Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 25 Jun 2010 18:25:18 +1000 (EST)
From:      Ian Smith <smithi@nimnet.asn.au>
To:        Stefan Farfeleder <stefan@fafoe.narf.at>
Cc:        freebsd-acpi@freebsd.org
Subject:   Re: Dell Precision m4400 acline problems
Message-ID:  <20100625174549.T9227@sola.nimnet.asn.au>
In-Reply-To: <20100624151811.GK1735@mole.fafoe.narf.at>
References:  <20100620161124.GA1692@mole.fafoe.narf.at> <20100622010612.T9227@sola.nimnet.asn.au> <20100624140454.GJ1735@mole.fafoe.narf.at> <20100625005304.M9227@sola.nimnet.asn.au> <20100624151811.GK1735@mole.fafoe.narf.at>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 24 Jun 2010, Stefan Farfeleder wrote:
 > On Fri, Jun 25, 2010 at 01:05:38AM +1000, Ian Smith wrote:
 > > On Thu, 24 Jun 2010, Stefan Farfeleder wrote:
 > >  > 
 > >  > Thanks. I think I solved my problem. I do not use power profiles and
 > >  > thus the dev.cpu.0.freq stays at 2801. However going offline evidently
 > >  > causes the frequency to decrease internally, ie. without dev.cpu.0.freq
 > >  > knowing, so when I manually reset the frequency to 2801, the system is
 > >  > fast again.
 > > 
 > > Goodo.  But adding these to /etc/rc.conf would save that manual bother:
 > >  performance_cpu_freq="HIGH"
 > >  economy_cpu_freq="HIGH"
 > > since /etc/rc.d/power_profile is run on ACAD events anyway, at least 
 > > with the default /etc/devd.conf
 > 
 > No, unfortunately that's not sufficient.  I really have to decrease and
 > increase the frequency.  I think the reason is that the ACPI kernel code
 > simply ignores setting the frequency to the same value.

Stefan, I'm afraid that this isn't making a lot of sense, to me anyway.

You say that switching to battery "evidently causes the frequency to 
decrease internally, ie. without dev.cpu.0.freq knowing" which, if true, 
would indicate a serious bug, and not one I've seen mentioned before.

You said earlier that, on switching to battery, performance decreases to 
50% or less, and stays there when switching back to AC.  How exactly are 
you measuring that performance, if dev.cpu.0.freq is then still showing 
your top freq of 2801?

How are you lowering then raising the frequency? (show, don't tell)

Perhaps if you provide some more information when the machine is in both 
AC and battery state, it may provide some clues as to what's going on.

Off the top of my head, posting the below for both states might help:

# sysctl dev.cpu hw.acpi
# vmstat -i
# whatever tool you are using to measure 'performance'

(Does anyone else have any better ideas on how to instrument this?)

cheers, Ian



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100625174549.T9227>