powerd doesn't decrease CPU frequency in some cases
smithi at nimnet.asn.au
Mon Dec 24 03:43:26 PST 2007
On Mon, 24 Dec 2007, Andrey wrote:
> Thank you, Aragon.
> Actually, your patch does the same as mine (but in a different way).
> Thus we have at least 2 cases so far when the challenge is appeared.
> But I'm not sure if our solutions are the best way out.
> And... I believe, that it would be wonderful if Nate Lawson take a look
> here and evaluate the things we are talking through.
I've watched with amazement over the months seeing various problems with
machines having such as 2201 and 2200, or 2001 and 2000MHz frequencies.
It makes no sense to me at all that manufacturers would bother having a
separate frequency just 1MHz higher. It seems ridiculous, and bogus.
Just an observation ..
> Aragon Gouveia wrote:
> > Hi,
> > I recently experienced the exact same behaviour as you are experiencing.
> > I'm running an HP Pavilion dv2610ei. In my case, dev.cpu.0.freq_levels
> > shows frequencies 2201 and 2200 (2.2 GHz Core2Duo CPU), and setting
> > frequency to 2200 jumps to 2201. This was completely messing with powerd as
> > it does not expect a frequency change to jump to another level.
> > I wrote a powerd patch a while ago which adds a check condition and removes
> > a frequency level if it fails to get set. It's attached. Please feel free
> > to test it and report back. :)
> > Regards,
> > Aragon
> Andrey Kosachenko
More information about the freebsd-acpi