From esr at thyrsus.com Fri Feb 4 19:05:01 2005 Content-Type: multipart/mixed; boundary="===============5992750353806537772==" MIME-Version: 1.0 From: Eric S. Raymond To: devel at lists.fedoraproject.org Subject: Re: x86_64 in FC4? Date: Fri, 04 Feb 2005 19:03:47 -0500 Message-ID: <20050205000347.GA21368@thyrsus.com> In-Reply-To: 1107560637.31589.10.camel@prospect.inetint.com --===============5992750353806537772== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable eli.carter(a)exgate.tek.com : > > Indeed. I have a single Opteron 3400 that is cranking a measured 1.65 = > > times better performance than my previous machine on doclifter, my = > > monster Python app -- which doesn't sound impressive until you = > > consider that the previous machine was a 2-processor SMP with 1.8GHz = > > Athlons. So my real-world throughput is as good or (allowing for = > > Amdahl's Law inefficiencies) better than what I'd get from a 6 GHz = > > Athlon, if such a thing existed. = > = > I was under the impression that the Python interpreter was serialized > and thus unable to take good advantage of SMP systems. > If that's the case, then you'd only be getting one of those 1.8GHz > processors beating on 'doclifter'... and if that's the case, you'd be > looking at a 3GHz Althon instead of 6GHz. Hmm...you're probably mostly right about this. The Python app would get = some gain from background daemons and X traffic migrating to the other processor, though. -- = Eric S. Raymond --===============5992750353806537772==--