On Tue, 21 Aug 2007, Wei Yang wrote:
> Hi Stephen,
>
> Can you try rhel4-32 and rhel4-64? Do you need LSF on them?
No LSF required.
> You might have to run klog after login. If the above machines don't fit your
> need and you want a boer for a while, I think we can arrange it.
Sasah Telnov is also logged in to the rhel4-* machines. They are also the
same as yakut. I wanted a boer-like machine to do a similar comparison
with dual-cores. So if you could put one of those aside that would be
good.
> Wei Yang | [log in to unmask] | 650-926-3338(O)
>
>
> Stephen J. Gowdy wrote:
>> On Tue, 21 Aug 2007, Young, Charles C. wrote:
>>
>> > The 15% edge over Opteron is for running one job, right? When using all
>> > 8 cores, that becomes roughly 1/2 or 7.5%. The yakut number is for one
>> > job, but there are probably other loads. So it's not clear which number
>> > is more relevant.
>>
>> Well, the yakut was busy I used so you can't compare the Wall Time. The
>> 15% edge I expected to be reduced by the 5% loss, so down to 10%. I should
>> try a multicore Opteron and repeat multiple jobs there. Can I have one of
>> the new-ish machines to do that?
>
>
--
/------------------------------------+-------------------------\
|Stephen J. Gowdy, SLAC | CERN Office: 32-2-A22|
|http://www.slac.stanford.edu/~gowdy/ | CH-1211 Geneva 23 |
| | Switzerland |
|EMail: [log in to unmask] | Tel: +41 22 767 5840 |
\------------------------------------+-------------------------/
|