Re: powerd(8)

> From: "Daniel O'Connor" <doconnor_at_gsoft.com.au>
> Date: Tue, 19 Apr 2005 10:26:55 +0930
> 
> --nextPart20628293.OurAszTEo5
> Content-Type: text/plain;
>   charset="utf-8"
> Content-Transfer-Encoding: quoted-printable
> Content-Disposition: inline
> 
> On Tue, 19 Apr 2005 08:59, Kevin Oberman wrote:
> > I also have found that changing the polling interval to 150 is an
> > improvement. Half a second is just too long, IMHO. I also discovered
> > that with my system (P4M) that some settings can use substantially more
> > power than faster settings, so I have kludged an ugly hack to avoid
> > those settings. These changes make a significant difference in power
> > consumption.
> 
> Can you elaborate on these?
> It may be worth adding a general algorithm to cull the frequency list power> d 
> uses based on this info.
> 
> -- 
> Daniel O'Connor software and network engineer
> for Genesis Software - http://www.gsoft.com.au

You can see the thread in acpi_at_ archives, but simply states, lower the
actual CPU reduces power far more than throttling with TCC. I have two
available clock rates (1.8 GHz and 1.2 GHz) and TCC provides 8 evenly
spaced steps of 12.5% each. The impact on performance, though, is nearly
identical. As a result, if there is a choice between the two, you want to
run with the lower frequency nd less throttle, but the current code does
it the other way. As a result I am limited to 1800, 1575, 1350, 1200,
1050, 750, 600, 300, and 150.

cpu.dev.0.freq	Temperature	CPU Speed
1800		>_PSV		1800
1575		>_PSV		1800
1350		85		1800
1200		73		1200
1125		82		1800
1050		69		1200
900		77		1800
750		64		1200
675		72		1800
600		62		1200
450		66		1800
300		56		1200
225		61		1800
150		54		1200

Temperature is the steady state CPU temperature when running the CPU at
full speed (md5) for 5 minutes. It should be proportional to power
consumption.

I don't know how well this plays with EST or standard throttling
(non-P4TCC) and, since my system does not provide energy numbers (it's
not EST), there is no easy way to know how it should run except testing.

When I get a little more time, I hope to do some more work on this, but,
for now, go with PHK's mods and, if energy numbers are available, base
the algorithm on that.
-- 
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
-- 
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