From: SJ Park <sj@kernel.org>
To: Anton Gavriliuk <antosha20xx@gmail.com>
Cc: SJ Park <sj@kernel.org>, damon@lists.linux.dev
Subject: Re: Memory tiering with DAMON/DAMOS auto-tuning
Date: Tue, 29 Sep 2026 00:49:32 -0700 [thread overview]
Message-ID: <20260929074932.40176-1-sj@kernel.org> (raw)
In-Reply-To: <CAAiJnjrJxoKy3AdXHiCSEGZyDwSqDJa4riO5FZeQX9uZKfjgmg@mail.gmail.com>
On Mon, 28 Sep 2026 20:02:43 +0300 Anton Gavriliuk <antosha20xx@gmail.com> wrote:
> > Thank you for sharing your use case and question!
>
> Thank you for your quick response and your answer/comments.
>
> We already had memory tiering implemented at hardware level 5-7 years
> ago with Intel's Optane PEM when configured in Memory Mode.
> Now we do the same thing at Linux level :-)
Interesting!
> I'm new at DAMON/DAMOS, there are lots of new tunables to me.
Indeed there are. I'd recommend starting with existing DAMON-based memory
tiering solutions including SK Hynix HMSDK and my auto-tune based memory
tiering script [1] and ask questions to the authors.
>
> > The example memory tiering script [1] uses 4% as intervals goal. Is there a
> > reason to use 97% as the goal instead?
>
> Oohhh... I moved to 5%.
I understand you mean 4%?
>
> > Is the above node_mum_used_bp what you really want? That means you want to
> > promote hot pages from node 2 to node 0, until the node 0 memory utilization
> > becomes 99.7%. That overlaps with the demotion goal (50% free memory of node
> > 0) quite a lot. I'd suggest smaller overlap, say, 50.3%, to keep healthy
> > circulation of hot/cold pages while not consuming too much resource under
> > stabilized access pattern.
>
> > DAMOS quota auto-tuning supports two tuner algorithms [2], consistent and
> > temporal. As the documentation [2] explains, 'consistent' tuner assumes it
> > should keep applying the action in a level to keep the goal achieved. In this
> > case, for example, the demotion scheme assumes there will be continued
> > promotion and therefore it keeps demoting. If there is no appropriate
> > promotion, it could result in demoting more than expected amount.
>
> So let me firstly understand what I really want to test :-)
Sure, and please feel free to ask any question in any ways. If you prefer to,
you could also ask questions privately to me or use DAMON Beer/Coffee/Tea Chat
series [2].
[1] https://github.com/damonitor/damo/blob/next/scripts/mem_tier.sh
[2] https://docs.google.com/document/d/1v43Kcj3ly4CYqmAkMaZzLiM2GEnWfgdGbZAH3mi2vpM/edit?usp=sharing
Thanks,
SJ
[...]
next prev parent reply other threads:[~2026-09-29 7:49 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-27 17:26 Memory tiering with DAMON/DAMOS auto-tuning Anton Gavriliuk
2026-09-28 8:15 ` SJ Park
2026-09-28 17:02 ` Anton Gavriliuk
2026-09-29 7:49 ` SJ Park [this message]
2026-09-29 8:41 ` Anton Gavriliuk
2026-09-29 9:25 ` SJ Park
2026-09-29 13:54 ` Anton Gavriliuk
2026-09-29 17:37 ` SJ Park
2026-09-30 4:03 ` Anton Gavriliuk
2026-09-30 8:35 ` SJ Park
2026-09-30 16:26 ` Anton Gavriliuk
2026-09-30 17:43 ` SJ Park
2026-10-01 9:58 ` Anton Gavriliuk
2026-10-01 10:26 ` SJ Park
2026-10-01 12:08 ` Anton Gavriliuk
2026-10-01 13:27 ` SJ Park
2026-10-01 16:54 ` Anton Gavriliuk
2026-10-02 8:34 ` SJ Park
2026-10-02 10:02 ` Anton Gavriliuk
2026-10-02 11:02 ` SJ Park
2026-10-02 13:06 ` Anton Gavriliuk
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=20260929074932.40176-1-sj@kernel.org \
--to=sj@kernel.org \
--cc=antosha20xx@gmail.com \
--cc=damon@lists.linux.dev \
/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