On 9 January 2014 02:30, Mathieu Poirier <mathieu.poirier@xxxxxxxxxx> wrote:
> On 8 January 2014 13:35, Sriram V <vshrirama@xxxxxxxxx> wrote:
>> I am seeing a behaviour where the load variations are very frequency
>> between 1ghz and 300mhz in a very short interval of time giving the
>> ping-pong effect. \
>> Is there a way of avoiding this Also, is it possible to find out which
>> task is causing this behaviour.
>>
>> All this happens in a couple of sampling periods of ondemand.
> I am not specifically familiar with the ondemand governor but have
> amply dealt with governor matters. On way to deal with the ping pong
> effect is to increase the amount of time the frequency needs to stay
> at a certain level before going back down.
These are the tunables present with ondemand governor:
struct od_dbs_tuners {
unsigned int ignore_nice_load;
unsigned int sampling_rate;
unsigned int sampling_down_factor;
unsigned int up_threshold;
unsigned int powersave_bias;
unsigned int io_is_busy;
};
And none of them tracks the time for which you should stay at a particular
freq :)
Though you can play with sampling_down_factor and up_threshold to
achieve what you want. For their details, please refer:
Documentation/cpu-freq/governors.txt
Please give us values of these tunables from sysfs for your case, if you
aren't able to make your system better by playing with above tunables.
--
To unsubscribe from this list: send the line "unsubscribe cpufreq" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
