[Beowulf] Query regarding Linux clustering for CFD

[Reuti]

Hi,

Quoting "Lombard, David N" <david.n.lombard at intel.com>:

<snip>
> > I wish to know do i need to have any other software
> > for clustering or the lam mpi software can provide the
> > all the features for the scientific computation??
> 
> It depends.  How many people will be using the cluster?  If only you, or
> a small number of people, you could just manually schedule the jobs.
> Otherwise, consider a resource manager, like Torque, SGE, or other to
> schedule the cluster.

I'd would go for a queuing system in any case. It's also advantageous to have 
it on a single machine, just to serialize your workflow (instead of using the 
built in "batch" command in Linux or throwing n applications on the machine at 
once). You have of course in this case the overhead of getting familiar with 
the queuing system, unless you installed it already in a cluster - then it may 
be a couple minutes to put it to work on a dedicated workstation.

And thinking in long terms, I'd suggest to use SGE instead of Torque. On my own 
installations I use SGE, but I have access to some clusters with OpenPBS/Torque 
+ Maui. With this combination I see many idling nodes and endless waiting jobs; 
and checking with "qstat" and "showq", seems that Maui has a different view of 
the cluster than the underlying resource manager.

Cheers - Reuti