[ltp] thinkpad r50 problems

[Uwe Walter]

On So, 2004-06-27 at 14:52, Frederik Wagner wrote:
> > happened to get the problem that the fan doesn't stop spinning. i resea=
rched=20
> > on the net and people reported the fan with the same model under window=
s xp=20
> > hard spins. i have tried both ACPI and APM, but either improved the=20
> > situation. certianly this is just my guess, perhpas power mangement hav=
e no=20
> > relevence to fan at all. does any one happen to share the problem?

The problem is well-known but I do not know of a good solution yet
(everybody, please enlighten me! ;-)).

I have a T40p, Linux 2.6, APM, ATI's fglrx driver, speedstep running.

Even when idle, the fan starts spinning after some time and does not
really shutdown very often.


But, as Nils wrote in a side-message:

On So, 2004-06-27 at 17:16, pub@wigginghaus.de wrote:=20
> Hey Chris,
> I have the same machine, same problem. But after installing ATIs fglrx-dr=
ivers=20
> on Debian Unstable and 2.6.7 the fan now sometimes stops under LInux as i=
t=20

Sometimes, my fan does also stop!

My personal opinion is that IBM uses too conservative temperature
thresholds in its APM BIOS. I feel, the fan starts too early and shuts
down too late. The CPU should easily be able to withstand temperatures,
which are a few degrees higher.

I just get this assumption, because, when I cool my system really down
(long time idle and HD standby), the fan does indeed stop.


Windows XP should be able to control the fan itself (using ACPI) and
circumvent the APM BIOS thresholds.

(The same should be possible for Linux ACPI using the fan module, but I
have absolutely no experience with that.)


Another hint is, that people using ACPI with Linux, report that the CPU
temperatures are really low, while the fan still keeps running (around
35=B0C).


And a final problem with APM on Linux is that the CPU can not enter the
power-saving C3 state, which is only possible by using ACPI (and
resulting in some other problems). This is also part of the reason, why
APM-Linux-Users do not reach the same battery runtime as ACPI Windows
users. And shorter runtime, means more power drain, which should also
result in more heat (back to the topic ;-)).


More user reports are available anywhere on the web (or the list
archive):


For example, see
http://cgi.zdnet.de/forum/viewtopic.php?t=3D25
(unfortunately in german, but as I said, similar fan noise discussions
should be available in the list archive or via google...)


Some people even change the fan:
http://cgi.cse.unsw.edu.au/~jmd000/r50/
But this is of course a radical "solution"...



I can only suggest to keep the laptop as cool as possible:

- Use speedstep a lot.
- Shutdown the HD (using the laptop mode integrated since 2.6.6)
- Disable the WiFi card if it is IBM's atheros model!
Only the most current cvs madwifi (WPA) drivers began to enable some
powersaving features. Before, the fan would definitely start if the card
was active and it would burn quite some power.
- I can't judge, if ATI's fglrx or the Open Source X11 is more
power-conservative. We have side-threads in the list currently,
discussing this topic. I remember from some time ago, it was suggested
to use ATI's driver to save power (and heat of course).
- Watch your system load. If possible, disable as much as possible
(e.g. for me tpb burnt a lot of cycles with its default nvram polling
interval, so I set it to a much bigger value, saving 10% of CPU load)

And, if possible, you might consider the following:

* Report the problem to IBM. Complain!
They say, the are aware of the problem and are currently resolving it
(see side-message), but this "currently" is getting longer and longer...


Just my 2c.

More insights always appreciated.


Greetings, UW(e)