From owner-freebsd-questions@FreeBSD.ORG  Wed Aug  3 09:23:12 2005
Return-Path: <owner-freebsd-questions@FreeBSD.ORG>
X-Original-To: freebsd-questions@freebsd.org
Delivered-To: freebsd-questions@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id 999E816A41F
	for <freebsd-questions@freebsd.org>;
	Wed,  3 Aug 2005 09:23:12 +0000 (GMT)
	(envelope-from victor@vmpbg.com)
Received: from 123bg.com (123bg.com [193.68.120.66])
	by mx1.FreeBSD.org (Postfix) with SMTP id 923EA43D4C
	for <freebsd-questions@freebsd.org>;
	Wed,  3 Aug 2005 09:23:11 +0000 (GMT)
	(envelope-from victor@vmpbg.com)
Received: (qmail 21491 invoked from network); 3 Aug 2005 09:23:09 -0000
Received: from host-201.teranet.evro.net (HELO neon.devian.bg) (83.148.125.201)
	by 0 with SMTP; 3 Aug 2005 09:23:09 -0000
From: Victor Semionov <victor@vmpbg.com>
Organization: Devian
To: Chuck Swiger <cswiger@mac.com>,
 freebsd-questions@freebsd.org
Date: Wed, 3 Aug 2005 12:23:19 +0300
User-Agent: KMail/1.8.1
References: <200508021537.26986.victor@vmpbg.com>
	<20050802173421.GA34971@alexis.mi.celestial.com>
	<42EFBF4A.2010209@mac.com>
In-Reply-To: <42EFBF4A.2010209@mac.com>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200508031223.20065.victor@vmpbg.com>
Cc: 
Subject: Re: polling decreases throughput ~50%
X-BeenThere: freebsd-questions@freebsd.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: User questions <freebsd-questions.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-questions>, 
	<mailto:freebsd-questions-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-questions>
List-Post: <mailto:freebsd-questions@freebsd.org>
List-Help: <mailto:freebsd-questions-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-questions>, 
	<mailto:freebsd-questions-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Wed, 03 Aug 2005 09:23:12 -0000

> >> Why is that? I thought polling should decrease CPU usage by avoiding too
> >> many context switches when a hw irq is generated frequently, but it
> >> shouldn't make the transfer slower if there are no other jobs running.
>
> You have to poll often enough to keep the pipe full, otherwise your max
> throughput can be limited.  Also, rl hardware isn't the greatest and
> probably requires a lot more CPU than a device with working buffer/DMA
> design.

HZ is 1000, which I guess should be more than enough with 
kern.polling.burst_max=150.

Indeed, it was hardware's fault - my other NIC is a fxp and I got much better 
results with it - less CPU, while throughput stayed the same as without 
polling.