From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756992AbaIEMY0 (ORCPT ); Fri, 5 Sep 2014 08:24:26 -0400 Received: from mail-oi0-f43.google.com ([209.85.218.43]:59726 "EHLO mail-oi0-f43.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756778AbaIEMYZ (ORCPT ); Fri, 5 Sep 2014 08:24:25 -0400 MIME-Version: 1.0 In-Reply-To: <5409A73B.4050402@linux.vnet.ibm.com> References: <1409051215-16788-1-git-send-email-vincent.guittot@linaro.org> <1409051215-16788-9-git-send-email-vincent.guittot@linaro.org> <5409A73B.4050402@linux.vnet.ibm.com> From: Vincent Guittot Date: Fri, 5 Sep 2014 14:24:04 +0200 Message-ID: Subject: Re: [PATCH v5 08/12] sched: move cfs task on a CPU with higher capacity To: Preeti U Murthy Cc: Peter Zijlstra , Ingo Molnar , linux-kernel , Russell King - ARM Linux , LAK , Rik van Riel , Morten Rasmussen , Mike Galbraith , Nicolas Pitre , "linaro-kernel@lists.linaro.org" , Daniel Lezcano , Dietmar Eggemann Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 5 September 2014 14:06, Preeti U Murthy wrote: > Hi Vincent, > > On 08/26/2014 04:36 PM, Vincent Guittot wrote: >> If the CPU is used for handling lot of IRQs, trig a load balance to check if >> it's worth moving its tasks on another CPU that has more capacity. >> >> As a sidenote, this will note generate more spurious ilb because we already >> trig an ilb if there is more than 1 busy cpu. If this cpu is the only one that >> has a task, we will trig the ilb once for migrating the task. >> >> The nohz_kick_needed function has been cleaned up a bit while adding the new >> test >> >> Signed-off-by: Vincent Guittot > > So I see that there are added checks in your previous patches on if the > cpu capacity for CFS tasks is good enough to run tasks on the cpu. My > concern is although they appear sensible, would they trigger an increase > in the number of times we load balance to a large extent. > > Ebizzy would not test this aspect right? There are no real time > tasks/interrupts that get generated. yes, ebizzy doesn't test this part but check for non regression The scp test is the one that i use to check this patch and the previous one but a test with some cfs and rt tasks should also do the jobs. we can see an increase of 82% for the dual core when CONFIG_IRQ_TIME_ACCOUNTING is enable > > Besides, what is the column that says patchset+irq? What is the irq > accounting patchset that you refer to in your cover letter? it refers to CONFIG_IRQ_TIME_ACCOUNTING which includes the time spent under interrupt context to compute the scale_rt_capacity Regards, Vincent > > Regards > Preeti U Murthy > From mboxrd@z Thu Jan 1 00:00:00 1970 From: vincent.guittot@linaro.org (Vincent Guittot) Date: Fri, 5 Sep 2014 14:24:04 +0200 Subject: [PATCH v5 08/12] sched: move cfs task on a CPU with higher capacity In-Reply-To: <5409A73B.4050402@linux.vnet.ibm.com> References: <1409051215-16788-1-git-send-email-vincent.guittot@linaro.org> <1409051215-16788-9-git-send-email-vincent.guittot@linaro.org> <5409A73B.4050402@linux.vnet.ibm.com> Message-ID: To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On 5 September 2014 14:06, Preeti U Murthy wrote: > Hi Vincent, > > On 08/26/2014 04:36 PM, Vincent Guittot wrote: >> If the CPU is used for handling lot of IRQs, trig a load balance to check if >> it's worth moving its tasks on another CPU that has more capacity. >> >> As a sidenote, this will note generate more spurious ilb because we already >> trig an ilb if there is more than 1 busy cpu. If this cpu is the only one that >> has a task, we will trig the ilb once for migrating the task. >> >> The nohz_kick_needed function has been cleaned up a bit while adding the new >> test >> >> Signed-off-by: Vincent Guittot > > So I see that there are added checks in your previous patches on if the > cpu capacity for CFS tasks is good enough to run tasks on the cpu. My > concern is although they appear sensible, would they trigger an increase > in the number of times we load balance to a large extent. > > Ebizzy would not test this aspect right? There are no real time > tasks/interrupts that get generated. yes, ebizzy doesn't test this part but check for non regression The scp test is the one that i use to check this patch and the previous one but a test with some cfs and rt tasks should also do the jobs. we can see an increase of 82% for the dual core when CONFIG_IRQ_TIME_ACCOUNTING is enable > > Besides, what is the column that says patchset+irq? What is the irq > accounting patchset that you refer to in your cover letter? it refers to CONFIG_IRQ_TIME_ACCOUNTING which includes the time spent under interrupt context to compute the scale_rt_capacity Regards, Vincent > > Regards > Preeti U Murthy >