Skip to contentSkip to conversation

usb still sucks battery in -rc7-mm1

DISCUSSION Pavel Machek 5 participants 9 messages Sep 24, 2006 – Sep 27, 2006

lore
Share & tools
lore
#0 lore raw

Hi!

I made some quick experiments, and usb still eats 4W of battery power. (With whole machine eating 9W, that's kind of a big deal)...

This particular machine has usb bluetooth, but it can be disabled by firmware, and appears unplugged. (I did that). It also has fingerprint scanner, that can't be disabled, but that does not have driver (only driven by useland, and was unused in this experiment).

Any ideas?
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#1 lore raw

Perhaps your cpu cannot go into deep c states? At least my IBm T40p has that problem, when USB is enabled.

regards
Alex

Am Sonntag, 24. September 2006 11:08 schrieb Pavel Machek:

Quoted 12 lines

Hi!

I made some quick experiments, and usb still eats 4W of battery power. (With whole machine eating 9W, that's kind of a big deal)...

This particular machine has usb bluetooth, but it can be disabled by firmware, and appears unplugged. (I did that). It also has fingerprint scanner, that can't be disabled, but that does not have driver (only driven by useland, and was unused in this experiment).

Any ideas?
Pavel

Signature · 3 lines
Encrypted Mails welcome.
PGP-Key at http://www.grans.eu/misc/pgpkey.asc | Key-ID: 0x6D7DD291
More info at http://www.grans.eu/Alexander_Gran.html
#2 lore raw
Quoted 2 lines · Alexander Gran

Perhaps your cpu cannot go into deep c states? At least my IBm T40p has that problem, when USB is enabled.

Yes, and latest -mm's were expected to fix that...

Pavel

Quoted 22 lines

regards
Alex

Am Sonntag, 24. September 2006 11:08 schrieb Pavel Machek:

Quoted 12 lines

Hi!

I made some quick experiments, and usb still eats 4W of battery power. (With whole machine eating 9W, that's kind of a big deal)...

This particular machine has usb bluetooth, but it can be disabled by firmware, and appears unplugged. (I did that). It also has fingerprint scanner, that can't be disabled, but that does not have driver (only driven by useland, and was unused in this experiment).

Any ideas?
Pavel

Signature · 3 lines
Encrypted Mails welcome.
PGP-Key at http://www.grans.eu/misc/pgpkey.asc | Key-ID: 0x6D7DD291
More info at http://www.grans.eu/Alexander_Gran.html
Signature · 1 line
Thanks for all the (sleeping) penguins.
#3 lore raw
In reply to Pavel Machek
Quoted 9 lines · Pavel Machek

I made some quick experiments, and usb still eats 4W of battery power. (With whole machine eating 9W, that's kind of a big deal)...

This particular machine has usb bluetooth, but it can be disabled by firmware, and appears unplugged. (I did that). It also has fingerprint scanner, that can't be disabled, but that does not have driver (only driven by useland, and was unused in this experiment).

Any ideas?

(I have not followed the thread, so bear with me if I say anything irrelavant)

I have encountered at least 3 hubs (2 usb2 & 1 usb1) that will consume a lot of power (about 2-2.5W if the laptop power consumption readings are to be trusted) and heat up a lot (to the point of being too hot to touch for more than a few seconds) even when no devices are connected, at least on Linux. I have not tested them on a Windows machine to see if this is the case there. The USB2 ones used Cypress chips. I do not know what your h/w config is, but perhaps this is a similar case ?

Regards,

Theodoros Kalamatianos

PS: Note that the 2-2.5W consumption is the maximum that the laptop USB port can provide, and I have noticed that with the hub's external power supply connected it heats up even more.

#4 lore raw

Hi!

Quoted 21 lines
Quoted 9 lines

I made some quick experiments, and usb still eats 4W of battery power. (With whole machine eating 9W, that's kind of a big deal)...

This particular machine has usb bluetooth, but it can be disabled by firmware, and appears unplugged. (I did that). It also has fingerprint scanner, that can't be disabled, but that does not have driver (only driven by useland, and was unused in this experiment).

Any ideas?

