From owner-freebsd-net Wed Feb 5 1:19:30 2003 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 AAEAB37B405 for ; Wed, 5 Feb 2003 01:19:29 -0800 (PST) Received: from sunu007.rz.ruhr-uni-bochum.de (sunu007.rz.ruhr-uni-bochum.de [134.147.64.14]) by mx1.FreeBSD.org (Postfix) with SMTP id 4298443FC3 for ; Wed, 5 Feb 2003 01:19:28 -0800 (PST) (envelope-from jens.wiggers@ruhr-uni-bochum.de) Received: (qmail 15612 invoked by uid 82); 5 Feb 2003 09:19:21 -0000 Received: from jens.wiggers@ruhr-uni-bochum.de by mailhost with qmail-scanner-1.00 (uvscan: v4.1.40/v4245. . Clean. Processed in 0.468764 secs); 05 Feb 2003 09:19:22 -0000 Received: (qmail 15557 invoked from network); 5 Feb 2003 09:19:21 -0000 Received: from act6.etdv.ruhr-uni-bochum.de (134.147.40.121) by mailhost.rz.ruhr-uni-bochum.de with SMTP; 5 Feb 2003 09:19:21 -0000 From: "Jens Wiggers" To: , Subject: Correlation between CPU load and network usage Date: Wed, 5 Feb 2003 10:20:50 +0100 Message-ID: <000001c2ccf7$dfcdcc70$1d01a8c0@iis2> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook, Build 10.0.2627 Importance: Normal X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200 X-Virus-Scanned: by AMaViS perl-11 Sender: owner-freebsd-net@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org Hi, I'm using two FreeBSD machines in an small switched network. One of them blows out tcp packets as fast as possible for five seconds. The other machine just receives the packets and does nothing else. What I see is that for short packets the senders CPU load is just 100 percent, but for larger packets the senders CPU load drops down rapidly. >From the applications point of view the data to send gets into the socket buffer and via tcp_output, ip_output and ether_output into the interface buffer. The rest of work is done by the interrupt handler of the network controller. Therefore the application returns from the socket send syscall and calls it again. So where is sparetime for the CPU in this cycle? Any help would be appreciated. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message