On 7 March 2014 14:35, Christopher J. Walker <[log in to unmask]> wrote:
> On 06/03/14 15:57, Christopher J. Walker wrote:
>
>>
>> Multicore
>> =========
>>
>> I'm not sure whether it's worth summarizing status on this - but if it's
>> easy/thought to be worth it, I will.
>>
>
> As the info is on the ATLAS twiki, I may do a slide on that too. Any
> corrections, please get the twiki changed and let me know. CMS sites
> should also let me know.
>
> https://twiki.cern.ch/twiki/bin/view/AtlasComputing/AtlasMulticore
>
> says:
> RAL
> HTCondor Production and Analysis Multi-Core panda queues dynamic
> yes any corecount, default 8
>
> UKI-LT2-QMUL
> SGE one queue, CPUNumber=8 dynamic Yes
>
> UKI-NORTHGRID-LANCS-HEP
> Torque+Maui one queue, CPUNumber=8 static Yes 64 8-core nodes plus
> (maybe) fraction of larger SGE cluster
>
> UKI-NORTHGRID-MAN-HEP
> Torque+Maui one queue, CPUNumber=8 In test
>
> UKI-SCOTGRID-ECDF
> SGE one queue, CPUNumer=8 dynamic yes but agis needs reconfiguring
>
> UKI-SCOTGRID-GLASGOW
> Torque+Maui one queue, CPUNumber=8 dynamic configured but currently
> disabled disabled since SL6 move; on the workplan for the new year to
> recommission.
More detail on the Glasgow situation: we were actually going to rename
the queue to something else (it's currently called something like
wholenode, and was originally used for whole-node scheduling tests for
Andy Washbrook), and I just need to get around to doing that. (We do
have a local queue, not advertised in the BDII, for parallel jobs,
that currently works for MPI jobs happily, so there's no real problem
with the overall batch config.)
Sam
>
> Chris
|