From owner-freebsd-stable@FreeBSD.ORG  Mon Aug  9 14:38:18 2004
Return-Path: <owner-freebsd-stable@FreeBSD.ORG>
Delivered-To: freebsd-stable@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id E9D6716A4CE
	for <freebsd-stable@freebsd.org>;
	Mon,  9 Aug 2004 14:38:18 +0000 (GMT)
Received: from ux11.ltcm.net (ux11.ltcm.net [64.215.98.6])
	by mx1.FreeBSD.org (Postfix) with ESMTP id 39EFA43D3F
	for <freebsd-stable@freebsd.org>;
	Mon,  9 Aug 2004 14:38:18 +0000 (GMT)	(envelope-from mipam@ibb.net)
Received: from ux11.ltcm.net (mipam@localhost.ltcm.net [IPv6:::1])
	by ux11.ltcm.net (8.12.9/8.12.9/UX11TT) with ESMTP id i79EcGIo016647
	for <freebsd-stable@freebsd.org>;
	Mon, 9 Aug 2004 16:38:16 +0200 (MEST)
Received: from localhost (mipam@localhost)
	by ux11.ltcm.net (8.12.9/8.12.9/Submit) with ESMTP id i79EcE5k004855
	for <freebsd-stable@freebsd.org>;
	Mon, 9 Aug 2004 16:38:15 +0200 (MEST)
X-Authentication-Warning: ux11.ltcm.net: mipam owned process doing -bs
Date: Mon, 9 Aug 2004 16:38:14 +0200 (MEST)
From: Mipam <mipam@ibb.net>
X-X-Sender: mipam@ux11.ltcm.net
To: freebsd-stable@freebsd.org
Message-ID: <Pine.BSO.4.56.0408091628030.31478@ux11.ltcm.net>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Subject: 82541 Gi + device polling
X-BeenThere: freebsd-stable@freebsd.org
X-Mailman-Version: 2.1.1
Precedence: list
List-Id: Production branch of FreeBSD source code
	<freebsd-stable.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-stable>,
	<mailto:freebsd-stable-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-stable>
List-Post: <mailto:freebsd-stable@freebsd.org>
List-Help: <mailto:freebsd-stable-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-stable>,
	<mailto:freebsd-stable-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Mon, 09 Aug 2004 14:38:19 -0000

Hi,

I saw that the 82541 GI intel card is supported:
dev/em/if_em_hw.h:#define E1000_DEV_ID_82541GI             0x1076
in 5.2.1
However, with lots of traffic, lots of interrupts are being generated and
so lots of cpu power is consumed. Device polling would be a nice way 
around 
this. However, in the machine which contains these cards, SMP is enabled, 
cause its a hyperthreading machine. I saw that device polling is
not possible with smp enabled? And irq mitigation is also not included in 
these cards, so the old interrupt problem is still present with the modern 
card then? I think that the amount of interrupts consume more cpu, than
calculation of checksums. Maybe im wrong in this?
Any idea why intel quitted irq mitigation.
A combination of irq mitigation and checksum offloading would
really save cpu cycles.
Bye,

Mipam.