[ltp] New PowerTOP utility for Linux 2.6.21+ from Intel

Chris Hanson linux-thinkpad@linux-thinkpad.org
Tue, 15 May 2007 00:20:08 -0400


Henrique de Moraes Holschuh wrote:
> On Sun, 13 May 2007, Chris Hanson wrote:
>> output.  hdapsd and the hdaps module, however, are both significant,
> 
> That can't be avoided.  hdaps needs to pool the EC at somewhat high
> frequencies to be effective (say, 50Hz or so), and that also disturbs the
> EC, yes, it is nowhere near cheap in power.

Hmmm... but PowerTOP seems to indicate that the kernel module and the
userspace daemon are each independently polling at 50Hz, which could be
improved.

>> problems are: USB at #1 (23%), fglrx at #2 (14%), and ipw3945 at #5
> 
> Always remove uhci-hcd unless you *really* need usb 1.1.  ehci-hcd is less
> of a power hog at runtime

I didn't know that it was possible to use USB 1.1 devices with ehci-hcd.
   Can I just blacklist uhci-hcd in module-utils?

> (but do check if you should not be removing it on sleep-to-ram).

How do I check that?  By trying it and looking for lossage?  I thought
that unloading the USB drivers was standard practice for any kind of
suspend.

That said, I've never even tried to use sleep-to-ram or sleep-to-disk on
this machine.  My previous experience was that making it work required a
lot of tweaking and rebooting, and I haven't been able to justify the
necessary time.  But that was a year or two ago, on my previous laptop
(an X31).  Maybe things are better now.

By the way, it turns out that the high USB interrupts go away if the
wireless kill switch is changed to disable wireless.  I'm guessing the
interrupts have something to do with bluetooth.