Hi!
I have installed Ubuntu Hardy Aplha 1, this is the output of PowerTop v1.9

Notice C2 state is the lowest available, this is because recent F.23A
BIOS update for HP Pavilion laptop series started preventing CPU
entering lower power states with AC plug plugged in (I wonder how this
could be considered a wise implementation...).

Notice C2 state is at 50% average, even though all I'm actually doing is
typing emails.
No task running in background. With same hardware and drivers but 2.6.23
kernel I could get 90% C2 residency.
> Cn                Avg residency       P-states (frequencies)
> C0 (cpu running)        (44.3%)        2.01 Ghz     1.9%
> C1                0.0ms ( 0.0%)              2.00 Ghz     0.0%
> C2                0.1ms (55.7%)             1.60 Ghz     0.0%
>                                                          800 Mhz    98.1%
Notice the HUGE amount of wakeups. I never ever had so much of them!
(around 60-80 with 2.6.23 kernel)
There is also a mismatch between this amount of wakeups and the detailed
list here beneath.
> Wakeups-from-idle per second : 6268.4   interval: 10.0s
> Power usage (5 minute ACPI estimate) :   0.1 W (597.5 hours left)
Notice the power usage row, I'm with AC so why does it get there (it
always does after some minutes I keep powertop running, despite AC
plugged in)?

Notice also the exaggerate amount of wakeups  for keyboard and  touchpad
(they were lower in 2.6.23 kernel) and in particular the   4.2%
(107.3)      <kernel IPI> : Rescheduling interrupts
I know it's something related to the CPU rescheduling, but the CPU is
far from under load (with ondemand governor it is for 98.1% time at
lowest speed, and also in C2 state for half the time).

Notice the <interrupt> : extra timer interrupt which simply wasn't there
with 2.6.23 kernel. Also all the other programs requested less wakeups,
to sum up to 60-80 wakeups.

Why is this kernel behaving sooo bad??
> Top causes for wakeups:
>   61.6% (464.0)       <interrupt> : PS/2 keyboard/mouse/touchpad
>   14.2% (107.3)      <kernel IPI> : Rescheduling interrupts
>   10.3% ( 77.5)       <interrupt> : extra timer interrupt
>    3.1% ( 23.4)       <interrupt> : iwl4965
>    2.8% ( 20.9)              Xorg : schedule_timeout (process_timeout)
>    2.5% ( 18.7)              Xorg : do_setitimer (it_real_fn)
>    1.8% ( 13.2)       compiz.real : schedule_timeout (process_timeout)
>    0.8% (  6.2)          trackerd : schedule_timeout (process_timeout) 

Thank you for yours replies.

_______________________________________________
Power mailing list
[email protected]
http://www.bughost.org/mailman/listinfo/power

Reply via email to