All of lore.kernel.org
 help / color / mirror / Atom feed
From: Len Brown <lenb@kernel.org>
To: Zdenek Kabelac <zdenek.kabelac@gmail.com>
Cc: Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	cpufreq@vger.kernel.org
Subject: Re: Perfomance governor is still changing frequency
Date: Mon, 15 Dec 2008 12:59:15 -0500 (EST)	[thread overview]
Message-ID: <alpine.LFD.2.00.0812151242430.16823@localhost.localdomain> (raw)
In-Reply-To: <c4e36d110812150515q1c4b2458wa61c79c5a85c624e@mail.gmail.com>



On Mon, 15 Dec 2008, Zdenek Kabelac wrote:

> Hi
> 
> I'm noticing strange behavior on my T61.
> 
> When I select performance governor  for my CPU I still get occasional
> drop on CPU frequency from 2.2.GHz to 800MHz.
> 
> I'm attaching my .config file.
> 
> How it could happen that perfomance is actually able to change
> frequency from it's highest mode - is it because of some overheating ?

performance governor itself will not change the frequency.

> (It usually happens when I run   make -j2 -  even just for a minute -
> it suddenly start to drop the CPU to lowest clock rate -
> imho I think the machine is quite calm to be 'overheated')
> 
> What other debug info is needed ?
> 
> 
> affected_cpus
> 0 1
> cpuinfo_cur_freq
> 2201000
> cpuinfo_max_freq
> 2201000
> cpuinfo_min_freq
> 800000
> related_cpus
> 0 1
> scaling_available_frequencies
> 2201000 2200000 1600000 1200000 800000
> scaling_available_governors
> ondemand performance
> scaling_cur_freq
> 2201000
> scaling_driver
> acpi-cpufreq
> scaling_governor
> performance
> scaling_max_freq
> 2201000
> scaling_min_freq
> 800000
> scaling_setspeed
> <unsupported>
> 
> 
>  stats/time_in_state
> 2201000 331831
> 2200000 2876
> 1600000 1739
> 1200000 3398
> 800000 512768
> 
> 
> (btw why do I have two states  - 2201000 and 2200000 ??)


2201000 is the marketing speed of your processor, plus 1MHz.
This is used by the ACPI BIOS to enable "Turbo Mode",
aka Intel Dynamic Acceleration Technology --
which gives the harware license to run faster than
2.2GHz if the conditions are right.
ie. when there is still no idle time available at 2.2GHz,
ondemand chooses 2.21 Ghz, and the HW may give it more...

The right conditions depend on the part -- could be that
some cores need to be in deep idle to let other cores
go fast, or it could be based on current and thermal
constraints, or both.

I have a 2.0 GHz T61, and I'm unable to reproduce
your observation.

You can eliminate turbo from the picture by
echo 2200000 > time_in_state

and that will tell us if turbo is related to your
occasional obervance of low-frequency-mode.

You can also observe the temperature in
/proc/acpi/thermal*/*/temperature

cheers,
-Len


  reply	other threads:[~2008-12-15 17:59 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-12-15 13:15 Perfomance governor is still changing frequency Zdenek Kabelac
2008-12-15 17:59 ` Len Brown [this message]
2008-12-23 17:56   ` Matthew Garrett
2008-12-15 20:53 ` Tim Blechmann

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=alpine.LFD.2.00.0812151242430.16823@localhost.localdomain \
    --to=lenb@kernel.org \
    --cc=cpufreq@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=zdenek.kabelac@gmail.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.