From owner-freebsd-net@FreeBSD.ORG  Tue Aug 23 11:26:30 2005
Return-Path: <owner-freebsd-net@FreeBSD.ORG>
X-Original-To: freebsd-net@freebsd.org
Delivered-To: freebsd-net@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id AE15B16A41F;
	Tue, 23 Aug 2005 11:26:30 +0000 (GMT) (envelope-from _pppp@mail.ru)
Received: from f31.mail.ru (f31.mail.ru [194.67.57.70])
	by mx1.FreeBSD.org (Postfix) with ESMTP id 4B08F43D45;
	Tue, 23 Aug 2005 11:26:30 +0000 (GMT) (envelope-from _pppp@mail.ru)
Received: from mail by f31.mail.ru with local 
	id 1E7WvD-000LZD-00; Tue, 23 Aug 2005 15:26:27 +0400
Received: from [212.5.170.174] by win.mail.ru with HTTP;
	Tue, 23 Aug 2005 15:26:27 +0400
From: dima <_pppp@mail.ru>
To: Mao Shou Yan <david.mao@thomson.net>
Mime-Version: 1.0
X-Mailer: mPOP Web-Mail 2.19
X-Originating-IP: [212.5.170.174]
Date: Tue, 23 Aug 2005 15:26:27 +0400
In-Reply-To: <31021C278A7A6B4AB95E9A085C3552180FF7C5@bjngsmail01>
Content-Type: text/plain; charset=koi8-r
Content-Transfer-Encoding: 8bit
Message-Id: <E1E7WvD-000LZD-00._pppp-mail-ru@f31.mail.ru>
Cc: freebsd-net@freebsd.org, freebsd-stable@freebsd.org
Subject: Re: polling in 4.11 vs 5.4
X-BeenThere: freebsd-net@freebsd.org
X-Mailman-Version: 2.1.5
Precedence: list
Reply-To: dima <_pppp@mail.ru>
List-Id: Networking and TCP/IP with FreeBSD <freebsd-net.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-net>,
	<mailto:freebsd-net-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-net>
List-Post: <mailto:freebsd-net@freebsd.org>
List-Help: <mailto:freebsd-net-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-net>,
	<mailto:freebsd-net-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Tue, 23 Aug 2005 11:26:30 -0000

>          I hope this is the right place that I post to!
> 
>  
> 
>          When I enabled polling in 5.4 and 4.11, I found that CPU load
> in 5.4 is much higher than 4.11. For example, suppose HZ is 5000, in 5.4
> the idle CPU is about 87%, but in 4.11, the idle CPU is about 99.9%. Is
> this right? Or the statistic is wrong in 4.11?
> 
>          BTW, ACPI is disabled under 5.4.
> 
> BTW, I found that context switch in 5.4 is much higher than 5.4 while
> enabling polling. (Using systat -vmstat 1)
> 
>          I'm looking forward your reply.
The polling code hasn't been changed since 4.11. I've posted a patch to resolve giant lock issue in it, but it wasn't merged. I don't have enough time by now to rewrite it according the opinions mentioned. But the cost of context switch is definitely higher on 5.x
HZ == 5000 is too much i think. 1000 is quite enough for network-intensive applications.

> 
>          Any response is welcome!