For the kernel IPI, it's the inter processor interrupt between CPU cores. Does it really mean the interrupts to wake up the system? I think it should prevent the CPU to go to sleep instead of wake up the CPU. To wake up CPU, normally the interrupts should be some devices interrupts. How do you think?
And I'm not sure if different cores in Intel Core2 Due CPU can sleep separately. Thanks Jackie -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Arjan van de Ven Sent: 2008-05-30 11:14 To: shadowkernel Cc: [email protected] Subject: Re: [Discuss] Very high wakeups in PowerTOP shadowkernel wrote: > Greetings, > > I am using a ThinkPad T61(T7500 CPU and NVS 140M Graphic Card), and > Ubuntu Linux with kernel version 2.6.24. Here is my powertop. I notice > that kernel_IPI takes a huge amount of wakeups. I have tried several > methods in Google,such as hpet=force kernel parameter and echo 1 > > /sys/devices/system/cpu/sched_mc_power_savings > in rc.local, but none of them works. 2.6.22 kernel does not have the > problem. > > Any suggestions or comments? > > Thanks, > shadowkernel > > ---------------------------------------------------------- > PowerTOP version 1.9 (C) 2007 Intel Corporation > > Cn Avg residency P-states (frequencies) > C0 (cpu running) (11.2%) 2.21 Ghz 9.6% > C1 0.0ms ( 0.0%) 2.21 Ghz 0.0% > C2 0.0ms ( 0.1%) 1.60 Ghz 0.0% > C3 1.2ms (88.7%) 800 Mhz 90.4% > > > Wakeups-from-idle per second : 755.4 interval: 10.0s > no ACPI power usage estimate available > > Top causes for wakeups: > 49.9% (556.0) <kernel IPI> : Rescheduling interrupts > 15.5% (172.7) audacious : schedule_timeout (process_timeout) > 8.6% ( 96.0) audacious : do_nanosleep (hrtimer_wakeup) very likely the IPI's are a consequence of the audcious process doing a lot of work.. I bet if you stop audacious they'll either go away or way down. _______________________________________________ Discuss mailing list [email protected] http://mail.lesswatts.org/mailman/listinfo/discuss _______________________________________________ Discuss mailing list [email protected] http://mail.lesswatts.org/mailman/listinfo/discuss
