[rescue] Perverse Question

[Francisco Javier Mesa-Martinez]

On Thu, 19 Jun 2003, Charles Shannon Hendrix wrote:
>
> Sometimes it is, sometimes it isn't.
>
> Embarrasingly parallel is what they are good at, but even then there is
> no garantee it will be better than something like an Origin.

You'd be amazed, that is one of the reasons why SGI is have a hard time
selling them. Origins can no longer keep up, in most EP fields. Their lack
of processing power per node can not longer be compesated by the nice
internodal bandwidth.

So yeah, for the most part beowulfs do make sense for these sorts of
problems.

> The whole beowulf thing is surrounded by a lot of dangerous hype spread
> by clueless people who have never actually set one up.

Same goes for almost any sort of high performance computing...

> Some locals got burned bad on beowulf systems, because they read all the
> $50K supercomputer hype and believed it.

Again, the correct tool for the correct job. My previous lab got burned
with a $250K SGI... so the same goes for every platform.

> In any case, I wasn't suggesting anyone here is an idiot at all, I just
> said a lot of people don't understand much about either type of system,
> and I stand by that.

Most of the clusters I have seen deployed in Academia have been done by
very competent people. I understand what your point is, but that is true
for almost every computing platform.