Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 22 May 2001 23:44:50 -0500
From:      Mike Meyer <mwm@mired.org>
To:        Kent Stewart <kstewart@urx.com>
Cc:        questions@freebsd.org
Subject:   Re: Dual Ppro motherboards
Message-ID:  <15115.16450.249394.115359@guru.mired.org>
In-Reply-To: <6338897@toto.iv>

next in thread | previous in thread | raw e-mail | index | archive | help
Kent Stewart <kstewart@urx.com> types:
> I have always thought of MP systems as something that didn't run anything
> faster but what they would do is run additional applications equally slowly.

Yup, that's what they are. They're really nice if you have lots of
applications that are otherwise i/o bound. For typical single-user
use, they aren't so great. For Unix power users - who have a tendency
to have other things going in the background - an extra processor is
also a nice thing to have.

Of course, if you're developing applications that run in one of those
environments, having a second processor so you can do some testing in
the environment is also nice. That's why I bought one. When it's not
fast enough to be a workstation anymore, I'll probably go buy a
single-CPU system, and keep this one around for a test system. It's
going to be interesting to see what causes that to happen.

> The compiler's on Cray's were the best the world had to offer in 1990.
> However, when you turned on full MP and vector support, a 486-PC would out
> compile it on a line for line basis. The PC was fast because it didn't have
> to worry about pipelines and producing do-loops that could be run on more
> than one processor.

Another thing to consider is the architecture of the two machines.
Crays was designed for crunching numbers, generally in large arrays.
Doing character manipulation was a lot of work - you had to shift the
word to the character you wanted then mask it off, so that while you
were doing a lot more instructions per second than a conventional
machine, you were also doing a lot more instructions for each unit of
work. The end result was that unix character-crunching benchmarks on a
Cray performed about as well as a fast minicomputer of the time.

	<mike
--
Mike Meyer <mwm@mired.org>			http://www.mired.org/home/mwm/
Independent WWW/Perforce/FreeBSD/Unix consultant, email for more information.

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?15115.16450.249394.115359>