From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933199AbbFWNCs (ORCPT ); Tue, 23 Jun 2015 09:02:48 -0400 Received: from e23smtp03.au.ibm.com ([202.81.31.145]:46381 "EHLO e23smtp03.au.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932861AbbFWNCT (ORCPT ); Tue, 23 Jun 2015 09:02:19 -0400 X-Helo: d23dlp02.au.ibm.com X-MailFrom: srikar@linux.vnet.ibm.com X-RcptTo: linux-kernel@vger.kernel.org Date: Tue, 23 Jun 2015 18:31:14 +0530 From: Srikar Dronamraju To: Ingo Molnar Cc: Rik van Riel , Peter Zijlstra , linux-kernel@vger.kernel.org, Mel Gorman Subject: Re: [PATCH v2 2/4] sched:Consider imbalance_pct when comparing loads in numa_has_capacity Message-ID: <20150623130114.GD32412@linux.vnet.ibm.com> Reply-To: Srikar Dronamraju References: <1434455762-30857-1-git-send-email-srikar@linux.vnet.ibm.com> <1434455762-30857-3-git-send-email-srikar@linux.vnet.ibm.com> <55803511.1060601@redhat.com> <20150622162958.GB32412@linux.vnet.ibm.com> <20150623081038.GA26231@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline In-Reply-To: <20150623081038.GA26231@gmail.com> User-Agent: Mutt/1.5.23 (2014-03-12) X-TM-AS-MML: disable X-Content-Scanned: Fidelis XPS MAILER x-cbid: 15062313-0009-0000-0000-000001974B73 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org * Ingo Molnar [2015-06-23 10:10:39]: > > Please let me know if there are any better ways to observe the > > spread. [...] > > There are. I see you are using prehistoric tooling, but see the various NUMA > convergence latency measurement utilities in 'perf bench numa': > > vega:~> cat numa01-THREAD_ALLOC > > perf bench numa mem --no-data_rand_walk -p 2 -t 16 -G 0 -P 0 -T 192 -l 1000 -zZ0c $@ > > You can generate very flexible setups of NUMA access patterns, and measure their > behavior accurately. > > It's all so much more capable and more flexible than autonumabench ... Okay, thanks for the hint, I will try this out in future. > > Also, when you are trying to report numbers for multiple runs, please use > something like: > > perf stat --null --repeat 3 ... > > This will run the workload 3 times (doing only time measurement) and report the > stddev in a human readable form. > Thanks again for this hint. Wouldnt system time/ user time also matter? I guess once Mel did point out that it was important to make sure that system time and user time dont increase when elapsed time decreases. But I cant find the email though. -- Thanks and Regards Srikar Dronamraju