[Aces-support] processor speed of hardware groups

[Peter H Israelsson]

Hi Ed,

Thanks for your reply.  I am surprised to hear that there are no appreciable
difference in cpu speeds between individual machines.  I'll look into 
profiling
statistics when I get a chance, although I am judging the code as running
"slowly" relative to my experience running on other PCs and linux clusters
(i.e., I am satisfied with the code's performance on other machines).  Don't
think I'll be able to stop by during office hours this week, but perhaps next
week.

Thanks again.

Peter


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  Peter H. Israelsson
  Massachusetts Institute of Technology
  Department of Civil & Environmental Engineering
  48-114, 15 Vassar Street, Cambridge, MA 02139, USA
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


Quoting Ed Hill <ed at eh3.com>:

> On Fri, 2005-07-29 at 10:54 -0400, Peter H Israelsson wrote:
>> Hello,
>>
>> I am new to the ACES system, and have begun running simulations on
>> ao.acesgrid.org.  My simulations are running quite slowly, which I 
>> imagine has
>> to do with the processor speed of the hardware group that I am 
>> using.  When I
>> run my simulations, I have not been specifying a hardware group, 
>> just the queue
>> (I have been using "four" and "long").  Looking at the various 
>> hardware groups
>> available, I am not clear on which ones have the fastest processors.  I run
>> single processor jobs only.  Any recommendations?
>>
>> Obviously I am not looking to hog the fastest machines, but if they are just
>> sitting around idle I would like to make use of them (and from what 
>> I can see
>> from "qstat" there have not been many jobs running in the past few days).
>>
>> Thanks in advance for your help.
>
>
> Hi Peter,
>
> For the most part, the differences between the machines within the ACES
> queues are rather small -- the CPUs are relatively homogeneous.  So
> there probably isn't a lot to be gained (in terms of CPUs) by fussing
> with hardware groups within the queues.
>
> In terms of performance, there are a number of things that could be
> causing bottlenecks for you and CPU speed is only one of them.  Have you
> spent any time profiling your code and looking for bottlenecks?  Are you
> certain that your performance problems aren't related to something
> entirely different such as I/O, memory footprint (swapping), or memory
> bandwidth?
>
> If you haven't already, please spend a little time profiling your code
> and then come to the next weekly ACES office hours where it'll be easier
> to discuss this and perhaps find a better setup (or better machines) for
> you.
>
> Ed
>
> --
> Edward H. Hill III, PhD
> office:  MIT Dept. of EAPS;  Rm 54-1424;  77 Massachusetts Ave.
>             Cambridge, MA 02139-4307
> emails:  eh3 at mit.edu                ed at eh3.com
> URLs:    http://web.mit.edu/eh3/    http://eh3.com/
> phone:   617-253-0098
> fax:     617-253-4464
>
> _______________________________________________
> Aces-support mailing list
> Aces-support at acesgrid.org
> http://acesgrid.org/mailman/listinfo/aces-support
>