From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753603AbbB0D2z (ORCPT ); Thu, 26 Feb 2015 22:28:55 -0500 Received: from mga14.intel.com ([192.55.52.115]:17923 "EHLO mga14.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753209AbbB0D2y (ORCPT ); Thu, 26 Feb 2015 22:28:54 -0500 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.09,656,1418112000"; d="scan'208";a="533547336" Message-ID: <1425007731.10337.50.camel@intel.com> Subject: Re: [LKP] [futex] 76835b0ebf8: -8.1% will-it-scale.per_thread_ops From: Huang Ying To: Catalin Marinas Cc: Linus Torvalds , LKML , LKP ML Date: Fri, 27 Feb 2015 11:28:51 +0800 In-Reply-To: <20150226095812.GA32538@localhost> References: <1424927813.10337.19.camel@intel.com> <20150226095812.GA32538@localhost> Content-Type: text/plain; charset="UTF-8" X-Mailer: Evolution 3.12.9-1+b1 Mime-Version: 1.0 Content-Transfer-Encoding: 8bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi, Catalin, On Thu, 2015-02-26 at 09:58 +0000, Catalin Marinas wrote: > On Thu, Feb 26, 2015 at 05:16:53AM +0000, Huang Ying wrote: > > FYI, we noticed the below changes on > > > > git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master > > commit 76835b0ebf8a7fe85beb03c75121419a7dec52f0 ("futex: Ensure get_futex_key_refs() always implies a barrier") > > > > > > testbox/testcase/testparams: lkp-wsx01/will-it-scale/performance-futex4 > > > > 0429fbc0bdc297d6 76835b0ebf8a7fe85beb03c751 > > ---------------- -------------------------- > > %stddev %change %stddev > > \ | \ > > 6314259 ± 0% -8.1% 5800079 ± 0% will-it-scale.per_thread_ops > > 6274871 ± 0% -8.1% 5768747 ± 0% will-it-scale.per_process_ops > > This was raised before but I'm not sure we can do much about it: > > https://lkml.org/lkml/2014/10/26/189 Sorry for duplicated report. > It would be interesting to know the difference in performance with the > code prior to commits b0c29f79ecea and 11d4616bd07f which optimised the > futex code but missed a barrier. I will do some test for that. Best Regards, Huang, Ying