From: Dave Chinner <david@fromorbit.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Mel Gorman <mel@csn.ul.ie>,
Linux Kernel List <linux-kernel@vger.kernel.org>,
linux-mm@kvack.org, Rik van Riel <riel@redhat.com>,
Johannes Weiner <hannes@cmpxchg.org>,
Minchan Kim <minchan.kim@gmail.com>,
Christoph Lameter <cl@linux-foundation.org>,
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
Wu Fengguang <fengguang.wu@intel.com>,
David Rientjes <rientjes@google.com>
Subject: Re: [PATCH 3/3] mm: page allocator: Drain per-cpu lists after direct reclaim allocation fails
Date: Sat, 4 Sep 2010 17:58:40 +1000 [thread overview]
Message-ID: <20100904075840.GE705@dastard> (raw)
In-Reply-To: <20100903202101.f937b0bb.akpm@linux-foundation.org>
On Fri, Sep 03, 2010 at 08:21:01PM -0700, Andrew Morton wrote:
> On Sat, 4 Sep 2010 12:25:45 +1000 Dave Chinner <david@fromorbit.com> wrote:
>
> > Still, given the improvements in performance from this patchset,
> > I'd say inclusion is a no-braniner....
>
> OK, thanks.
>
> It'd be interesting to check the IPI frequency with and without -
> /proc/interrupts "CAL" field. Presumably it went down a lot.
Maybe I suspected you would ask for this. I happened to dump
/proc/interrupts after the livelock run finished, so you're in
luck :)
The lines below are:
before: before running the single 50M inode create workload
after: the numbers after the run completes
livelock: the numbers after two runs with a livelock in the second
Vanilla 2.6.36-rc3:
before: 561 350 614 282 559 335 365 363
after: 10472 10473 10544 10681 9818 10837 10187 9923
.36-rc3 With patchset:
before: 452 426 441 337 748 321 498 357
after: 9463 9112 8671 8830 9391 8684 9768 8971
The numbers aren't that different - roughly 10% lower on average
with the patchset. I will state that vanilla kernel runs I ijust did
had noticably more consistent performance than the previous results
I had acheived, so perhaps it wasn't triggering the livelock
conditions as effectively this time through.
And finally:
livelock: 59458 58367 58559 59493 59614 57970 59060 58207
So the livelock case tends to indicate roughly 40,000 more IPI
interrupts per CPU occurred. The livelock occurred for close to 5
minutes, so that's roughly 130 IPIs per second per CPU....
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
WARNING: multiple messages have this Message-ID (diff)
From: Dave Chinner <david@fromorbit.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Mel Gorman <mel@csn.ul.ie>,
Linux Kernel List <linux-kernel@vger.kernel.org>,
linux-mm@kvack.org, Rik van Riel <riel@redhat.com>,
Johannes Weiner <hannes@cmpxchg.org>,
Minchan Kim <minchan.kim@gmail.com>,
Christoph Lameter <cl@linux-foundation.org>,
KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>,
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>,
Wu Fengguang <fengguang.wu@intel.com>,
David Rientjes <rientjes@google.com>
Subject: Re: [PATCH 3/3] mm: page allocator: Drain per-cpu lists after direct reclaim allocation fails
Date: Sat, 4 Sep 2010 17:58:40 +1000 [thread overview]
Message-ID: <20100904075840.GE705@dastard> (raw)
In-Reply-To: <20100903202101.f937b0bb.akpm@linux-foundation.org>
On Fri, Sep 03, 2010 at 08:21:01PM -0700, Andrew Morton wrote:
> On Sat, 4 Sep 2010 12:25:45 +1000 Dave Chinner <david@fromorbit.com> wrote:
>
> > Still, given the improvements in performance from this patchset,
> > I'd say inclusion is a no-braniner....
>
> OK, thanks.
>
> It'd be interesting to check the IPI frequency with and without -
> /proc/interrupts "CAL" field. Presumably it went down a lot.
