Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On 8 December 2016 at 15:09, Matt Fleming wrote: > On Mon, 05 Dec, at 01:35:46PM, Matt Fleming wrote: >> On Mon, 05 Dec, at 10:27:36AM, Vincent Guittot wrote: >> > >> > Hi Matt, >> > >> > Thanks for the results. >> > >> > During the review, it has been pointed out by

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On 8 December 2016 at 15:09, Matt Fleming wrote: > On Mon, 05 Dec, at 01:35:46PM, Matt Fleming wrote: >> On Mon, 05 Dec, at 10:27:36AM, Vincent Guittot wrote: >> > >> > Hi Matt, >> > >> > Thanks for the results. >> > >> > During the review, it has been pointed out by Morten that the test >> >

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Mon, 05 Dec, at 01:35:46PM, Matt Fleming wrote: > On Mon, 05 Dec, at 10:27:36AM, Vincent Guittot wrote: > > > > Hi Matt, > > > > Thanks for the results. > > > > During the review, it has been pointed out by Morten that the test condition > > (100*this_avg_load < imbalance_scale*min_avg_load)

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Mon, 05 Dec, at 01:35:46PM, Matt Fleming wrote: > On Mon, 05 Dec, at 10:27:36AM, Vincent Guittot wrote: > > > > Hi Matt, > > > > Thanks for the results. > > > > During the review, it has been pointed out by Morten that the test condition > > (100*this_avg_load < imbalance_scale*min_avg_load)

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Mon, 05 Dec, at 10:27:36AM, Vincent Guittot wrote: > > Hi Matt, > > Thanks for the results. > > During the review, it has been pointed out by Morten that the test condition > (100*this_avg_load < imbalance_scale*min_avg_load) makes more sense than > (100*min_avg_load >

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Mon, 05 Dec, at 10:27:36AM, Vincent Guittot wrote: > > Hi Matt, > > Thanks for the results. > > During the review, it has been pointed out by Morten that the test condition > (100*this_avg_load < imbalance_scale*min_avg_load) makes more sense than > (100*min_avg_load >

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

Le Saturday 03 Dec 2016 à 21:47:07 (+), Matt Fleming a écrit : > On Fri, 02 Dec, at 07:31:04PM, Brendan Gregg wrote: > > > > For background, is this from the "A decade of wasted cores" paper's > > patches? > > No, this patch fixes an issue I originally reported here, > >

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

Le Saturday 03 Dec 2016 à 21:47:07 (+), Matt Fleming a écrit : > On Fri, 02 Dec, at 07:31:04PM, Brendan Gregg wrote: > > > > For background, is this from the "A decade of wasted cores" paper's > > patches? > > No, this patch fixes an issue I originally reported here, > >

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Fri, 02 Dec, at 07:31:04PM, Brendan Gregg wrote: > > For background, is this from the "A decade of wasted cores" paper's > patches? No, this patch fixes an issue I originally reported here, https://lkml.kernel.org/r/20160923115808.2330-1-m...@codeblueprint.co.uk Essentially, if you have

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Fri, 02 Dec, at 07:31:04PM, Brendan Gregg wrote: > > For background, is this from the "A decade of wasted cores" paper's > patches? No, this patch fixes an issue I originally reported here, https://lkml.kernel.org/r/20160923115808.2330-1-m...@codeblueprint.co.uk Essentially, if you have

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Fri, Nov 25, 2016 at 7:34 AM, Vincent Guittot wrote: > > find_idlest_group() only compares the runnable_load_avg when looking for > the least loaded group. But on fork intensive use case like hackbench > where tasks blocked quickly after the fork, this can lead to

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Fri, Nov 25, 2016 at 7:34 AM, Vincent Guittot wrote: > > find_idlest_group() only compares the runnable_load_avg when looking for > the least loaded group. But on fork intensive use case like hackbench > where tasks blocked quickly after the fork, this can lead to selecting the > same CPU

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Fri, 02 Dec, at 04:20:54PM, Vincent Guittot wrote: > > Matt, > > Have you been able to get some results for the patchset ? Sorry, I messed up the test config and the tests never ran :( I've redone everything and fast-tracked the results on the SUSE grid now. Should have results by morning.

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Fri, 02 Dec, at 04:20:54PM, Vincent Guittot wrote: > > Matt, > > Have you been able to get some results for the patchset ? Sorry, I messed up the test config and the tests never ran :( I've redone everything and fast-tracked the results on the SUSE grid now. Should have results by morning.

