From: Bryan Rittmeyer <bryan@staidm.org>
To: Matt Mackall <mpm@selenic.com>
Cc: Andrew Morton <akpm@osdl.org>,
arjanv@redhat.com, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] shrink core hashes on small systems
Date: Mon, 5 Apr 2004 23:31:31 -0700 [thread overview]
Message-ID: <20040406063131.GA5186@staidm.org> (raw)
In-Reply-To: <20040405225911.GJ6248@waste.org>
On Mon, Apr 05, 2004 at 05:59:11PM -0500, Matt Mackall wrote:
> On the large end, we obviously have diminishing returns for larger
> hashes and lots of dirty cachelines for hash misses. We almost
> certainly want sublinear growth here, but sqrt might be too
> aggressive.
Hand wavy.
Memory size is not necessarily predictive of optimal hash size;
certain embedded workloads may want huge TCP hashes but
render farms may only need a few dozen dentries. Why not
just start small and rehash when chains get too long (or too short)?
That gives better cache behavior _and_ memory usage at the
expensive of increased latency during rehash. Maybe that's OK?
> For 2.7, I've been thinking about pulling out a generic lookup API,
> which would allow replacing hashing with something like rbtree, etc.,
> depending on space and performance criterion.
rbtrees have different performance characteristics than a hash, and
hashing seems pretty optimal in the places it's currently used.
But, I'd love to be wrong if it means a faster kernel.
-Bryan
next prev parent reply other threads:[~2004-04-06 23:04 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-04-05 20:49 [PATCH] shrink core hashes on small systems Matt Mackall
2004-04-05 21:02 ` Andrew Morton
2004-04-05 21:19 ` Matt Mackall
2004-04-05 21:38 ` Andrew Morton
2004-04-05 22:59 ` Matt Mackall
2004-04-06 6:31 ` Bryan Rittmeyer [this message]
2004-04-07 5:21 ` Marcelo Tosatti
2004-04-07 18:10 ` Matt Mackall
2004-04-07 22:53 ` Benjamin Herrenschmidt
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=20040406063131.GA5186@staidm.org \
--to=bryan@staidm.org \
--cc=akpm@osdl.org \
--cc=arjanv@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mpm@selenic.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 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).