Re: CONFIG_SUSPEND and power consumption

Re: CONFIG_SUSPEND and power consumption

[Rafael J. Wysocki]

Hi,

On Sunday, 19 August 2007 15:32, Jean Delvare wrote:
> Hi all, hi Rafael,
> 
> Running kernel 2.6.23-rc3-git1, I noticed yesterday that my CPU (AMD
> Sempron 2600+) was running at a much lower temperature when
> CONFIG_SUSPEND was enabled.

Hm, interesting.

> The temperature difference was quite significant, about 6 degrees Celsius at
> idle. Measuring the power consumption of my system confirmed that the energy
> savings were real: with CONFIG_SUSPEND=n, the system consumes 80 W of power
> (idle), while with CONFIG_SUSPEND=y, the system consumes only 69 W (idle)!
> Can anyone explain how this works?

I can't.

> I didn't expect CONFIG_SUSPEND to make any difference before actually
> switching the system to standby or suspend state.

Yes, that's the expected behavior.

> I tried the same trick on two Intel motherboards I use for testing, but
> this option didn't seem to make any difference in the power consumption
> for these.

Do you have CONFIG_HIBERNATION set?  If not, please see if setting it instead
of CONFIG_SUSPEND leads to the same result on the affected box (ie. running
at lower temperatures).

Greetings,
Rafael

Re: CONFIG_SUSPEND and power consumption

[Jean Delvare]

Hi Rafael,

On Sun, 19 Aug 2007 22:42:43 +0200, Rafael J. Wysocki wrote:
> On Sunday, 19 August 2007 15:32, Jean Delvare wrote:
> > Hi all, hi Rafael,
> > 
> > Running kernel 2.6.23-rc3-git1, I noticed yesterday that my CPU (AMD
> > Sempron 2600+) was running at a much lower temperature when
> > CONFIG_SUSPEND was enabled.
> 
> Hm, interesting.
> 
> > The temperature difference was quite significant, about 6 degrees Celsius at
> > idle. Measuring the power consumption of my system confirmed that the energy
> > savings were real: with CONFIG_SUSPEND=n, the system consumes 80 W of power
> > (idle), while with CONFIG_SUSPEND=y, the system consumes only 69 W (idle)!
> > Can anyone explain how this works?
> 
> I can't.
> 
> > I didn't expect CONFIG_SUSPEND to make any difference before actually
> > switching the system to standby or suspend state.
> 
> Yes, that's the expected behavior.
> 
> > I tried the same trick on two Intel motherboards I use for testing, but
> > this option didn't seem to make any difference in the power consumption
> > for these.
> 
> Do you have CONFIG_HIBERNATION set?  If not, please see if setting it instead
> of CONFIG_SUSPEND leads to the same result on the affected box (ie. running
> at lower temperatures).

I've made some more tests. More importantly, the additional power
consumption comes from my (new) external hard disk drive. When it's not
connected, the system consume 69 W at idle, regardless of the config
options.

I guess that the drive was being autosuspended (by CONFIG_USB_SUSPEND)
only when CONFIG_SUSPEND was set. But as I found that my drive never
comes back from autosuspend, it's now blacklisted, and my system
consumes 80 W again even with CONFIG_SUSPEND=y.

If I rmmod "ehci-hcd" then the power consumption is back to 69 W. This
confirms that this is really USB-related. I have to admit that I did
not expect an external drive to eat that much power from the system,
especially when not used. I am told that VIA chips are notoriously bad
at this kind of things. I'll try the same external drive on an Intel
system later today.

The last mystery remaining is how USB "activity" can cause my CPU to
heat. I would expect the south bridge to heat, not the CPU.

Thanks,
-- 
Jean Delvare

Re: CONFIG_SUSPEND and power consumption

[Oliver Neukum]

Am Montag 20 August 2007 schrieb Jean Delvare:
> If I rmmod "ehci-hcd" then the power consumption is back to 69 W. This
> confirms that this is really USB-related. I have to admit that I did
> not expect an external drive to eat that much power from the system,
> especially when not used. I am told that VIA chips are notoriously bad
> at this kind of things. I'll try the same external drive on an Intel
> system later today.
> 
> The last mystery remaining is how USB "activity" can cause my CPU to
> heat. I would expect the south bridge to heat, not the CPU.

