Fw: [Beowulf] Correct networking solution for 16-core nodes

[Greg Lindahl]

On Fri, Aug 04, 2006 at 08:52:06PM +0100, Vincent Diepeveen wrote:

> Read the subject. you posted onto this list answerring a private email of 
> mine
> which i had shipped to the guy starting this thread.

Vincent, please go into the archive and find where *I* posted a
private email. *I* did not. Nor do I retaliate against people who post
such emails; I call them out in public, just like I'm calling you out.

> You don't design that thing to let it look good on paper.

No, the goal was to do great on real applications. And that
performance does not depend on answering your theoretical
objections.

If you'd like to see how we perform on real apps up to 512 cores,
please see our whitepapers. By SC, we'll have published results, real
application results, for much larger systems.

If you think about it, castigating the interconnect vendor which
publishes the most performance data (microbenchmarks, HPCC, real apps)
for not understanding real apps is, well, funny.

-- greg
(stuck in Phoenix on the way to vacation, at least there's free wireless)