Maybe I suspected you would ask for this. I happened to dump
/proc/interrupts after the livelock run finished, so you're in
luck :)
The lines below are:
before: before running the single 50M inode create workload
after: the numbers after the run completes
livelock: the numbers after two runs with a livelock in the second
Vanilla 2.6.36-rc3:
before: 561 350 614 282 559 335 365 363
after: 10472 10473 10544 10681 9818 10837 10187 9923
.36-rc3 With patchset:
before: 452 426 441 337 748 321 498 357
after: 9463 9112 8671 8830 9391 8684 9768 8971
The numbers aren't that different - roughly 10% lower on average
with the patchset. I will state that vanilla kernel runs I ijust did
had noticably more consistent performance than the previous results
I had acheived, so perhaps it wasn't triggering the livelock
conditions as effectively this time through.
And finally:
livelock: 59458 58367 58559 59493 59614 57970 59060 58207
So the livelock case tends to indicate roughly 40,000 more IPI
interrupts per CPU occurred. The livelock occurred for close to 5
minutes, so that's roughly 130 IPIs per second per CPU....
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2010-09-04 8:00 UTC|newest]
Thread overview: 104+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-09-03 9:08 [PATCH 0/3] Reduce watermark-related problems with the per-cpu allocator V4 Mel Gorman
2010-09-03 9:08 ` Mel Gorman
2010-09-03 9:08 ` [PATCH 1/3] mm: page allocator: Update free page counters after pages are placed on the free list Mel Gorman
2010-09-03 9:08 ` Mel Gorman
2010-09-03 22:38 ` Andrew Morton
2010-09-03 22:38 ` Andrew Morton
2010-09-05 18:06 ` Mel Gorman
2010-09-05 18:06 ` Mel Gorman
2010-09-03 9:08 ` [PATCH 2/3] mm: page allocator: Calculate a better estimate of NR_FREE_PAGES when memory is low and kswapd is awake Mel Gorman
2010-09-03 9:08 ` Mel Gorman
2010-09-03 22:55 ` Andrew Morton
2010-09-03 22:55 ` Andrew Morton
2010-09-03 23:17 ` Christoph Lameter
2010-09-03 23:17 ` Christoph Lameter
2010-09-03 23:28 ` Andrew Morton
2010-09-03 23:28 ` Andrew Morton
2010-09-04 0:54 ` Christoph Lameter
2010-09-04 0:54 ` Christoph Lameter
2010-09-05 18:12 ` Mel Gorman
2010-09-05 18:12 ` Mel Gorman
2010-09-03 9:08 ` [PATCH 3/3] mm: page allocator: Drain per-cpu lists after direct reclaim allocation fails Mel Gorman
2010-09-03 9:08 ` Mel Gorman
2010-09-03 23:00 ` Andrew Morton
2010-09-03 23:00 ` Andrew Morton
2010-09-04 2:25 ` Dave Chinner
2010-09-04 2:25 ` Dave Chinner
2010-09-04 3:21 ` Andrew Morton
2010-09-04 3:21 ` Andrew Morton
2010-09-04 7:58 ` Dave Chinner [this message]
2010-09-04 7:58 ` Dave Chinner
2010-09-04 8:14 ` Dave Chinner
2010-09-04 8:14 ` Dave Chinner
[not found] ` <20100905015400.GA10714@localhost>
[not found] ` <20100905021555.GG705@dastard>
[not found] ` <20100905060539.GA17450@localhost>
[not found] ` <20100905131447.GJ705@dastard>
2010-09-05 13:45 ` Wu Fengguang
2010-09-05 13:45 ` Wu Fengguang
2010-09-05 23:33 ` Dave Chinner
2010-09-05 23:33 ` Dave Chinner
2010-09-06 4:02 ` Dave Chinner
2010-09-06 4:02 ` Dave Chinner
2010-09-06 8:40 ` Mel Gorman
2010-09-06 8:40 ` Mel Gorman
2010-09-06 21:50 ` Dave Chinner
2010-09-06 21:50 ` Dave Chinner