USB, or strictly speaking EHCI, OHCI and UHCI, use DMA. To allow
that the cache coherency logic has to be active. Therefore your CPU
cannot go to C3. Therefore it draws more power. The problem we are
facing in USB is that to get great savings, our coverage has to be perfect.
One device that cannot be autosuspended and we lose most savings.

	Regards
		Oliver

Re: CONFIG_SUSPEND and power consumption

[Jean Delvare]

On Mon, 20 Aug 2007 12:11:34 +0200, Oliver Neukum wrote:
> Am Montag 20 August 2007 schrieb Jean Delvare:
> > If I rmmod "ehci-hcd" then the power consumption is back to 69 W. This
> > confirms that this is really USB-related. I have to admit that I did
> > not expect an external drive to eat that much power from the system,
> > especially when not used. I am told that VIA chips are notoriously bad
> > at this kind of things. I'll try the same external drive on an Intel
> > system later today.
> > 
> > The last mystery remaining is how USB "activity" can cause my CPU to
> > heat. I would expect the south bridge to heat, not the CPU.
> 
> USB, or strictly speaking EHCI, OHCI and UHCI, use DMA. To allow
> that the cache coherency logic has to be active. Therefore your CPU
> cannot go to C3. Therefore it draws more power. The problem we are
> facing in USB is that to get great savings, our coverage has to be perfect.
> One device that cannot be autosuspended and we lose most savings.

Ah, OK, thanks for the clarification, it explains a lot.

I've made some more tests on two Intel boards and another VIA board.
The bottom line is that both VIA boards see a bump in power consumption
when plugging my USB 2.0 hard disk drive (10 W on one board, 4 W on the
other) while none of the Intel boards exhibit any change in power
consumption. I wonder if I should blame VIA for eating extra power when
the disk is plugged, or thank them for saving power when it's not. Or
maybe I am looking at things the wrong way, and I should thank AMD for
saving more power in C3 than Intel does?

-- 
Jean Delvare

Re: CONFIG_SUSPEND and power consumption

[Pavel Machek]

Hi!

> > If I rmmod "ehci-hcd" then the power consumption is back to 69 W. This
> > confirms that this is really USB-related. I have to admit that I did
> > not expect an external drive to eat that much power from the system,
> > especially when not used. I am told that VIA chips are notoriously bad
> > at this kind of things. I'll try the same external drive on an Intel
> > system later today.
> > 
> > The last mystery remaining is how USB "activity" can cause my CPU to
> > heat. I would expect the south bridge to heat, not the CPU.
> 
> USB, or strictly speaking EHCI, OHCI and UHCI, use DMA. To allow
> that the cache coherency logic has to be active. Therefore your CPU
> cannot go to C3. Therefore it draws more power. The problem we are
> facing in USB is that to get great savings, our coverage has to be perfect.
> One device that cannot be autosuspended and we lose most savings.

Ok.. but CONFIG_USB_SUSPEND should not really have anything to do with
CONFIG_SUSPEND (= s2ram). Perhaps it should depend on CONFIG_PM
instead?
							Pavel

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

Re: CONFIG_SUSPEND and power consumption

[Rafael J. Wysocki]

On Monday, 20 August 2007 19:29, Pavel Machek wrote:
> Hi!
> 
> > > If I rmmod "ehci-hcd" then the power consumption is back to 69 W. This
> > > confirms that this is really USB-related. I have to admit that I did
> > > not expect an external drive to eat that much power from the system,
> > > especially when not used. I am told that VIA chips are notoriously bad
> > > at this kind of things. I'll try the same external drive on an Intel
> > > system later today.
> > > 
> > > The last mystery remaining is how USB "activity" can cause my CPU to
> > > heat. I would expect the south bridge to heat, not the CPU.
> > 
> > USB, or strictly speaking EHCI, OHCI and UHCI, use DMA. To allow
> > that the cache coherency logic has to be active. Therefore your CPU
> > cannot go to C3. Therefore it draws more power. The problem we are
> > facing in USB is that to get great savings, our coverage has to be perfect.
> > One device that cannot be autosuspended and we lose most savings.
> 
> Ok.. but CONFIG_USB_SUSPEND should not really have anything to do with
> CONFIG_SUSPEND (= s2ram). Perhaps it should depend on CONFIG_PM
> instead?

CONFIG_USB_SUSPEND doesn't depend on CONFIG_SUSPEND.

Greetings,
Rafael

Re: CONFIG_SUSPEND and power consumption

[Pavel Machek]

Hi!