(I have not followed the thread, so bear with me if I say anything irrelavant)

I have encountered at least 3 hubs (2 usb2 & 1 usb1) that will consume a lot of power (about 2-2.5W if the laptop power consumption readings are to be trusted) and heat up a lot (to the point of being too hot to touch for more than a few seconds) even when no devices are connected, at least on Linux. I have not tested them on a Windows machine to see if this is the case there. The USB2 ones used Cypress chips. I do not know what your h/w config is, but perhaps this is a similar case ?

Quoted 3 lines

PS: Note that the 2-2.5W consumption is the maximum that the laptop USB port can provide, and I have noticed that with the hub's external power supply connected it heats up even more.

I'm not using any external hubs. I believe it is old "uhci keeps cpu from reaching deep c states" problem. I just want to ping greg/list, because greg told me problem is solved in -mm.
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#5 lore raw
In reply to Theodoros V. Kalamatianos
Quoted 7 lines · Theodoros V. Kalamatianos

I have encountered at least 3 hubs (2 usb2 & 1 usb1) that will consume a lot of power (about 2-2.5W if the laptop power consumption readings are to be trusted) and heat up a lot (to the point of being too hot to touch for more than a few seconds) even when no devices are connected, at least on Linux. I have not tested them on a Windows machine to see if this is the case there. The USB2 ones used Cypress chips. I do not know what your h/w config is, but perhaps this is a similar case ?

Probably not unrelated. I have one of those Cypress-based hubs; it's nice except for the heat/power, which precludes using them in bus-powered mode with a laptop (or anything) ... I don't much like wall-warts!

It might be too much to expect that when that hub's upstream port is suspended, its power usage goes to something reasonable.

#6 lore raw
In reply to Pavel Machek
Quoted 11 lines · Pavel Machek

Hi!

I made some quick experiments, and usb still eats 4W of battery power. (With whole machine eating 9W, that's kind of a big deal)...

This particular machine has usb bluetooth, but it can be disabled by firmware, and appears unplugged. (I did that). It also has fingerprint scanner, that can't be disabled, but that does not have driver (only driven by useland, and was unused in this experiment).

Any ideas?

The USB autosuspend patches are still not entirely in -mm. They contain a couple of bugs that have to get fixed first. When they do get merged you should see considerable improvement. Note that although they will reduce the amount of power being used by the USB controllers and will stop the DMA activity (thus allowing your CPU to enter C2), they won't put the controllers into D3. For that you'll have to get PCI autosuspend... :-)

In the meantime, if all you care about is power consumption there are some things you can do. The easiest is simply to rmmod ehci-hcd, ohci-hcd, and uhci-hcd. After all, if you're not using USB there's no reason to let the drivers eat up memory, CPU time, and power.

Alan Stern

#7 lore raw

Hi!

Quoted 10 lines
Quoted 9 lines

I made some quick experiments, and usb still eats 4W of battery power. (With whole machine eating 9W, that's kind of a big deal)...

This particular machine has usb bluetooth, but it can be disabled by firmware, and appears unplugged. (I did that). It also has fingerprint scanner, that can't be disabled, but that does not have driver (only driven by useland, and was unused in this experiment).

Any ideas?

Quoted 7 lines

The USB autosuspend patches are still not entirely in -mm. They contain a couple of bugs that have to get fixed first. When they do get merged you should see considerable improvement. Note that although they will reduce the amount of power being used by the USB controllers and will stop the DMA activity (thus allowing your CPU to enter C2), they won't put the controllers into D3. For that you'll have to get PCI autosuspend... :-)

I have not measured that, but I _hope_ drain by controller itself will not be big enough.

Quoted 4 lines

In the meantime, if all you care about is power consumption there are some things you can do. The easiest is simply to rmmod ehci-hcd, ohci-hcd, and uhci-hcd. After all, if you're not using USB there's no reason to let the drivers eat up memory, CPU time, and power.

Are autosuspend patches available somewhere? (Relative to -mm, or relative to 2.6.18?) I'd like to play with them...
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#8 lore raw
Quoted 2 lines · Pavel Machek

Are autosuspend patches available somewhere? (Relative to -mm, or relative to 2.6.18?) I'd like to play with them...

The versions in the mailing list archives are a little old. I'll post some updates later this week.

Alan Stern