lkml.org 
[lkml]   [2012]   [Nov]   [21]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 00/46] Automatic NUMA Balancing V4
On Wed, Nov 21, 2012 at 07:21:58PM +0100, Ingo Molnar wrote:
>
> * Mel Gorman <mgorman@suse.de> wrote:
>
> > On Wed, Nov 21, 2012 at 06:33:16PM +0100, Ingo Molnar wrote:
> > >
> > > * Mel Gorman <mgorman@suse.de> wrote:
> > >
> > > > On Wed, Nov 21, 2012 at 06:03:06PM +0100, Ingo Molnar wrote:
> > > > >
> > > > > * Mel Gorman <mgorman@suse.de> wrote:
> > > > >
> > > > > > On Wed, Nov 21, 2012 at 10:21:06AM +0000, Mel Gorman wrote:
> > > > > > >
> > > > > > > I am not including a benchmark report in this but will be posting one
> > > > > > > shortly in the "Latest numa/core release, v16" thread along with the latest
> > > > > > > schednuma figures I have available.
> > > > > > >
> > > > > >
> > > > > > Report is linked here https://lkml.org/lkml/2012/11/21/202
> > > > > >
> > > > > > I ended up cancelling the remaining tests and restarted with
> > > > > >
> > > > > > 1. schednuma + patches posted since so that works out as
> > > > >
> > > > > Mel, I'd like to ask you to refer to our tree as numa/core or
> > > > > 'numacore' in the future. Would such a courtesy to use the
> > > > > current name of our tree be possible?
> > > > >
> > > >
> > > > Sure, no problem.
> > >
> > > Thanks!
> > >
> > > I ran a quick test with your 'balancenuma v4' tree and while
> > > numa02 and numa01-THREAD-ALLOC performance is looking good,
> > > numa01 performance does not look very good:
> > >
> > > mainline numa/core balancenuma-v4
> > > numa01: 340.3 139.4 276 secs
> > >
> > > 97% slower than numa/core.
> > >
> >
> > It would be. numa01 is an adverse workload where all threads
> > are hammering the same memory. The two-stage filter in
> > balancenuma restricts the amount of migration it does so it
> > ends up in a situation where it cannot balance properly. [...]
>
> Do you mean this "balancenuma v4" patch attributed to you:
>
> Subject: mm: Numa: Use a two-stage filter to restrict pages being migrated for unlikely task<->node relationships
> From: Mel Gorman <mgorman@suse.de>
> Date: Wed, 21 Nov 2012 10:21:42 +0000
>

Yes.

> ...
>
> Signed-off-by: Mel Gorman <mgorman@suse.de>
>
> which has:
>
> /*
> * Multi-stage node selection is used in conjunction
> * with a periodic migration fault to build a temporal
> * task<->page relation. By using a two-stage filter we
> * remove short/unlikely relations.
> *
> * Using P(p) ~ n_p / n_t as per frequentist
> * probability, we can equate a task's usage of a
> * particular page (n_p) per total usage of this
> * page (n_t) (in a given time-span) to a probability.
> *
> * Our periodic faults will sample this probability and
> * getting the same result twice in a row, given these
> * samples are fully independent, is then given by
> * P(n)^2, provided our sample period is sufficiently
> * short compared to the usage pattern.
> *
> * This quadric squishes small probabilities, making
> * it less likely we act on an unlikely task<->page
> * relation.
>
> This looks very similar to the code and text that Peter wrote
> for numa/core:
>
> /*
> * Multi-stage node selection is used in conjunction with a periodic
> * migration fault to build a temporal task<->page relation. By
> * using a two-stage filter we remove short/unlikely relations.
> *
> * Using P(p) ~ n_p / n_t as per frequentist probability, we can
> * equate a task's usage of a particular page (n_p) per total usage
> * of this page (n_t) (in a given time-span) to a probability.
> *
> * Our periodic faults will then sample this probability and getting
> * the same result twice in a row, given these samples are fully
> * independent, is then given by P(n)^2, provided our sample period
> * is sufficiently short compared to the usage pattern.
> *
> * This quadric squishes small probabilities, making it less likely
> * we act on an unlikely task<->page relation.
> *
> * Return the best node ID this page should be on, or -1 if it should
> * stay where it is.
> */
>
> see commit:
>
> 30f93abc6cb3 sched, numa, mm: Add the scanning page fault machinery
>
> ?
>
> I think it's the very same concept - yours is taken from an
> older sched/numa commit and attributed to yourself? [If so then
> please fix the attribution.]

Yes, it's completely based on earlier sched/numa patches. In many of the
patches you'll see notes where I documented what patches I originally
based on -- be it from sched/numa, autonuma or some combination of both.
In many cases I could not keep the signed-off-by because the end result
was simply too different to claim that the author was happy with it. I was
hoping that these notes would convert to signed-offs-by after review from
the original authors who were cc'd at all times.

> We have the same filter in numa/core - because we wrote it (FYI,
> I wrote bits of the last_cpu variant in numa/core), yet our
> numa01 performance is much better than the one of balancenuma.
>

Yes, the lack of a note was a mistake. I've added the following note to
the top of this patch now

Note: This two-stage filter was taken directly from the sched/numa patch
"sched, numa, mm: Add the scanning page fault machinery" but is
only a partial extraction. As the end result is not necessarily
recognisable, the signed-offs-by had to be removed. Will be
added back if requested.

Thanks and apologies in advance for any other patch where I failed to
document the history correctly.

--
Mel Gorman
SUSE Labs


\
 
 \ /
  Last update: 2012-11-21 20:41    [W:0.088 / U:0.500 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site