> > > > If I rmmod "ehci-hcd" then the power consumption is back to 69 W. This
> > > > confirms that this is really USB-related. I have to admit that I did
> > > > not expect an external drive to eat that much power from the system,
> > > > especially when not used. I am told that VIA chips are notoriously bad
> > > > at this kind of things. I'll try the same external drive on an Intel
> > > > system later today.
> > > > 
> > > > The last mystery remaining is how USB "activity" can cause my CPU to
> > > > heat. I would expect the south bridge to heat, not the CPU.
> > > 
> > > USB, or strictly speaking EHCI, OHCI and UHCI, use DMA. To allow
> > > that the cache coherency logic has to be active. Therefore your CPU
> > > cannot go to C3. Therefore it draws more power. The problem we are
> > > facing in USB is that to get great savings, our coverage has to be perfect.
> > > One device that cannot be autosuspended and we lose most savings.
> > 
> > Ok.. but CONFIG_USB_SUSPEND should not really have anything to do with
> > CONFIG_SUSPEND (= s2ram). Perhaps it should depend on CONFIG_PM
> > instead?
> 
> CONFIG_USB_SUSPEND doesn't depend on CONFIG_SUSPEND.

Strange... what is going on here, then?

config USB_SUSPEND
        bool "USB selective suspend/resume and wakeup (EXPERIMENTAL)"
        depends on USB && PM && EXPERIMENTAL
        help
          If you say Y here, you can use driver calls or the sysfs
          "power/state" file to suspend or resume individual USB
          peripherals.

          Also, USB "remote wakeup" signaling is supported, whereby
some
          USB devices (like keyboards and network adapters) can wake
up
          their parent hub.  That wakeup cascades up the USB tree, and
          could wake the system from states like suspend-to-RAM.

          If you are unsure about this, say N here.

config USB_OTG
        bool
        depends on USB && EXPERIMENTAL
        select USB_SUSPEND
        default n

hmmm, it looks like USB_OTG can be selected without CONFIG_PM, but it
selects USB_SUSPEND. Is that okay?
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: CONFIG_SUSPEND and power consumption

[Rafael J. Wysocki]

On Monday, 20 August 2007 22:19, Pavel Machek wrote:
> Hi!
> 
> > > > > If I rmmod "ehci-hcd" then the power consumption is back to 69 W. This
> > > > > confirms that this is really USB-related. I have to admit that I did
> > > > > not expect an external drive to eat that much power from the system,
> > > > > especially when not used. I am told that VIA chips are notoriously bad
> > > > > at this kind of things. I'll try the same external drive on an Intel
> > > > > system later today.
> > > > > 
> > > > > The last mystery remaining is how USB "activity" can cause my CPU to
> > > > > heat. I would expect the south bridge to heat, not the CPU.
> > > > 
> > > > USB, or strictly speaking EHCI, OHCI and UHCI, use DMA. To allow
> > > > that the cache coherency logic has to be active. Therefore your CPU
> > > > cannot go to C3. Therefore it draws more power. The problem we are
> > > > facing in USB is that to get great savings, our coverage has to be perfect.
> > > > One device that cannot be autosuspended and we lose most savings.
> > > 
> > > Ok.. but CONFIG_USB_SUSPEND should not really have anything to do with
> > > CONFIG_SUSPEND (= s2ram). Perhaps it should depend on CONFIG_PM
> > > instead?
> > 
> > CONFIG_USB_SUSPEND doesn't depend on CONFIG_SUSPEND.
> 
> Strange... what is going on here, then?
> 
> config USB_SUSPEND
>         bool "USB selective suspend/resume and wakeup (EXPERIMENTAL)"
>         depends on USB && PM && EXPERIMENTAL
>         help
>           If you say Y here, you can use driver calls or the sysfs
>           "power/state" file to suspend or resume individual USB
>           peripherals.
> 
>           Also, USB "remote wakeup" signaling is supported, whereby
> some
>           USB devices (like keyboards and network adapters) can wake
> up
>           their parent hub.  That wakeup cascades up the USB tree, and
>           could wake the system from states like suspend-to-RAM.
> 
>           If you are unsure about this, say N here.
> 
> config USB_OTG
>         bool
>         depends on USB && EXPERIMENTAL
>         select USB_SUSPEND
>         default n
> 
> hmmm, it looks like USB_OTG can be selected without CONFIG_PM, but it
> selects USB_SUSPEND. Is that okay?

Well, I don't think so.  It should depend on the same things as USB_SUSPEND,
IMO.

Greetings,
Rafael