WinXP laptop dynamic speed power management problem

[Yousuf Khan]

Been using a Gateway laptop with a single-core Turion for a while now. I
used to keep its power management setting on either "Portable/Laptop" or
on "Minimal Power Management" most of the time. Using a utility called
MobileMeter, I used to notice its speed reduce from full (1.8Ghz) to a
slow power-saving frequency (800Mhz), due to the AMD PowerNow power
management feature (it's the equivalent of SpeedStep on Intel
processors). That was all fine and good, that's what I wanted. But
recently I started noticing that it lasts a lot less time while under
battery power. So I watched it under MobileMeter again, and noticed that
now the laptop stays 100% of the time on its full speed rather than
going into power saving speed, even if there is not too much activity on
the system.

So then I switched from the "Portable/Laptop" or the "Minimal Power
Management" schemes to the "Max Battery" scheme. And now battery life is
back to normal. But I wonder what could've happened to make the two
previous suitable schemes become unsuitable? I can't think of any major
changes on the laptop, and the CPU loads seem to be idle most of the
time, just as it always used to be. I consider this to be a workaround
rather than a fix.

How does Windows figure out when to switch to lower speed mode, and what
could've changed to interfere with it?

Yousuf Khan

 

[Paul]

Been using a Gateway laptop with a single-core Turion for a while now. I
used to keep its power management setting on either "Portable/Laptop" or
on "Minimal Power Management" most of the time. Using a utility called
MobileMeter, I used to notice its speed reduce from full (1.8Ghz) to a
slow power-saving frequency (800Mhz), due to the AMD PowerNow power
management feature (it's the equivalent of SpeedStep on Intel
processors). That was all fine and good, that's what I wanted. But
recently I started noticing that it lasts a lot less time while under
battery power. So I watched it under MobileMeter again, and noticed that
now the laptop stays 100% of the time on its full speed rather than
going into power saving speed, even if there is not too much activity on
the system.

So then I switched from the "Portable/Laptop" or the "Minimal Power
Management" schemes to the "Max Battery" scheme. And now battery life is
back to normal. But I wonder what could've happened to make the two
previous suitable schemes become unsuitable? I can't think of any major
changes on the laptop, and the CPU loads seem to be idle most of the
time, just as it always used to be. I consider this to be a workaround
rather than a fix.

How does Windows figure out when to switch to lower speed mode, and what
could've changed to interfere with it?

Yousuf Khan

This is just a guess, but could it be related to this ? I know you
have a single core processor, but maybe this change affects power
management.

http://support.microsoft.com/kb/896256

Paul

 

[Yousuf Khan]

This is just a guess, but could it be related to this ? I know you
have a single core processor, but maybe this change affects power
management.

http://support.microsoft.com/kb/896256

Paul

Hi, thanks for the link, but it didn't turn out to be the solution.
However, it inspired me to google along those lines on the Microsoft
support website, and I found the following article:

Different ways to determine CPU speed in Windows XP or in Windows Server
2003
http://support.microsoft.com/kb/888282

Which then led me to the following article:

How to use Powercfg.exe in Windows Server 2003
http://support.microsoft.com/kb/324347/

And that last article turned out to be the exact answer I needed!
Powercfg is a command-line utility version of "Power Options" Control
Panel applet, but it is far more powerful than the CPL. Even though it
says that utility is for Windows Server 2003, I've found it in both XP
(SP2) Home and Pro. The CLI util allows you to display some hidden
parameters in power management not available within the CPL applet, such
as processor throttling states.

I queried the settings of each of the power schemes, and found that a
lot of the throttle states had gotten messed up somehow (don't know
how). Anyways, many of them were showing "none" as the throttle
settings, when they should've been showing something like "adaptive" or
"degrade". I compared the schemes to another computer, and then I reset
them all properly.

Here's examples how you can use the powercfg.exe program from the
command-line:

-powercfg /l {list schemes}
-powercfg /q "portable/laptop" {query parameters of "portable/laptop"
scheme}
-powercfg /x "portable/laptop" /processor-throttle-dc adaptive {change
throttle state to "adaptive" when running under batteries (dc)}

It was really cool finding out about this. Wish I knew what caused the
hidden parts of the power schemes to get screwed up in the first place.
It's probably in some little-known registry variable.

Yousuf Khan