From: Balbir Singh <balbir@linux.vnet.ibm.com>
To: Anton Blanchard <anton@samba.org>
Cc: mel@csn.ul.ie, cl@linux-foundation.org, linuxppc-dev@lists.ozlabs.org
Subject: Re: [PATCH] powerpc: Set a smaller value for RECLAIM_DISTANCE to enable zone reclaim
Date: Fri, 19 Feb 2010 21:13:16 +0530 [thread overview]
Message-ID: <661de9471002190743t5385d97hf38c544f0007879c@mail.gmail.com> (raw)
In-Reply-To: <20100218222923.GC31681@kryten>
On Fri, Feb 19, 2010 at 3:59 AM, Anton Blanchard <anton@samba.org> wrote:
>
> I noticed /proc/sys/vm/zone_reclaim_mode was 0 on a ppc64 NUMA box. It ge=
ts
> enabled via this:
>
> =A0 =A0 =A0 =A0/*
> =A0 =A0 =A0 =A0 * If another node is sufficiently far away then it is bet=
ter
> =A0 =A0 =A0 =A0 * to reclaim pages in a zone before going off node.
> =A0 =A0 =A0 =A0 */
> =A0 =A0 =A0 =A0if (distance > RECLAIM_DISTANCE)
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0zone_reclaim_mode =3D 1;
>
> Since we use the default value of 20 for REMOTE_DISTANCE and 20 for
> RECLAIM_DISTANCE it never kicks in.
>
> The local to remote bandwidth ratios can be quite large on System p
> machines so it makes sense for us to reclaim clean pagecache locally befo=
re
> going off node.
>
> The patch below sets a smaller value for RECLAIM_DISTANCE and thus enable=
s
> zone reclaim.
>
A reclaim distance of 10 implies a ratio of 1, that means we'll always
do zone_reclaim() to free page cache and slab cache before moving on
to another node?
Balbir Singh.
next prev parent reply other threads:[~2010-02-19 15:43 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-02-18 22:29 [PATCH] powerpc: Set a smaller value for RECLAIM_DISTANCE to enable zone reclaim Anton Blanchard
2010-02-19 0:07 ` Anton Blanchard
2010-02-19 14:55 ` Mel Gorman
2010-02-19 15:12 ` Christoph Lameter
2010-02-19 15:41 ` Balbir Singh
2010-02-19 15:51 ` Christoph Lameter
2010-02-19 17:39 ` Balbir Singh
2010-02-23 1:55 ` Anton Blanchard
2010-02-23 16:23 ` Mel Gorman
2010-02-24 15:43 ` Christoph Lameter
2010-03-01 12:06 ` Mel Gorman
2010-03-01 15:19 ` Christoph Lameter
2010-02-19 15:43 ` Balbir Singh [this message]
2010-02-23 1:38 ` Anton Blanchard
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=661de9471002190743t5385d97hf38c544f0007879c@mail.gmail.com \
--to=balbir@linux.vnet.ibm.com \
--cc=anton@samba.org \
--cc=cl@linux-foundation.org \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=mel@csn.ul.ie \
/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).