[ltp] pcie_aspm=force - module/pcie_aspm/parameters/policy = powersave any effects?

Martin Steigerwald linux-thinkpad@linux-thinkpad.org
Tue, 28 Jun 2011 12:59:22 +0200


Hallo!

After making a lot of fuss out of it, Michael Larabel finally disclosed=20
results on finding the cause for the main power regression he claimed to=20
have found:

The Leading Cause Of The Recent Linux Kernel Power Problems
http://www.phoronix.com/scan.php?page=3Darticle&item=3Dlinux_2638_aspm&num=
=3D2

I added pcie_aspm=3Dforce to kernel parameters and

module/pcie_aspm/parameters/policy =3D powersave

to sysfs.conf.

I do not have a power meter.

So far I noticed powertop displays around 2610-40 rpm for the fan=20
regarding about 2800-2850 before. I didn=B4t test several times between=20
different settings tough yet. This in on a ThinkPad T520 with Intel i5=20
Sandybridge, connected to AC, full brightness.

Unfortunately in powertop 1.97 I do not see that power consumption in=20
watts anymore that earlier powertop 1 versions displayed when I=20
disconnected AC. And I have not enough comparable data to judge on:

merkaba:/etc> date ; acpi -b
Di 28. Jun 12:53:32 CEST 2011
Battery 0: Discharging, 98%, 03:40:25 remaining
merkaba:/etc> date ; acpi -b
Di 28. Jun 12:53:34 CEST 2011
Battery 0: Discharging, 98%, 03:54:28 remaining
merkaba:/etc> date ; acpi -b
Di 28. Jun 12:53:41 CEST 2011
Battery 0: Discharging, 98%, 04:20:53 remaining
merkaba:/etc> date ; acpi -b
Di 28. Jun 12:53:50 CEST 2011
Battery 0: Discharging, 98%, 04:38:33 remaining
merkaba:/etc> date ; acpi -b
Di 28. Jun 12:53:57 CEST 2011
Battery 0: Discharging, 98%, 04:47:23 remaining
merkaba:/etc> date ; acpi -b
Di 28. Jun 12:54:00 CEST 2011
Battery 0: Discharging, 98%, 04:40:00 remaining

But from what I remember with some earlier tests this actually looks like=20
quite an improvement. Unfortunately my previous with some naive acpi -b=20
measurements is hidden deep somewhere in debian-user-german. This is with=20
the 6 cell battery.

What do you find? Do you know of a tool that measures actual power=20
consumptions via ACPI?

Ciao,
=2D-=20
Martin 'Helios' Steigerwald - http://www.Lichtvoll.de
GPG: 03B0 0D6C 0040 0710 4AFA  B82F 991B EAAC A599 84C7