From owner-freebsd-hardware Fri Oct 18 10: 5:52 2002 Delivered-To: freebsd-hardware@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 5BFA337B401; Fri, 18 Oct 2002 10:05:50 -0700 (PDT) Received: from search.sparks.net (d-207-5-180-136.gwi.net [207.5.180.136]) by mx1.FreeBSD.org (Postfix) with ESMTP id EE45843E9C; Fri, 18 Oct 2002 10:05:49 -0700 (PDT) (envelope-from dmiller@search.sparks.net) Received: by search.sparks.net (Postfix, from userid 100) id D7197D999; Fri, 18 Oct 2002 12:59:27 -0400 (EDT) Received: from localhost (localhost [127.0.0.1]) by search.sparks.net (Postfix) with ESMTP id D1CADD98C; Fri, 18 Oct 2002 12:59:27 -0400 (EDT) Date: Fri, 18 Oct 2002 12:59:27 -0400 (EDT) From: David Miller To: Attila Nagy Cc: "Kenneth D. Merry" , Christopher Smith , hardware@FreeBSD.ORG, net@FreeBSD.ORG Subject: Re: High interrupt load on firewalls In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-hardware@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org On Wed, 9 Oct 2002, Attila Nagy wrote: With a dc ethernet card and ~45K packets per second, an XP1700 system went from > 50% interrupt to < 1%. I was astounded at the change! If all it takes to get Gb interfaces polling is to send Luigi a card then he needs to send me his shipping address:) --- David > Hello, > > > You might want to try out some of the Intel gigabit boards. At least > > we've got an engineer from Intel who maintains the driver. > I'm far from being a FreeBSD expert, but Luigi Rizzo's polling patch > helped me a lot in similar cases to get better performance. > > >From POLLING(4): > DESCRIPTION > "Device polling" (polling for brevity) refers to a technique to handle > devices that does not rely on the latter to generate interrupts when they > need attention, but rather lets the CPU poll devices to service their > needs. This might seem inefficient and counterintuitive, but when done > properly, polling gives more control to the operating system on when and > how to handle devices, with a number of advantages in terms of system > responsivity and performance. > > AFAIK there are only two problems about that: > SUPPORTED DEVICES > Polling requires explicit modifications to the device drivers. As of > this writing, the dc, fxp, rl and sis devices are supported, with other > in the works. > > (no Gigabit cards) > > and > sys/kern/kern_poll.c: > [...] > #ifdef SMP > #include "opt_lint.h" > #ifndef COMPILING_LINT > #error DEVICE_POLLING is not compatible with SMP > #endif > #endif > [...] > > (no SMP support) > > As stated above, I'm not an expert, but my opinion is that polling would > be the best on the Gigabit cards (due to their capacity) and not the Fast > Ethernet ones. > > Maybe we could start sending GE hardware to Luigi :) > > ----------[ Free Software ISOs - http://www.fsn.hu/?f=download ]---------- > Attila Nagy e-mail: Attila.Nagy@fsn.hu > Free Software Network (FSN.HU) phone @work: +361 210 1415 (194) > cell.: +3630 306 6758 > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-hardware" in the body of the message > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hardware" in the body of the message