[Linrad] ACPI bugs in modern computers.

[Leif Asbrink <sm5bsz.com; leif@xxxxxxxxxxxxxxxx>]

Hello All,

This posting might be of interest to PC users in general, not only
to the relatively small group interested in software defined radio.
The general lesson is this: Poor performance can be due to ACPI bugs
in the BIOS of your specific hardware. This shows up as about 90%
cpu load on one processor while the computer is idle.

First a remainder. The Linrad archives are searchable with Google,
Yahoo, Bing and other search engines but with a rather long delay.
(Containing the word LINRADDARNIL to help limiting search results.)

ACPI (read Wikipedia) is a standard for platform independent
hardware discovery, configuration, power management and monitoring.
Modern operating systems and modern hardwares are intended to
follow this standard but it is complicated and the ACPI control
suffers from many bugs that may degrade the performance of a
computer.

I have run into an ACPI problem on my Compaq 6510b laptop
(Intel Centrino Duo). It used to have a good battery life, 
but I use it on batteries only in the winter when I travel 
to a place with mild climate so I did not know when the
problem with high energy consumption started. I just found
the computer half useless when running on battery after
nearly one year on mains power.

I am mainly using Linux and I found that one processor
was running at 2 GHz while the other one was running at
800 MHz with a total power consumption of 30W.

Searching for Compaq 6510b and phrases like "short battery
life" did not help. The problem was the computer because
Microsoft Windows XP also suffered from the same problem.

Under Windows one can not see any process that is running
with high cpu load, but yet the system monitor shows about
90% cpu just as the Linux monitor does when the ACPI bug
is active.

Under Linux there is a process kacpid that clearly is 
responsible. It is a common problem with modern computers
that the kacpid process uses about 90% of one processor
due to a bug. There are presumably many different bugs.
Frequently reported is one related to the fan and the cpu 
temperature.

I have not been able to find any way to kill the kapcid 
process or to disable ACPI. Putting noacpi in the bootloader
(grub) does not help.

Since the problem is related to hardware the cure to it is
to try all possible BIOS settings to try to find one that
does not activate the ACPI bug. This is of course different
for different platforms and for the Compaq 6510b the important
BIOS setting is: "Network Interface Controller (LAN)" DO NOT
DISABLE this one even if you are not using wired Ethernet. 

I had disabled everything I am not using to save energy.
That was the mistake. Other platforms require other special
knowledge on how to avoid the specific ACPI bugs that they 
may suffer from.

Here is the estimation from the power manager of Windows XP:
 LAN      Battery life
enable       2 hours
disable      6 hours
  
Under Linux one can use powertop to get ACPI information:
 LAN       Power
enable      12W
disable     30W

The ACPI bug uses all available CPU time from processor 1.
By doing that it forces the governor (which is ondemand
by default) to set the cpu speed to its maximum at 2 GHz.
By switching to the powersave governor one can limit the speed
to 800 MHz and then the power stays at 20W when the bug
uses up its 90% of the cpu time.
 
Here is some more. For search engines.....
kacpid cpu-hog uses 90% of one processor.
The computer is slow (because one cpu is always busy.)
It has slow response, can not even play a DVD,
a faster computer is needed (seemingly.)
Disable acpi and disable kacpid are things that many with
ACPI problems tried over the years. I do not think that
is possible any more. Excessive power consumption, short
battery life are relevant to laptops. 

It will take a long time until computers having ACPI 
bugs are obsolete. I hope this mail will help others to 
fix ACPI problems faster than I did:-) 

  73

  Leif / SM5BSZ

--

You received this message because you are subscribed to the Google Groups "Linrad" group.
To post to this group, send email to linrad@xxxxxxxxxxxxxxxxx
To unsubscribe from this group, send email to linrad+unsubscribe@xxxxxxxxxxxxxxxxx
For more options, visit this group at http://groups.google.com/group/linrad?hl=en.

LINRADDARNIL