Date: Wed, 12 Jul 2006 14:58:09 -0400 From: John Baldwin <jhb@freebsd.org> To: "Matthew D. Fuller" <fullermd@over-yonder.net> Cc: freebsd-amd64@freebsd.org Subject: Re: Turion 64 X2 support in future versions of FreeBSD. Message-ID: <200607121458.10311.jhb@freebsd.org> In-Reply-To: <20060712183343.GM98476@over-yonder.net> References: <93f8f44b0607061447k27adb556u306063b09b3d9b0d@mail.gmail.com> <200607121418.58293.jhb@freebsd.org> <20060712183343.GM98476@over-yonder.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wednesday 12 July 2006 14:33, Matthew D. Fuller wrote: > On Wed, Jul 12, 2006 at 02:18:57PM -0400 I heard the voice of > John Baldwin, and lo! it spake thus: > > On Wednesday 12 July 2006 00:37, Matthew D. Fuller wrote: > > > On Tue, Jul 11, 2006 at 11:57:49PM +0400 I heard the voice of > > > Michael, and lo! it spake thus: > > > > > > > > cpu0: timer 19005 129 > > > > cpu1: timer 9016 61 > > > > > > That looks a little odd... > > > > Possibly. Because cpu0's timer starts up sooner, it will generally > > have a higher total count (and uptime rate which is what vmstat -i > > shows you) than the other CPUs. It's hard to say if 10000 > > interrupts is normal for the differential though. That seems high. > > Well, it's not just the 10000; I've got a thousand differential on my > box (and that's with HZ=100). Of course, a 10,000 differential with > one of them not even having reached 10,000 yet is pretty big. But > look at the rate: 129? HZ=65? I doubt it... that sounds like > something REALLY weird going on. Probably using the default HZ of 1000 in which case the 10000 differential would seem sane according to your numbers (HZ 100 yields 1000 differential). The rate in 'vmstat -i' is misleading because it is the total interrupts divided by the uptime of the box. Recall that part of the boot is spent with interrupts disabled. systat -vmstat gives a much more usable interrupt rate. -- John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200607121458.10311.jhb>