linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Pavel Machek <pavel@ucw.cz>
To: Arjan van de Ven <arjan@infradead.org>
Cc: linux-kernel@vger.kernel.org, akpm@linux-foundation.org,
	mingo@elte.hu, peterz@infradead.org, tglx@linutronix.de,
	davej@redhat.com, cpufreq@vger.kernel.org
Subject: Re: [PATCH 7/7] ondemand: Solve the big performance issue with ondemand during disk IO
Date: Fri, 23 Apr 2010 07:24:39 +0200	[thread overview]
Message-ID: <20100423052439.GB4829@ucw.cz> (raw)
In-Reply-To: <20100418120346.1b478410@infradead.org>

Hi!

> The ondemand cpufreq governor uses CPU busy time (e.g. not-idle time) as
> a measure for scaling the CPU frequency up or down.
> If the CPU is busy, the CPU frequency scales up, if it's idle, the CPU
> frequency scales down. Effectively, it uses the CPU busy time as proxy
> variable for the more nebulous "how critical is performance right now"
> question.
> 
> This algorithm falls flat on its face in the light of workloads where
> you're alternatingly disk and CPU bound, such as the ever popular
> "git grep", but also things like startup of programs and maildir using
> email clients... much to the chagarin of Andrew Morton.
> 
> This patch changes the ondemand algorithm to count iowait time as busy,
> not idle, time. As shown in the breakdown cases above, iowait is performance
> critical often, and by counting iowait, the proxy variable becomes a more
> accurate representation of the "how critical is performance"
> question.

Well, and now, if you do something like cat /dev/<your usb1.1 hdd> >
/dev/null, you'll keep cpu on max frequency. Not a problem for new
core i7, but probably big deal for athlon 64.

Maybe modern cpus can and should simply react faster?

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

  parent reply	other threads:[~2010-04-23  5:25 UTC|newest]

Thread overview: 53+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-04-18 18:59 [PATCH 0/7] Fix performance issue with ondemand governor Arjan van de Ven
2010-04-18 19:00 ` [PATCH 1/7] sched: add a comment to get_cpu_idle_time_us() Arjan van de Ven
2010-04-26 19:25   ` Rik van Riel
2010-04-18 19:01 ` [PATCH 2/7] sched: introduce a function to update the idle statistics Arjan van de Ven
2010-04-26 20:11   ` Rik van Riel
2010-04-18 19:01 ` [PATCH 3/7] sched: update the idle statistics in get_cpu_idle_time_us() Arjan van de Ven
2010-04-26 20:36   ` Rik van Riel
2010-04-18 19:02 ` [PATCH 4/7] sched: fold updating of the last update time into update_ts_time_stats() Arjan van de Ven
2010-04-26 20:58   ` Rik van Riel
2010-04-18 19:02 ` [PATCH 5/7] sched: eliminate the ts->idle_lastupdate field Arjan van de Ven
2010-04-26 21:00   ` Rik van Riel
2010-04-18 19:03 ` [PATCH 6/7] sched: introduce get_cpu_iowait_time_us() Arjan van de Ven
2010-04-26 21:05   ` Rik van Riel
2010-04-18 19:03 ` [PATCH 7/7] ondemand: Solve the big performance issue with ondemand during disk IO Arjan van de Ven
2010-04-19  8:29   ` Éric Piel
2010-04-19 13:43     ` Arjan van de Ven
2010-04-19 14:30       ` Éric Piel
2010-04-19 14:47         ` Arjan van de Ven
2010-04-20  9:24       ` Thomas Renninger
2010-04-27  0:29         ` Mike Chan
2010-04-27 13:01           ` Pavel Machek
2010-04-27 18:10             ` Mike Chan
2010-04-19  9:09   ` Tvrtko Ursulin
2010-04-19 13:46     ` Arjan van de Ven
2010-04-19 15:29       ` Tvrtko Ursulin
2010-04-20  0:47         ` Arjan van de Ven
2010-04-20  9:10           ` Tvrtko Ursulin
2010-04-20 11:02             ` Arjan van de Ven
2010-04-28  8:57               ` Tvrtko Ursulin
2010-04-23  5:26           ` Pavel Machek
2010-04-20  9:29   ` Thomas Renninger
2010-04-20 11:07     ` Arjan van de Ven
2010-04-23  5:24   ` Pavel Machek [this message]
2010-04-23  5:38     ` Willy Tarreau
2010-04-23  8:50       ` Thomas Renninger
2010-04-23 16:08         ` Arjan van de Ven
2010-04-27 11:39           ` Thomas Renninger
2010-05-04  3:48             ` [PATCH 8/7] cpufreq: make the iowait-is-busy-time a sysfs tunable Arjan van de Ven
2010-05-04  4:16               ` Willy Tarreau
2010-05-04  5:43               ` Pavel Machek
2010-05-04 13:51               ` Rik van Riel
2010-04-23 14:10       ` [PATCH 7/7] ondemand: Solve the big performance issue with ondemand during disk IO Arjan van de Ven
2010-04-23 15:35         ` Willy Tarreau
2010-04-23 13:52     ` Arjan van de Ven
2010-04-23  8:38       ` Pavel Machek
2010-04-23 16:06         ` Arjan van de Ven
2010-04-24  4:56       ` Pavel Machek
2010-05-01 23:29         ` Arjan van de Ven
2010-04-26 21:30   ` Rik van Riel
2010-04-26 21:41 ` [PATCH 0/7] Fix performance issue with ondemand governor Dave Jones
2010-04-26 21:45   ` Dominik Brodowski
2010-04-26 21:59     ` Rik van Riel
2010-04-26 22:05       ` Dominik Brodowski

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=20100423052439.GB4829@ucw.cz \
    --to=pavel@ucw.cz \
    --cc=akpm@linux-foundation.org \
    --cc=arjan@infradead.org \
    --cc=cpufreq@vger.kernel.org \
    --cc=davej@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=peterz@infradead.org \
    --cc=tglx@linutronix.de \
    /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).