Date: Tue, 29 Apr 1997 11:23:27 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: brownie@earthling.net Cc: terry@lambert.org, FreeBSD-SMP@FreeBSD.org Subject: Re: SMP Message-ID: <199704291823.LAA04826@phaeton.artisoft.com> In-Reply-To: <33659096.10DA@earthling.net> from "Chris Browning" at Apr 28, 97 11:09:26 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > Compare this to Sequent's Dynix, which scales well to 32 processors, > > or Thinking Machines, which scales into the thousands of processors. > > The Sequent S81 at Georgia Tech, which uses Dynix, is not an SMP > machine. Some of it's processor us a master-slave relationship. This must be an older machine with the I/O processor relationship. I assume it's a 386 based box? > Scheduling occurs on the masters. I believe the CM's also had > some dedicated nodes for different purposes, thus not making them > SMP. The Connection Machines have processor local memory. You're right that they aren't true SMP, at least at the OS level. The issue was concurrency and scalability, though, not ethnic purity of the implementation. I think the Sandia labs machine isn't very "pure" either (I think it has cluster local memory). For what it's worth, I think that a hybrid system is probably the best trade off, in any case, for doing things which are inherently parallelizable (fluidic modelling, laminar air flow, factoring, etc.). This is getting off topic, but I admit I started it when I whined about the PC architecture. 8-). Regards, Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199704291823.LAA04826>