Date: Sat, 30 Jul 2005 02:15:03 -0500 From: "Matthew D. Fuller" <fullermd@over-yonder.net> To: Vivek Khera <vivek@khera.org> Cc: freebsd-stable@freebsd.org Subject: Re: i386 vs amd64 - benchmark results Message-ID: <20050730071503.GA35495@over-yonder.net> In-Reply-To: <03E81E5E-7292-490E-8540-66B8CC4F98C4@khera.org> References: <42E67972.4000700@nurfuerspam.de> <03E81E5E-7292-490E-8540-66B8CC4F98C4@khera.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Jul 27, 2005 at 11:52:30AM -0400 I heard the voice of Vivek Khera, and lo! it spake thus: > > The amd64 memory architecture is NUMA -- that is, depending on how > your RAM is layed out, some of it is faster to access for each > processor. Accessing RAM "local" to the other processor(s) is > slower. On the other hand, I've heard from various sources (this is all pure hearsay, so trust it as much as it deserves) that in practice NUMAization in this case isn't really a gain. It's non-uniform, but it's not nearly as non-uniform as a lot of applications of the term, and the performance penalty is so small in absolute terms that the added complexity that comes with NUMA awareness can actually be enough to make it a net loss. But then, I usually don't know what I'm talking about 8-} -- Matthew Fuller (MF4839) | fullermd@over-yonder.net Systems/Network Administrator | http://www.over-yonder.net/~fullermd/ On the Internet, nobody can hear you scream.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050730071503.GA35495>