linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Mel Gorman <mgorman@suse.de>
To: Linux Kernel <linux-kernel@vger.kernel.org>,
	Linux-MM <linux-mm@kvack.org>,
	Linux-FSDevel <linux-fsdevel@vger.kernel.org>
Cc: Johannes Weiner <hannes@cmpxchg.org>,
	Jens Axboe <axboe@kernel.dk>, Jeff Moyer <jmoyer@redhat.com>,
	Dave Chinner <david@fromorbit.com>, Mel Gorman <mgorman@suse.de>
Subject: [PATCH 0/6] Improve sequential read throughput v2
Date: Wed, 25 Jun 2014 08:58:43 +0100	[thread overview]
Message-ID: <1403683129-10814-1-git-send-email-mgorman@suse.de> (raw)

Changelog since v1
o Rebase to v3.16-rc2
o Move CFQ patch to end of series where it can be rejected easier if necessary
o Introduce page-reclaim related patch related to kswapd/fairzone interactions
o Rework fast zone policy patch

IO performance since 3.0 has been a mixed bag. In many respects we are
better and in some we are worse and one of those places is sequential
read throughput. This is visible in a number of benchmarks but I looked
at tiobench the closest. This is using ext3 on a mid-range desktop and
comparing against 3.0.

                                      3.16.0-rc2            3.16.0-rc2                 3.0.0
                                         vanilla                cfq600               vanilla
Min    SeqRead-MB/sec-1         120.96 (  0.00%)      140.43 ( 16.10%)      134.04 ( 10.81%)
Min    SeqRead-MB/sec-2         100.73 (  0.00%)      118.18 ( 17.32%)      120.76 ( 19.88%)
Min    SeqRead-MB/sec-4          96.05 (  0.00%)      110.84 ( 15.40%)      114.49 ( 19.20%)
Min    SeqRead-MB/sec-8          82.46 (  0.00%)       92.40 ( 12.05%)       98.04 ( 18.89%)
Min    SeqRead-MB/sec-16         66.37 (  0.00%)       76.68 ( 15.53%)       79.49 ( 19.77%)

This series does not fully restore throughput performance to 3.0 levels
but it brings it acceptably close. While throughput for higher numbers
of threads is lower, it is known that it can be tuned by increasing
target_latency or disabling low_latency giving higher overall throughput
at the cost of latency and IO fairness.

This series in ordered in ascending-likelihood-to-cause-controversary so
that a partial series can still potentially be merged even if parts of it
are naked (e.g. CGQ). For reference, here is the series without the CFQ
patch at the end.

                                      3.16.0-rc2            3.16.0-rc2                 3.0.0
                                         vanilla             lessdirty               vanilla
Min    SeqRead-MB/sec-1         120.96 (  0.00%)      141.04 ( 16.60%)      134.04 ( 10.81%)
Min    SeqRead-MB/sec-2         100.73 (  0.00%)      116.26 ( 15.42%)      120.76 ( 19.88%)
Min    SeqRead-MB/sec-4          96.05 (  0.00%)      109.52 ( 14.02%)      114.49 ( 19.20%)
Min    SeqRead-MB/sec-8          82.46 (  0.00%)       88.60 (  7.45%)       98.04 ( 18.89%)
Min    SeqRead-MB/sec-16         66.37 (  0.00%)       69.87 (  5.27%)       79.49 ( 19.77%)


 block/cfq-iosched.c            |   2 +-
 include/linux/mmzone.h         | 210 ++++++++++++++++++++++-------------------
 include/linux/writeback.h      |   1 +
 include/trace/events/pagemap.h |  16 ++--
 mm/internal.h                  |   1 +
 mm/mm_init.c                   |   5 +-
 mm/page-writeback.c            |  15 +--
 mm/page_alloc.c                | 206 ++++++++++++++++++++++++++--------------
 mm/swap.c                      |   4 +-
 mm/vmscan.c                    |  16 ++--
 mm/vmstat.c                    |   4 +-
 11 files changed, 285 insertions(+), 195 deletions(-)

-- 
1.8.4.5

--
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>

             reply	other threads:[~2014-06-25  7:58 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-06-25  7:58 Mel Gorman [this message]
2014-06-25  7:58 ` [PATCH 1/6] mm: pagemap: Avoid unnecessary overhead when tracepoints are deactivated Mel Gorman
2014-06-25  7:58 ` [PATCH 2/6] mm: Rearrange zone fields into read-only, page alloc, statistics and page reclaim lines Mel Gorman
2014-06-25  7:58 ` [PATCH 3/6] mm: vmscan: Do not reclaim from lower zones if they are balanced Mel Gorman
2014-06-25 23:32   ` Andrew Morton
2014-06-26 10:17     ` Mel Gorman
2014-06-25  7:58 ` [PATCH 4/6] mm: page_alloc: Reduce cost of the fair zone allocation policy Mel Gorman
2014-06-25  7:58 ` [PATCH 5/6] mm: page_alloc: Reduce cost of dirty zone balancing Mel Gorman
2014-06-25 23:35   ` Andrew Morton
2014-06-26  8:43     ` Mel Gorman
2014-06-26 14:37       ` Johannes Weiner
2014-06-26 14:56         ` Mel Gorman
2014-06-26 15:11           ` Johannes Weiner
2014-06-25  7:58 ` [PATCH 6/6] cfq: Increase default value of target_latency Mel Gorman
2014-06-26 15:36   ` Jeff Moyer
2014-06-26 16:19     ` Mel Gorman
2014-06-26 16:50       ` Jeff Moyer
2014-06-26 17:45         ` Mel Gorman
2014-06-26 18:04           ` Jeff Moyer

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=1403683129-10814-1-git-send-email-mgorman@suse.de \
    --to=mgorman@suse.de \
    --cc=axboe@kernel.dk \
    --cc=david@fromorbit.com \
    --cc=hannes@cmpxchg.org \
    --cc=jmoyer@redhat.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).