linux-kernel - Re: Enabling power states for Core 2 Duo

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-Id: <200705240209.03198.maximlevitsky@gmail.com>
Date:	Thu, 24 May 2007 02:09:02 +0300
From:	Maxim Levitsky <maximlevitsky@...il.com>
To:	Jindrich Makovicka <makovick@...il.com>
Cc:	linux-kernel@...r.kernel.org
Subject: Re: Enabling power states for Core 2 Duo

On Tuesday 22 May 2007 20:35, Jindrich Makovicka wrote:
> On Tue, 22 May 2007 17:14:45 +0000
>
> "Paa Paa" <paapaa125@...mail.com> wrote:
> > >>But are you saying that with most desktop mobos one doesn't usually
> > >>have the
> > >>different power states available at all? So basically the only
> > >>means to conserve power is to scale the frequency?
> > >
> > >Please update your BIOS and try.
> >
> > I updated my Asus P5B Deluxe BIOS with no luck (this latest BIOS is
> > about 1 month old). Still no power states. I would be nice to know if
> > _any_ desktop C2D mobos have these C-states? (In x86_64 system).
> >
> > I think I never mentioned:
> >
> > I'm using 2.6.21.1 (actually gentoo-sources-2.6.21-r1)
> > My CPU is E6400.
>
> I observe the same problem with Gigabyte 965P-DS4, and there are at
> least two causes - 1) MP supported flag in FADT is missing (so
> CPU_HOTPLUG would be necessary), and 2) C2 latency is set to 101,
> while the maximum allowed in the kernel is 100.
>
> However, from the power consumption and the CPU temperature it seems
> that the power saving works anyway, so I'll live with it.

Hi,

I also don't have C-states on my DG965RY intel board with core 2 duo E6400

Any board can support C-states ether by fixed ACPI set of registers or by a 
set of functions in acpi tables.

On my system ICH8 (and I almost sure you got it too), doesn't support 
fixed-function C-states.

And there is no other magic chip that can put cpu in c-state (on my system) , 
so acpi has no "set of functions to do that"

So I guess that on my board there is no physical support for C-states, and 
this could be true for you too.

Anyway monitor/mwait can put cpu in C1, and having apic-stops-in-c2/c3 bug,
I don't think those states are that needed.

The Speedstep support is a different thing at all:
Agian acpi has a function that can change cpu clock speed to fixed values,
And yes on my board it also has only two values: 1500 and 2130.
This support is implemented by accessing the clock chip
(Actualy on my system acpi writes a magic value to one of ICH8 registers, and 
this triggers a SMM that changes the clock)

And the last thing is about throtting states your and mine system does 
support:
Throttling states are states that make system halt cpu for some % of time,
Yes there are 8 such states. You can enter one of those by writeing a state 
number to /proc/acpi/processor/*/throttling (I assume you have ICH8 or 
someting simular)
Also ICH*'s throtting states are system global, this is why I wrote "/*/"
If you change it for one core, it will be changed for other.
They are supported by fixed acpi function (by ICH8)

My 2 cents.

Regards
	Maxim Levitsky
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/