From owner-freebsd-current@FreeBSD.ORG  Mon Mar  3 22:19:10 2008
Return-Path: <owner-freebsd-current@FreeBSD.ORG>
Delivered-To: current@freebsd.org
Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34])
	by hub.freebsd.org (Postfix) with ESMTP id 785D310657B4;
	Mon,  3 Mar 2008 22:19:10 +0000 (UTC)
	(envelope-from kris@FreeBSD.org)
Received: from weak.local (freefall.freebsd.org [IPv6:2001:4f8:fff6::28])
	by mx1.freebsd.org (Postfix) with ESMTP id 8C73A8FC15;
	Mon,  3 Mar 2008 22:19:09 +0000 (UTC)
	(envelope-from kris@FreeBSD.org)
Message-ID: <47CC795C.9050600@FreeBSD.org>
Date: Mon, 03 Mar 2008 23:19:08 +0100
From: Kris Kennaway <kris@FreeBSD.org>
User-Agent: Thunderbird 2.0.0.12 (Macintosh/20080213)
MIME-Version: 1.0
To: Bakul Shah <bakul@bitblocks.com>
References: <20080303211731.D89A05B30@mail.bitblocks.com>
In-Reply-To: <20080303211731.D89A05B30@mail.bitblocks.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Cc: gnn@freebsd.org, current@freebsd.org
Subject: Re: Differences in malloc between 6 and 7?
X-BeenThere: freebsd-current@freebsd.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Discussions about the use of FreeBSD-current
	<freebsd-current.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-current>, 
	<mailto:freebsd-current-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-current>
List-Post: <mailto:freebsd-current@freebsd.org>
List-Help: <mailto:freebsd-current-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-current>,
	<mailto:freebsd-current-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Mon, 03 Mar 2008 22:19:10 -0000

Bakul Shah wrote:
> On Mon, 03 Mar 2008 15:23:33 EST gnn@freebsd.org  wrote:
>> One of the folks I'm working with found this.  The following code,
>> which yes, is just an example, is 1/2 as fast on 7.0-RELEASE as on
>> 6.3.  Where should I look to find out why?
> 
> Specifying malloc option K (double virtual memory chunk size)
> roughly halves the runtime.  Additional Ks reduce it more and
> more.  As this test spends most of its time in the kernel,
> may be this is just due to the mmap overhead? Or may be the
> defaults for 6.3 were different.

Well, the whole architecture of malloc is different.

I also see a big performance cliff (drops by a factor of 10) when malloc 
size exceeds the chunk size (1MB by default).  Also concurrent access to 
mmapped memory performs badly in FreeBSD right now.  I have patches that 
convert the vm_map lock to a sx to avoid this contention, but they need 
some more work.

Kris