Hi Antun,
On Mon, 24 Apr 2006, Mattias Wadenstein wrote:
> On Mon, 24 Apr 2006, Antun Balaz wrote:
>
>> Hi to all,
>>
>> I am trying to find out if there is any effort (official or unofficial) on
>> establishing a procedure for determining the value of GlueHostBenchmarkSI00
>> (SpecInt200) that should be published by an EGEE site. Especially for
>> heterogeneous sites this is not a trivial question.
>
> My preferred method is to run spec cpu2000 int and publish the result. I've
> run it with a "gcc -O2" compiler, since I felt it unlikely that a generic
> grid application is highly tuned for my CPU.
sometimes, with intel compiler, or some other commericial one, you have higher
specint value. and some vendor already have publish all possible model
available in spec.org, you always can turn to it if you're using widely
accepted machines. For atlas, as i know, they adopt this for basic ranking for
simul and reco jobs, that if site def didnt publish correct values accordingly,
you wont take jobs from robot. for your information only.
Br,
J
>
> Of course, I can do this only because I'm already sitting on a spec cpu2000
> license.
>
>> I am currently aware of http://goc.grid.sinica.edu.tw/gocwiki/kSpecInt and I
>> suppose that it is the unofficial effort.
>
> I'm not aware of a better one.
>
> /Mattias Wadenstein
>
-----
Jason Shih
ASGC/GD
Tel: +886-2-2788-0058 #1005 or 1006
Fax: +886-2-2789-6793
_______________________________________________
Roc mailing list
[log in to unmask]
http://lists.grid.sinica.edu.tw/mailman/listinfo/roc
|