From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964935AbWD0F1z (ORCPT ); Thu, 27 Apr 2006 01:27:55 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S964936AbWD0F1z (ORCPT ); Thu, 27 Apr 2006 01:27:55 -0400 Received: from smtp101.mail.mud.yahoo.com ([209.191.85.211]:37300 "HELO smtp101.mail.mud.yahoo.com") by vger.kernel.org with SMTP id S964935AbWD0F1z (ORCPT ); Thu, 27 Apr 2006 01:27:55 -0400 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com.au; h=Received:Message-ID:Date:From:User-Agent:X-Accept-Language:MIME-Version:To:CC:Subject:References:In-Reply-To:Content-Type:Content-Transfer-Encoding; b=Sz4SmPdG4dBWvJfn3l9nFcflkNMF2RIeACpzIt/wbZ/ivkbfa38EmFU4G6xzDXV1owfTpxLX+1yrMklRg/IHyMybgsUVPziEtZca/1MxesrHRg6WNvE7K/K//4jLa6WsjyW5YrqAEEC9QE6p4WIWVpRKeHnHZkH66in5Qop8REk= ; Message-ID: <4450551D.5050000@yahoo.com.au> Date: Thu, 27 Apr 2006 15:22:37 +1000 From: Nick Piggin User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.12) Gecko/20051007 Debian/1.7.12-1 X-Accept-Language: en MIME-Version: 1.0 To: Andrew Morton CC: Jens Axboe , linux-kernel@vger.kernel.org, npiggin@suse.de, linux-mm@kvack.org Subject: Re: Lockless page cache test results References: <20060426135310.GB5083@suse.de> <20060426095511.0cc7a3f9.akpm@osdl.org> <20060426174235.GC5002@suse.de> <20060426111054.2b4f1736.akpm@osdl.org> In-Reply-To: <20060426111054.2b4f1736.akpm@osdl.org> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Andrew Morton wrote: >>The top of the 4-client >>vanilla run profile looks like this: >> >>samples % symbol name >>65328 47.8972 find_get_page >> >>Basically the machine is fully pegged, about 7% idle time. > > > Most of the time an acquisition of tree_lock is associated with a disk > read, or a page-size memset, or a page-size memcpy. And often an > acquisition of tree_lock is associated with multiple pages, not just a > single page. Still, most of the times it is acquired would be once per page for read, write, nopage. For read and write, often it will be a full page memcpy but even such a memcpy operation can quickly become insignificant compared to tl contention. Anyway, whatever. What needs to be demonstrated are real world improvements at the end of the day. > > So although the graph looks good, I wouldn't view this as a super-strong > argument in favour of lockless pagecache. No. Cool numbers though ;) -- SUSE Labs, Novell Inc. Send instant messages to your online friends http://au.messenger.yahoo.com