> Although I do not think this has affected your test, for future
> comparisons it would be good to be able to re-use the same calibration
> constants
> Do you set the CCDB version in your yaml configuration?
>
> Thanks
>
> Best regards
> FX
>
> On Wed, Mar 14, 2018 at 10:00 PM, Sebastián Mancilla <smancill at jlab.org>
> wrote:
>
>> Hi all,
>>
>> The results of the multicore scaling test for CLARA with COAT 5a.1.3
>> services can be found here:
>>
>> https://claraweb.jlab.org/results/scaling/coat/clara-scaling
>> -coat-5a.1.3-20180314.html
>>
>> And the results of 5a.1.2 for comparison here:
>>
>> https://claraweb.jlab.org/results/scaling/coat/clara-scaling
>> -coat-5a.1.2-20180308.html
>>
>>
>> The event rate has increased 3x in 5a.1.3, with the same file, same node.
>>
>> The other important change is an increased execution time of the CVT
>> service when more cores are used.
>>
>>
>> Regards
>>
>> --
>> Sebastian Mancilla Matta
>> CCTVal, UTFSM
>> Valparaíso, Chile
>>
>> _______________________________________________
>> Clas12_software mailing list
>> Clas12_software at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/clas12_software
>>
>
>


-- 
Sebastian Mancilla Matta
CCTVal, UTFSM
Valparaíso, Chile
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/clas12_software/attachments/20180315/11505c87/attachment-0002.html>