From owner-freebsd-hackers Thu Jun 6 17: 2:34 2002 Delivered-To: freebsd-hackers@freebsd.org Received: from mail.speakeasy.net (mail16.speakeasy.net [216.254.0.216]) by hub.freebsd.org (Postfix) with ESMTP id 7867337B404 for ; Thu, 6 Jun 2002 17:02:26 -0700 (PDT) Received: (qmail 23759 invoked from network); 7 Jun 2002 00:02:25 -0000 Received: from unknown (HELO server.baldwin.cx) ([216.27.160.63]) (envelope-sender ) by mail16.speakeasy.net (qmail-ldap-1.03) with DES-CBC3-SHA encrypted SMTP for ; 7 Jun 2002 00:02:25 -0000 Received: from laptop.baldwin.cx (laptop.baldwin.cx [192.168.0.4]) by server.baldwin.cx (8.11.6/8.11.6) with ESMTP id g5702NF63163; Thu, 6 Jun 2002 20:02:23 -0400 (EDT) (envelope-from jhb@FreeBSD.org) Message-ID: X-Mailer: XFMail 1.5.2 on FreeBSD X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: <20020606230545.A78993@dogma.freebsd-uk.eu.org> Date: Thu, 06 Jun 2002 20:02:22 -0400 (EDT) From: John Baldwin To: j mckitrick Subject: RE: SMP/5.0 performance on single CPU? Cc: freebsd-hackers@freebsd.org Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG On 06-Jun-2002 j mckitrick wrote: > > Hi all, > > I'm sure the SMP features along with removal of the Big Giant Lock (TM) > are designed to sacrifice as little performance as possible across the > board. In practical terms, will a 5.0 kernel without SMP have any > performance loss at all on a single CPU machine compared with the 4.x > kernel we use now? I once read here on one of these lists that the > Solaris fine-grained locking was great for SMP, but was a lot of > overhead for single CPU machines (correct me if I misunderstood). At the moment it is slower because interrupts are effectively blocked in almost all of the kernel. However, the SMP work is still in progress and it will get better as time goes on. One thing to note is that in 4.x, if we are in the networking code, for example, we block all network interrupts. When the more-finely grained locking is implemented in 5.x, if we are working on network card A, we won't be effectively blocking network interrupts on card B, while in 4.x we would be blocking interrupts on card B. However, most of the kernel is still under Giant and so this does not apply to 5.0 yet. -- John Baldwin <>< http://www.FreeBSD.org/~jhb/ "Power Users Use the Power to Serve!" - http://www.FreeBSD.org/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message