Beowulf on Sparc

[kragen at pobox.com]

"Mattson, Timothy G" <timothy.g.mattson at intel.com> writes:
> If you look at those old workshop proceedings, you'll see that all the major
> components of a Beowulf cluster were available and in use in those days.

The major components of a Beowulf cluster are mass-market commodity
hardware (which wasn't fast enough in those days), free software to
run the thing (which didn't exist), dedicated nodes, and a dedicated
network.

> But in terms of what the user of a cluser sees --- the parallel
> algorithms, parallel programming environments, batch-queue/job-
> management, and cluster management systems -- we had it all way back
> in the early 90's.

There are really important things the users of a Beowulf see that the
users of Sun clusters didn't --- much lower startup and maintenance
costs per megaflop and the assurance that your cluster isn't dependent
on a particular vendor for its future, for example.

> P.S. So now I'm ready for someone from DEC to tell me that clustering even
> predates SPARC.  To a certain extent, that may be the case, but the old DEC
> clusters were not "shared nothing" clusters and were quite different beasts
> -- weren't they?  

How about Tandem?  Have they been doing shared-nothing
clusters-in-a-box for longer than the PVM people, or is it the other
way around?

> Disclaimer: The opinions in this message are my own and do not reflect the
> views of my employer.

Nor of Microsoft, despite what someone thought last time, right? :)