On 06/05/14 04:53, Johan Svensson wrote:

On 06/05/14 00:53, STeve Andre' wrote:
On 06/04/14 17:08, Johan Svensson wrote:
I'm trying to migrate from Linux to Openbsd on my laptop (thinkpad x201).

The first problem that i came across was that the Cpu fanspeed was running constantly at 3500RPM. After the acpithinkpad.c patch from jcs (and i modified to make it work on the openbsd-current(link: http://exclude.se/patch/jcs_mod_by_js.diff)

Another thing that i noticed is that the battery lifetime is really bad.
In Linux i get around ~5,5 hours.
In OpenBSD i get around 2 hours.

when i ran : sysctl hw.sensors | grep -i consumption.
the output of the cpu was 6W.

in Linux it's around 1,5W.

with: apmd -C and apmd -L it's the same.
dmesg: http://exclude.se/openbsd/dmesg.txt

Is there anyway to fix this?

Regards
Johan Svensson


Take a look at hw.setperf in sysctl.  I think you are running at the
maximum cpu speed?  On my 2.8GHz W500 I can run at 800, 1600,
2133 and 2801.  800MHz makes a huge difference.  You have to
try different values for setperf to see what happens.  sysctl will
also tell you the speed in hw.cpuspeed.

--STeve Andre'
This my output from sysctl and apm when running on the lowest clockspeed:
# sysctl hw | grep -iE "cpuspeed|setperf|fan|consumption"
hw.sensors.acpithinkpad0.fan0=1959 RPM
hw.sensors.itherm0.power0=6.00 W (CPU power consumption)
hw.cpuspeed=1199
hw.setperf=0
# apm
Battery state: high, 70% remaining, 111 minutes life estimate
A/C adapter state: not connected
Performance adjustment mode: manual (1199 MHz)


This is the output when i use apm -H:
# sysctl hw | grep -iE "cpuspeed|setperf|fan|consumption"
hw.sensors.acpithinkpad0.fan0=1972 RPM
hw.sensors.itherm0.power0=6.00 W (CPU power consumption)
hw.cpuspeed=2666
hw.setperf=100
# apm
Battery state: high, 68% remaining, 107 minutes life estimate
A/C adapter state: not connected
Performance adjustment mode: manual (2666 MHz)

The energy consumption is the same which is odd.

--Johan

Hmmm.  Smells like a bug, to me.  But by changing hw.setperf your
self you should be able to go to other speeds(?).  And of course, the
real test is to see if you get longer life at setperf 0.

--STeve Andre'

Reply via email to