Alessandra, the approximation of 4 is just to allow a rough comparison
between sites which are doing their accounting based on SI2K and those
who have measured HEPSPEC06. Once everyone is using HEPSPEC06 then
comparison with the old numbers is irrelevant. At any stage we can
change the portal to report HEPSPEC06-hours, converting the other way if
necessary.
GridPP has solved the problem internally by asking you all to coordinate
the change today. Thus all GridPP sites are basing the kSI2K numbers
they publish on HEPSPEC06 measurements so the comparisons are accurate
no matter what scale factor you use to convert.
John
> -----Original Message-----
> From: Testbed Support for GridPP member institutes [mailto:TB-
> [log in to unmask]] On Behalf Of Alessandra Forti
> Sent: 01 October 2009 09:50
> To: [log in to unmask]
> Subject: Re: HEPSPEC06 and kSI2k Publishing
>
> Hi John,
>
> nobody is against moving to spec06. It is the quite heavily
> approximated
> conversion factor of 4 that leaves me personally without words and the
> fact that that is considered more accurate than an educated guess on
an
> architecture. Manchester got the numbers from DELL and they are
> consistent with the WLCG table and we have 4.44, Sheffield has 4.5.
> Today I'll go through Lancaster and Liverpool.
>
> > Now what we do about this discrepancy in past accounting? That's a
> > different question.
>
> indeed.
>
> cheers
> alessandra
>
> John Gordon wrote:
> > Which is too high and which too low? You measured the HEPSPEC06 (I
> > hope!!). Did you measure SI2K or just take a number from some
> suppliers
> > web page?
> >
> > If we all actually measure our resources then we will have a
> consistent
> > set of numbers with which to compare sites. So some of the old ones
> were
> > wrong? That's what the exercise was about - SI2K was no longer
> accepted
> > as a good measure of the power of modern systems to run HEP codes.
> >
> >
> > John
> >
> >
> >> -----Original Message-----
> >> From: Testbed Support for GridPP member institutes
> >> [mailto:[log in to unmask]] On Behalf Of Alessandra Forti
> >> Sent: 01 October 2009 01:02
> >> To: [log in to unmask]
> >> Subject: Re: HEPSPEC06 and kSI2k Publishing
> >>
> >>
> >> > Looking at the SL4 kSI2k numbers we used and the HEPSPEC06
> >> numbers we got it seems to be more like 4.5 to 4.6 so > our
> >> recalculated kSI2k numbers are 10-15% too high.
> >>
> >> You are not the only one. A lot of sites have a different
> >> conversion factor.
> >>
> >> cheers
> >> alessandra
> >>
> >> --
> >> Mindmelds. The last time I heard the words "my mind to your
> >> mind", I had a headache for two weeks. (Janeway, ST Voyager)
> >>
> >> Northgrid Tier2 Technical Coordinator
> >> http://www.hep.manchester.ac.uk/computing/tier2
> >>
> >>
>
> --
> Mindmelds. The last time I heard the words "my mind to your mind", I
> had a headache for two weeks. (Janeway, ST Voyager)
>
> Northgrid Tier2 Technical Coordinator
> http://www.hep.manchester.ac.uk/computing/tier2
--
Scanned by iCritical.
|