2010-09-08 8:49 ` Dave Chinner
2010-09-08 8:49 ` Dave Chinner
2010-09-09 12:39 ` Mel Gorman
2010-09-09 12:39 ` Mel Gorman
2010-09-10 6:17 ` Dave Chinner
2010-09-10 6:17 ` Dave Chinner
2010-09-07 14:23 ` Christoph Lameter
2010-09-07 14:23 ` Christoph Lameter
2010-09-08 2:13 ` Wu Fengguang
2010-09-08 2:13 ` Wu Fengguang
2010-09-04 3:23 ` Wu Fengguang
2010-09-04 3:23 ` Wu Fengguang
2010-09-04 3:59 ` Andrew Morton
2010-09-04 3:59 ` Andrew Morton
2010-09-04 4:37 ` Wu Fengguang
2010-09-04 4:37 ` Wu Fengguang
2010-09-05 18:22 ` Mel Gorman
2010-09-05 18:22 ` Mel Gorman
2010-09-05 18:14 ` Mel Gorman
2010-09-05 18:14 ` Mel Gorman
2010-09-08 7:43 ` KOSAKI Motohiro
2010-09-08 7:43 ` KOSAKI Motohiro
2010-09-08 20:05 ` Christoph Lameter
2010-09-08 20:05 ` Christoph Lameter
2010-09-09 12:41 ` Mel Gorman
2010-09-09 12:41 ` Mel Gorman
2010-09-09 13:45 ` Christoph Lameter
2010-09-09 13:45 ` Christoph Lameter
2010-09-09 13:55 ` Mel Gorman
2010-09-09 13:55 ` Mel Gorman
2010-09-09 14:32 ` Christoph Lameter
2010-09-09 14:32 ` Christoph Lameter
2010-09-09 15:05 ` Mel Gorman
2010-09-09 15:05 ` Mel Gorman
2010-09-10 2:56 ` KOSAKI Motohiro
2010-09-10 2:56 ` KOSAKI Motohiro
2010-09-03 23:05 ` [PATCH 0/3] Reduce watermark-related problems with the per-cpu allocator V4 Andrew Morton
2010-09-03 23:05 ` Andrew Morton
2010-09-21 11:17 ` Mel Gorman
2010-09-21 11:17 ` Mel Gorman
2010-09-21 12:58 ` [stable] " Greg KH
2010-09-21 12:58 ` Greg KH
2010-09-21 14:23 ` Mel Gorman
2010-09-21 14:23 ` Mel Gorman
2010-09-23 18:49 ` Greg KH
2010-09-23 18:49 ` Greg KH
2010-09-24 9:14 ` Mel Gorman
2010-09-24 9:14 ` Mel Gorman
-- strict thread matches above, loose matches on Subject: below --
2010-08-31 17:37 [PATCH 0/3] Reduce watermark-related problems with the per-cpu allocator V3 Mel Gorman
2010-08-31 17:37 ` [PATCH 3/3] mm: page allocator: Drain per-cpu lists after direct reclaim allocation fails Mel Gorman
2010-08-31 17:37 ` Mel Gorman
2010-08-31 18:26 ` Christoph Lameter
2010-08-31 18:26 ` Christoph Lameter
2010-08-23 8:00 [PATCH 0/3] Reduce watermark-related problems with the per-cpu allocator V2 Mel Gorman
2010-08-23 8:00 ` [PATCH 3/3] mm: page allocator: Drain per-cpu lists after direct reclaim allocation fails Mel Gorman
2010-08-23 8:00 ` Mel Gorman
2010-08-23 23:17 ` KOSAKI Motohiro
2010-08-23 23:17 ` KOSAKI Motohiro
2010-08-16 9:42 [RFC PATCH 0/3] Reduce watermark-related problems with the per-cpu allocator Mel Gorman
2010-08-16 9:42 ` [PATCH 3/3] mm: page allocator: Drain per-cpu lists after direct reclaim allocation fails Mel Gorman
2010-08-16 14:50 ` Rik van Riel
2010-08-17 2:57 ` Minchan Kim
2010-08-18 3:02 ` KAMEZAWA Hiroyuki
2010-08-19 14:47 ` Minchan Kim
2010-08-19 15:10 ` Mel Gorman
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20100904075840.GE705@dastard \
--to=david@fromorbit.com \
--cc=akpm@linux-foundation.org \
--cc=cl@linux-foundation.org \
--cc=fengguang.wu@intel.com \
--cc=hannes@cmpxchg.org \
--cc=kamezawa.hiroyu@jp.fujitsu.com \
--cc=kosaki.motohiro@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mel@csn.ul.ie \
--cc=minchan.kim@gmail.com \
--cc=riel@redhat.com \
--cc=rientjes@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.