From mboxrd@z Thu Jan 1 00:00:00 1970 From: Hagen Paul Pfeifer Subject: Re: [PATCH next-next-2.6] netdev: better dev_name_hash Date: Sun, 25 Oct 2009 21:17:14 +0100 Message-ID: <20091025201714.GA20987@nuttenaction> References: <200910252158.53921.opurdila@ixiacom.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: netdev@vger.kernel.org To: Octavian Purdila Return-path: Received: from alternativer.internetendpunkt.de ([88.198.24.89]:51111 "EHLO geheimer.internetendpunkt.de" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1752327AbZJYUXz (ORCPT ); Sun, 25 Oct 2009 16:23:55 -0400 Content-Disposition: inline In-Reply-To: <200910252158.53921.opurdila@ixiacom.com> Sender: netdev-owner@vger.kernel.org List-ID: * Octavian Purdila | 2009-10-25 21:58:53 [+0200]: > >The current dev_name_hash is not very good at spreading entries when a >large number of interfaces of the same type (e.g. ethXXXXX) are used. > >Here are some performance numbers for creating 16000 dummy interfaces with >and without the patch (with per device sysctl entries disabled) > > With patch Without patch > > real 0m 2.27s real 0m 4.32s > user 0m 0.00s user 0m 0.00s > sys 0m 1.13s sys 0m 2.16s Can you rerun the test with jhash() as the hash function? Just for clearness - the spreading of jhash should be more uniformly distributed. At the end: where is the threshold where a more accurate hash function is superior. HGN -- Hagen Paul Pfeifer || http://jauu.net/ Telephone: +49 174 5455209 || Key Id: 0x98350C22 Key Fingerprint: 490F 557B 6C48 6D7E 5706 2EA2 4A22 8D45 9835 0C22