[Beowulf] Re: building a new cluster

[SC Huang]

A second graph, using 16M grid points, is posted:

http://www.geocities.com/schuang21/index.html

I have repeated the timing several times and confirmed all numbers.
However I still cannot come up with a "theory" to explain the curve for
Opteron in the case of 16M points: beyond 16 processors it goes up at
32 processors and then goes down at 64 processors. But going down is
good anyway... :-)



> 
> --- Tim Mattox <tmattox at gmail.com> wrote:
> 
> > Hello SCH,
> > That is an interesting graph.  Are you running a representative
> size
> > data set?
> > 2 Million gridpoints seems small compared to what I'm used to
> seeing
> > CFD
> > people run on our clusters.  But I don't know your code or problem
> > set,
> > so 2 Million may be just what you want to work with.
> > 
> > I say this because the wallclock times in your graph are rather
> > short, with
> > all but one of the runs on faster than 100 Mbit networks at under
> 10
> > seconds.
> > You may be seeing the job's startup overhead more than your actual
> > code's
> > behaviour.
> > 
> > If you can, try running a bigger job, and/or for more timesteps. 
> > More timesteps
> > will hide the startup overhead, and a bigger job (if that is
> > representative
> > of your desired runs) should allow you to scale to larger numbers
> of
> > nodes.
> > 







		
_______________________________
Do you Yahoo!?
Win 1 of 4,000 free domain names from Yahoo! Enter now.
http://promotions.yahoo.com/goldrush