Re: powerd adaptive mode latching

From: Kevin Oberman <oberman_at_es.net>
Date: Sat, 12 Jan 2008 14:37:52 -0800
> Date: Sat, 12 Jan 2008 13:27:46 -0800
> From: Maxim Sobolev <sobomax_at_digifonica.com>
> 
> I wonder if somebody did measurement of power consumption with powerd 
> and without it on typical tasks. There is very interesting idea in the 
> last issue of ACM Queue that it might be much more beneficial to run CPU 
> at the full speed and then switch it to low-power mode as soon as 
> possible in the idle loop than to run longer at reduced speed for a 
> longer period of time.

Now that you mention it, I did do this back when powerd first
appeared. My results lead me to believe the Queue article is likely
correct. 

Here is a summary of my results:

1. Of course, powerd does nothing during processor intensive
   operations. 

2. For many interactive operations such as editing/word processing,
   spreadsheets, reading mail, and the like, the savings from speedstep
   were fairly significant. 

3. For moderate CPU utilization tasks such as playing music (mp3), the
   savings from speedstep can be very significant. This was the spot
   where powerd did the best, but I could beat it by doing manual
   settings. 

4. Throttling/TCC were often counter-productive. TCC did slightly better
   than throttling, but i simply disable both. Playing music with TCC
   showed a steady deterioration in power consumption as TCC was applied
   more aggressively.

I will try to find my test results. They were done on an IBM T30 about
three years ago and I may not find them now. This system had only
speedstep and I suspect ESS will do much better. If I have time, I will
try some tests with my current T43 and, maybe a T61 to which I have
access.
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman_at_es.net			Phone: +1 510 486-8634
Key fingerprint:059B 2DDF 031C 9BA3 14A4  EADA 927D EBB3 987B 3751

Received on Sat Jan 12 2008 - 21:37:57 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:25 UTC