From owner-freebsd-net  Sat Jan  8 13:26:37 2000
Delivered-To: freebsd-net@freebsd.org
Received: from cs.rice.edu (cs.rice.edu [128.42.1.30])
	by hub.freebsd.org (Postfix) with ESMTP id 55A3A152AC
	for <freebsd-net@freebsd.org>; Sat,  8 Jan 2000 13:26:35 -0800 (PST)
	(envelope-from aron@cs.rice.edu)
Received: (from aron@localhost)
	by cs.rice.edu (8.9.0/8.9.0) id PAA17320;
	Sat, 8 Jan 2000 15:26:31 -0600 (CST)
From: Mohit Aron <aron@cs.rice.edu>
Message-Id: <200001082126.PAA17320@cs.rice.edu>
Subject: Re: performance of FreeBSD-current as SMP
To: wes@softweyr.com (Wes Peters)
Date: Sat, 8 Jan 2000 15:26:31 -0600 (CST)
Cc: freebsd-net@freebsd.org
In-Reply-To: <3877AB9F.E132C12C@softweyr.com> from "Wes Peters" at Jan 8, 2000 02:26:55 pm
X-Mailer: ELM [version 2.4 PL25]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-net@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org


> 
> Not if defining SMP support causes the kernel to compile in all of the lock
> acquisition and release code.  Even if it simply does "if (ncpus < 2) return"
> it would still impact performance somewhat.
> 


Right, but the fall shouldn't be as much as 22%. According to Garret's earlier
posting, there seem to be some very expensive Intel instructions being used for
locking. Or possibly, the lock acquisition is done at too many places.



- Mohit


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-net" in the body of the message