Hi Pete, Stephen,
Peter Gronbech wrote:
> Re: HEPSPEC06 and kSI2k Publishing
>
> Why should the apel numbers be over. Surely if your scaling is
> working the number of hours the cpu does will be reduced (multiplied
> by say 0.8) and hence hours times spec int 2000 figure should equal
> the equivalent of the real thing.
>
> Eg measured hepspec06 of 8 If the cpu works for 10 hours and was
> publishing the equivalent specint 2k of 2000 would equal 20ksi2k
> hours
>
> In your case. The scaling multiple woyld bring the hours down to 8
> multiplied by 2500 equals 20ksi2k
>
> Pete
I'm not too worried about the scaling part - I'm pretty sure I'm
publishing HEPSPEC06 Hours OK - it's the suspiciously round factor of 4
to convert between HEPSPEC06 and kSI2k. 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.
Stephen Burke wrote:
> Re: HEPSPEC06 and kSI2k Publishing
>
> Testbed Support for GridPP member institutes
> > [mailto:[log in to unmask]] On Behalf Of
> > [log in to unmask]
> said:
> > 3) We should be publishing: GlueHostBenchmarkSI00: 2500
> > GlueHostProcessorOtherDescription:
> > Cores=4,Benchmark=10.00-HEP-SPEC06 GlueCECapability:
> > CPUScalingReferenceSI00=2500
> >
> > Looking at the old scaling numbers and reference of 1000SI2k and
> > the new scaling numbers I think this will overestimate our APEL
> > numbers by about 25%.
> >
> > Is all that reasonable?
>
> Not entirely ... the thing you publish in Benchmark=10.00-HEP-SPEC06
> is not supposed to be the reference value but the average, i.e. it
> should be possible to calculate the total installed power by
> multiplying that value by whatever you have in LogicalCPUs (which
> should be the total number of cores). Potentially you could take the
> average as the reference but it seems that you don't.
>
> Stephen -- Scanned by iCritical.
>
OK, I'll correct the value. Taking a round number reference makes things
a lot easier to understand.
But I assume that before APEL starts to use the HEPSPEC06 numbers there
will be a new CPUScalingReferenceHS06 value published to correct our cpu
and wall times.
Yours,
Chris.
|