[ltp] Z61p GPU Thermal Issue

Brian D. Ropers-Huilman linux-thinkpad@linux-thinkpad.org
Fri, 31 Aug 2007 09:43:19 -0500


On 8/31/07, Richard Neill <rn214@hermes.cam.ac.uk> wrote:
> 128 C is *hot*. Can you actually feel it?

Yes, it is. I certainly feel heat on the bottom of the laptop, verging
on /hot/ at times, but it's never been too hot for me to keep the
machine on my lap.

> If the GPU hits 128 degrees very suddenly, ...

Well, it's not hitting it suddenly. See comments below.

> but if the GPU sits at over 100 degrees for a few minutes, there's no
> way you could keep the laptop on your knee.

On a resume from suspend-to-memory, the GPU temperature fairly quickly
rises to around 110 C and stays there. As I'd mentioned in the
original email, I can specifically cause an increase in the GPU
temperature by grabbing a "busy" window and rapidly dragging it around
the screen (I have my window manager set to show the contents of the
window on move or resize). This can push the GPU temperature up to 120
C.

I have not run the fglrx driver and closely monitored the GPU
temperatures (the laptop is my secondary machine and is at home all
day while I use my deskside machine here at work). I'll see if I can
make some time this weekend to do some tests, specifically trying to
quantify:

1) initial GPU temperatures on-boot after the machine has been off for
many hours (cold start)
2) GPU temperature profile under:
2a) vesa
2b) fglrx
2c) the new avivo driver (though when I tried this before, the display
quickly deteriorated [I know of no better term for this -- there were
many graphical artifacts and distortions on the screen] and eventually
the machine hard-locked)
3) results of initial work with GPU power (and hopefully thermal)
management under fglrx

I'll provide this to the list when I have the data.

> It's possible you have a broken sensor instead of a broken GPU.

I've wondered this, but my limited memory of my limited monitoring of
the temperature at least indicate the sensor is consistent, though it
may very well be completely wrong.

> Also, is the fan running as it should?

It seems to be. I took a wee detour down this road and used some of
the scripts from the thinkwiki pages RE: the fan and all seemed to be
fine.

> As for solutions, have you ever played with Liquid Nitrogen? ;-)

lol, yes, actually, I have (I worked for many years at the University
of Wisconsin's Center for Applied Microelectronics [WCAM] and often
had to go refill our dewars of liquid nitrogen).

> Or, http://www.phys.ncku.edu.tw/~htsu/humor/fry_egg.html

Another lol. :)

-- 
Brian D. Ropers-Huilman