[Beowulf] Re: Cluster newbie, power recommendations

[Douglas Eadline]

> Charlie Peck <charliep at cs.earlham.edu> wrote
>
>> I think clusters like the one Eric wants to build have /significant/
>> educational value, both in the building and the use.  How else does one
>> learn to do parallel/distributed programming if not on a cluster, even
>> a "toy" one?  Sure the single AMD64 will be more powerful but it won't
>> provide an opportunity to learn about message passing, speedup,
>> efficiency, problem decomposition, etc.
>
> Interesting question that, how else might this be done?
>
> I wonder if one couldn't set up a single modern computer,
> with a fast CPU and tons of memory, as N virtual machines, for
> instance using VMware, and then run a sort of virtual cluster?
> Obviously there wouldn't be any performance advantage to doing
> this but it might allow the subject to be taught when a real
> clusters isn't available.
>

Why bother with virtualization? You can do this with any of the freely
available MPI versions. If you want to learn about writing MPI codes,
it is easy enough to oversubscribe a single computer with multiple MPI
processes. Of course you will need to keep an eye on how much memory you
are using, but you could run a reasonable number of tasks to start
learning MPI programming. Speed-up and other "cluster concepts" can
come along later. Indeed, with multi-core systems it may become
more than just an academic exercise.

You can find more on this idea at the Monkey:

   MPI: Processes, Processors, and MPI, Oh My!
   (http://www.clustermonkey.net//content/view/26/32/)


-- 
Doug