From: Matthew Dobson <colpatch@us.ibm.com>
To: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Paul Jackson <pj@sgi.com>, "Martin J. Bligh" <mbligh@aracnet.com>,
Andrew Morton <akpm@osdl.org>,
ckrm-tech@lists.sourceforge.net,
LSE Tech <lse-tech@lists.sourceforge.net>,
LKML <linux-kernel@vger.kernel.org>,
simon.derr@bull.net, frankeh@watson.ibm.com
Subject: Re: [RFC PATCH] scheduler: Dynamic sched_domains
Date: Thu, 07 Oct 2004 15:20:05 -0700 [thread overview]
Message-ID: <1097187605.17473.37.camel@arrakis> (raw)
In-Reply-To: <4164A664.9040005@yahoo.com.au>
On Wed, 2004-10-06 at 19:13, Nick Piggin wrote:
> This is what I did in my first (that nobody ever saw) implementation of
> sched domains. Ie. no sched_groups, just use sched_domains as the balancing
> object... I'm not sure this works too well.
>
> For example, your bottom level domain is going to basically be a redundant,
> single CPU on most topologies, isn't it?
I forgot to respond to this part in my last mail... :(
My benchmarks haven't shown any real deviation in performance from stock
-mm. Granted, my benchmarking has been very limited, pretty much just
running kernbench on a few different machines with different configs
(ie: SMT, SMP & NUMA on/off). A performance NOOP is exactly what I was
hoping for, though. I don't really expect these changes to be either a
performance win or loss, but a functionality improvement.
The patch is pretty dense because of all the renaming, but there are no
single CPU domains. The lowest level domain would be:
1) Node domains, for NUMA w/ SMP
2) Sibling domains, for SMT or NUMA w/ SMT
3) System domain, for flat SMP
The pseudo code version of my arch_init_sched_domains() looks like:
cpu_usable_map = cpu_online_map & ~cpu_isolated_map
create system domain;
if NUMA
for_each_node()
create node domain, parented to system domain;
for_each_cpu(cpu_usable_map)
if SCHED_SMT
create sibling domain, parented to node domain;
attach sibling cpu to it's domain;
else
attach cpu to either its node domain or system domain;
So there shouldn't be any redundant CPU domains.
-Matt
next prev parent reply other threads:[~2004-10-07 22:30 UTC|newest]
Thread overview: 63+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-10-07 0:51 [RFC PATCH] scheduler: Dynamic sched_domains Matthew Dobson
2004-10-07 2:13 ` Nick Piggin
2004-10-07 17:01 ` Jesse Barnes
2004-10-08 5:55 ` [Lse-tech] " Takayoshi Kochi
2004-10-08 6:08 ` Nick Piggin
2004-10-08 16:43 ` Jesse Barnes
2004-10-07 21:58 ` Matthew Dobson
2004-10-08 0:22 ` Nick Piggin
2004-10-07 22:20 ` Matthew Dobson [this message]
2004-10-07 4:12 ` [ckrm-tech] " Marc E. Fiuczynski
2004-10-07 5:35 ` Paul Jackson
2004-10-07 22:06 ` Matthew Dobson
2004-10-07 9:32 ` Paul Jackson
2004-10-08 10:14 ` [Lse-tech] " Erich Focht
2004-10-08 10:40 ` Nick Piggin
2004-10-08 15:50 ` [ckrm-tech] " Hubertus Franke
2004-10-08 22:48 ` Matthew Dobson
2004-10-08 18:54 ` Matthew Dobson
2004-10-08 21:56 ` Peter Williams
2004-10-08 22:52 ` Matthew Dobson
2004-10-08 23:13 ` Erich Focht
2004-10-08 23:50 ` Nick Piggin
2004-10-10 12:25 ` Erich Focht
2004-10-08 22:51 ` Erich Focht
2004-10-09 1:05 ` Matthew Dobson
2004-10-10 12:45 ` Erich Focht
2004-10-12 22:45 ` Matthew Dobson
2004-10-08 18:45 ` Matthew Dobson
2005-04-18 20:26 ` [RFC PATCH] Dynamic sched domains aka Isolated cpusets Dinakar Guniguntala
2005-04-18 23:44 ` Nick Piggin
2005-04-19 8:00 ` Dinakar Guniguntala
2005-04-19 5:54 ` Paul Jackson
2005-04-19 6:19 ` Nick Piggin
2005-04-19 6:59 ` Paul Jackson
2005-04-19 7:09 ` Nick Piggin
2005-04-19 7:25 ` Paul Jackson
2005-04-19 7:28 ` Paul Jackson
2005-04-19 7:19 ` Paul Jackson
2005-04-19 7:57 ` Nick Piggin
2005-04-19 20:34 ` Paul Jackson
2005-04-23 23:26 ` Paul Jackson
2005-04-26 0:52 ` Matthew Dobson
2005-04-26 0:59 ` Paul Jackson
2005-04-19 9:52 ` Dinakar Guniguntala
2005-04-19 15:26 ` Paul Jackson
2005-04-20 7:37 ` Dinakar Guniguntala
2005-04-19 20:42 ` Paul Jackson
2005-04-19 8:12 ` Simon Derr
2005-04-19 16:19 ` Paul Jackson
2005-04-19 9:34 ` [Lse-tech] " Dinakar Guniguntala
2005-04-19 17:23 ` Paul Jackson
2005-04-20 7:16 ` Dinakar Guniguntala
2005-04-20 19:09 ` Paul Jackson
2005-04-21 16:27 ` Dinakar Guniguntala
2005-04-22 21:26 ` Paul Jackson
2005-04-23 7:24 ` Dinakar Guniguntala
2005-04-23 22:30 ` Paul Jackson
2005-04-25 11:53 ` Dinakar Guniguntala
2005-04-25 14:38 ` Paul Jackson
2005-04-21 17:31 ` [RFC PATCH] Dynamic sched domains aka Isolated cpusets (v0.2) Dinakar Guniguntala
2005-04-22 18:50 ` Paul Jackson
2005-04-22 21:37 ` Paul Jackson
2005-04-23 3:11 ` Paul Jackson
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=1097187605.17473.37.camel@arrakis \
--to=colpatch@us.ibm.com \
--cc=akpm@osdl.org \
--cc=ckrm-tech@lists.sourceforge.net \
--cc=frankeh@watson.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=lse-tech@lists.sourceforge.net \
--cc=mbligh@aracnet.com \
--cc=nickpiggin@yahoo.com.au \
--cc=pj@sgi.com \
--cc=simon.derr@bull.net \
/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).