new SGI Origin & Onyx 3k?

W Bauske wsb at paralleldata.com
Wed Jul 26 13:08:55 PDT 2000


On Wed, 26 Jul 2000, Greg Lindahl wrote:
> 
> > > does anyone have experience, or even comments on SGI's new machines?
> >
> > Nothing that new -- just more of the same. The O3k brick scheme isn't as
> > flexible as you would hope because the router chips aren't running at full
> > speed yet. But it sure don't have anything to do with Beowulf clusters,
> > other than that you can sleep soundly knowing that everyone's favorite
> > whipping boy for bad price/performance on "MPI jobs with modest
> > communications and I/O needs" is still alive and kicking.
> >
> > > ob-beowulf: SGI docs actually
> > > call this stuff "commodity-off-the-shelf"!
> >
> > Of course. And it's "highly scalable" even though it poops out waaaay before
> > 1,000 nodes.
> 

Exactly how often do you think people want to run 1000 nodes on
a single problem? When was the last time you did that?

Realistically, the majority of parallel jobs are in range 4-128 
processors, depending on what you want to do and how the code scales. 
SGI's do fine in that space. So do other non-beowulf systems or
no one with sense would buy them.


Wes




More information about the Beowulf mailing list