[ltp] Re: very high number of wakeups (and power consumption) on T60p
Ondrej Balaz
linux-thinkpad@linux-thinkpad.org
Fri, 29 Feb 2008 12:46:31 +0000 (UTC)
On 2008-02-28, Peter F. Patel-Schneider <pfps@research.bell-labs.com> wrote:
> Has anyone run into problems with tens of thousands of wakeups that are
> unexplained in powertop? This does appear to be real as battery drain
> does jump up about 30-40%.
>
> Here is typical powertop output. The machine was just about idle when
> this measurement was taken.
>
>
> PowerTOP version 1.9 (C) 2007 Intel Corporation
>
> Cn Avg residency P-states (frequencies)
> C0 (cpu running) (30.1%) 2.17 Ghz 5.1%
> C1 0.0ms ( 0.0%) 1.67 Ghz 0.0%
> C2 0.0ms (65.3%) 1333 Mhz 0.0%
> C3 0.3ms ( 4.6%) 1000 Mhz 94.9%
>
>
> Wakeups-from-idle per second : 46243.6 interval: 10.0s
> Power usage (ACPI estimate): 26.5W (0.8 hours) (long term: 28.0W,/0.8h)
>
> Top causes for wakeups:
> 12.1% ( 1.8) <interrupt> : extra timer interrupt
check your /proc/interrupts if NMI: line is 0. If not, remove nmi_watchdog=1
from kernel boot arguments.
> 8.7% ( 1.3) wpa_supplicant : schedule_timeout (process_timeout)
> 8.1% ( 1.2) X : do_setitimer (it_real_fn)
> 7.4% ( 1.1) <interrupt> : ahci
> 6.7% ( 1.0) firefox-bin : futex_wait (hrtimer_wakeup)
> 6.7% ( 1.0) tail : do_nanosleep (hrtimer_wakeup)
> 6.7% ( 1.0) multiload-apple : schedule_timeout (process_timeout)
> 6.7% ( 1.0) setroubleshootd : schedule_timeout (process_timeout)
> 6.0% ( 0.9) atieventsd : schedule_timeout (process_timeout)
> 3.4% ( 0.5) <interrupt> : eth0
> 3.4% ( 0.5) NetworkManager : e1000_intr_msi (e1000_watchdog)
> 3.4% ( 0.5) gconfd-2 : schedule_timeout (process_timeout)
> 2.7% ( 0.4) gnome-power-man : schedule_timeout (process_timeout)
> 2.0% ( 0.3) <kernel module> : neigh_table_init_no_netlink (neigh_periodic_
> 2.0% ( 0.3) <kernel core> : neigh_table_init_no_netlink (neigh_periodic_timer)
>
>
>
> Peter F. Patel-Schneider
>
--
Ondrej `blami' Balaz