Re: powerd adaptive mode latching

Hi Nate,

Nate Lawson wrote:
> Stefan Lambrev wrote:
>   
>> It works :)
>>
>> Nate Lawson wrote:
>>     
>>> Nate Lawson wrote:
>>>   Please test the patch.  I need at least one "it works" from someone who
>>> has duplicated cpufreq states to commit it to -current.  Be sure to
>>> first remove any acpi_perf patch that you previously applied.  Thanks.
>>>
>>>   
>>>       
>> I applied your patch on clean releng_7_0 from today an it works - here
>> is the output from sysctl:
>>
>> dev.cpu.0.freq_levels: 2201/35000 1925/30625 1650/26250 1600/23000
>> 1400/20125 1200/16000 1050/14000 900/12000 800/14000 700/12250 600/10500
>>     
>
> Thanks for testing.  Can you be sure the 2201 Mhz setting works
> properly.  Perhaps you can set dev.cpu.0.freq=2201 then 1600 and be sure
> it has approximately correct CPU performance.  One test I use is:
>
>   dd if=/dev/zero bs=128k count=N | sha256
>
> where "N" is some number that takes about 10 seconds to complete.
>
>   
dev.cpu.0.freq: 2201
dd if=/dev/zero bs=128k count=10000 | sha256 - takes 16.453 sec
at dev.cpu.0.freq: 1925 it takes 19.020 sec
at dev.cpu.0.freq: 1600 - 22.629 sec

Everything works as expected. :)

-- 

Best Wishes,
Stefan Lambrev
ICQ# 24134177