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On 30 November 2016 at 15:24, Morten Rasmussen wrote: > On Wed, Nov 30, 2016 at 02:54:00PM +0100, Vincent Guittot wrote: >> On 30 November 2016 at 14:49, Vincent Guittot >> wrote: >> > On 30 November 2016 at 13:49, Morten Rasmussen

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On 30 November 2016 at 15:24, Morten Rasmussen wrote: > On Wed, Nov 30, 2016 at 02:54:00PM +0100, Vincent Guittot wrote: >> On 30 November 2016 at 14:49, Vincent Guittot >> wrote: >> > On 30 November 2016 at 13:49, Morten Rasmussen >> > wrote: >> >> On Fri, Nov 25, 2016 at 04:34:33PM +0100,

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Wed, Nov 30, 2016 at 02:54:00PM +0100, Vincent Guittot wrote: > On 30 November 2016 at 14:49, Vincent Guittot > wrote: > > On 30 November 2016 at 13:49, Morten Rasmussen > > wrote: > >> On Fri, Nov 25, 2016 at 04:34:33PM +0100, Vincent

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Wed, Nov 30, 2016 at 02:54:00PM +0100, Vincent Guittot wrote: > On 30 November 2016 at 14:49, Vincent Guittot > wrote: > > On 30 November 2016 at 13:49, Morten Rasmussen > > wrote: > >> On Fri, Nov 25, 2016 at 04:34:33PM +0100, Vincent Guittot wrote: > >>> find_idlest_group() only compares

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Wed, Nov 30, 2016 at 02:49:11PM +0100, Vincent Guittot wrote: > On 30 November 2016 at 13:49, Morten Rasmussen > wrote: > > On Fri, Nov 25, 2016 at 04:34:33PM +0100, Vincent Guittot wrote: > >> find_idlest_group() only compares the runnable_load_avg when looking for

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Wed, Nov 30, 2016 at 02:49:11PM +0100, Vincent Guittot wrote: > On 30 November 2016 at 13:49, Morten Rasmussen > wrote: > > On Fri, Nov 25, 2016 at 04:34:33PM +0100, Vincent Guittot wrote: > >> find_idlest_group() only compares the runnable_load_avg when looking for > >> the least loaded

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On 30 November 2016 at 14:49, Vincent Guittot wrote: > On 30 November 2016 at 13:49, Morten Rasmussen > wrote: >> On Fri, Nov 25, 2016 at 04:34:33PM +0100, Vincent Guittot wrote: >>> find_idlest_group() only compares the runnable_load_avg

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On 30 November 2016 at 14:49, Vincent Guittot wrote: > On 30 November 2016 at 13:49, Morten Rasmussen > wrote: >> On Fri, Nov 25, 2016 at 04:34:33PM +0100, Vincent Guittot wrote: >>> find_idlest_group() only compares the runnable_load_avg when looking for >>> the least loaded group. But on fork

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On 30 November 2016 at 13:49, Morten Rasmussen wrote: > On Fri, Nov 25, 2016 at 04:34:33PM +0100, Vincent Guittot wrote: >> find_idlest_group() only compares the runnable_load_avg when looking for >> the least loaded group. But on fork intensive use case like hackbench

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On 30 November 2016 at 13:49, Morten Rasmussen wrote: > On Fri, Nov 25, 2016 at 04:34:33PM +0100, Vincent Guittot wrote: >> find_idlest_group() only compares the runnable_load_avg when looking for >> the least loaded group. But on fork intensive use case like hackbench [snip] >> +

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Fri, Nov 25, 2016 at 04:34:33PM +0100, Vincent Guittot wrote: > find_idlest_group() only compares the runnable_load_avg when looking for > the least loaded group. But on fork intensive use case like hackbench > where tasks blocked quickly after the fork, this can lead to selecting the > same

Re: [PATCH 2/2 v2] sched: use load_avg for selecting idlest group

On Fri, Nov 25, 2016 at 04:34:33PM +0100, Vincent Guittot wrote: > find_idlest_group() only compares the runnable_load_avg when looking for > the least loaded group. But on fork intensive use case like hackbench > where tasks blocked quickly after the fork, this can lead to selecting the > same

[PATCH 2/2 v2] sched: use load_avg for selecting idlest group

find_idlest_group() only compares the runnable_load_avg when looking for the least loaded group. But on fork intensive use case like hackbench where tasks blocked quickly after the fork, this can lead to selecting the same CPU instead of other CPUs, which have similar runnable load but a lower

[PATCH 2/2 v2] sched: use load_avg for selecting idlest group

find_idlest_group() only compares the runnable_load_avg when looking for the least loaded group. But on fork intensive use case like hackbench where tasks blocked quickly after the fork, this can lead to selecting the same CPU instead of other CPUs, which have similar runnable load but a lower