lkml.org 
[lkml]   [2008]   [Aug]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRE: [linux-pm] [RFC-PATCH] Improve Menu Governor Prediction of interrupted C states.
Date
From
> Hi,
>
>> I also found current cpuidle Menu governor should have some problems
>> while predicting next non-expected break-event after some expected
>> break-events. The measured_us/elapsed_us/predicted_us will become
larger
>> and larger once (measured_us + BREAK_FUZZ >= data->expected_us -
>> target->exit_latency). The major point is that it should be
>> last_residency, not measured_us, that need to be used to do comparing
>> and distinguish between expected & non-expected events.
>>
>> Below is my draft patch (not tested) for this. It is simple and
should
>> also be effective for high interrupt rates case.
>
> This does seem to give better guesses. However, first tests are not
showing
> as well as the irq time stamp version I posted.
>
> I'll need to grab some hardware trace to see what is going on.

Thanks for the trial. My original patch has already been merged into
another patch "[patch 2/3] cpuidle: Menu governor fix wrong usage of
measured_us" sent out by Venkatesh. The new patch should be more intact.

Jimmy


\
 
 \ /
  Last update: 2008-08-06 03:55    [W:0.041 / U:0.572 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site