The performance of jobs at CERN compared to those at Wigner are actively being investigated and Helge Meinhard mentioned this at HEPIX last week.
He said a group will be doing careful tests and that job efficiencies may appear different for many reasons, and warned against simple conclusions some have appeared to find.
I don't know the details and we should await the CERN groups findings. One example was; a jobs efficiency (measured by CPU time over Wall time) may appear better on an older CPU at CERN compared to one at Wigner as the CPU at Wigner is newer and faster and hence may have more IO wait. (Assuming the IO systems is equal).
I'm sure there are loads of variables though and that’s why they are investigating.
Cheers Pete
--
----------------------------------------------------------------------
Peter Gronbech GridPP Project Manager Tel No. : 01865 273389
Department of Particle Physics,
University of Oxford,
Keble Road, Oxford OX1 3RH, UK E-mail : [log in to unmask]
----------------------------------------------------------------------
> -----Original Message-----
> From: Testbed Support for GridPP member institutes [mailto:TB-
> [log in to unmask]] On Behalf Of David Colling
> Sent: 24 October 2014 14:24
> To: [log in to unmask]
> Subject: Apologies and a question ...
>
> First my apologies for missing the technical meeting. It turns out that
> Starbucks network and Vidyo are not compatible...
>
> The second thing is a question that I would have asked if they had been.
> What are Atlas and LHCbs experiences of using Wigner? We are doing lots
> of testing in CMS with very mixed results.
>
>
>
> Best,
> david
|