On 10/18/2011 02:53 PM, Joe S wrote:
This isn't a problem and I'm not complaining, I'm just a bit curious
as apmd didn't save me as much power as I hoped for. I noticed that
apmd couldn't throttle my cpu in 4.9-RELEASE (amd64). However, since
March 2011, -CURRENT recognizes the K10 cpus, so I wanted to try it
out apmd on my HP Microserver. So I upgraded my system from
4.9-RELEASE to a recent -CURRENT snapshot. When I run "apmd -C"
hw.setperf gets set to 0 and the hw.cpuspeed gets set to 800 MHz. That
was expected. I attached a Kill-a-Watt meter to my system to see what
kind of power savings I would experience with apmd enabled. I think my
expectations were a bit too high. I was only saving .5 watts when my
cpu was throttled down to 800 MHz from 1.3 GHz. I haven't seen many
posts on this subject, so I was wondering if the power savings of .5
watts sounds normal, or if something is wrong on my end.
Were you running a CPU-intensive workload on the CPU(s)? Changing the
clock speed of an idle chip won't change the power usage very much in
absolute terms. If the CPU has multiple cores, exercising them all at
once may maximize power usage so check with all of them running hard.
Memory chips also use more power when cycled, but if the CPU is stalled
waiting for memory it may use less power, so the interaction is not a
priori well defined.
Geoff Steckel
