ATI Mobility Radeon HD4670 Overheating

[asfalatus]

Greetings,

anything new about this? Did you tried any recent drivers?

I'm running Arch Linux and I can't continue with rolling updates because catalyst doesn't support 4xxx cards no more.
So I want to move to OSS drivers, but I don't want to get this "overheat feature" again (cause summer is here which means >30 dgs of C. temperatures around).

[123_qwe]

http://article.gmane.org/gmane.comp.video.dri.devel/69915

Alex is working on advanced power management, so I'm not so deep into it, but
AFAIK this should be improving in the near future.

At least its coming

[pandev92]

Mhh I have the same problem with hp pavilion dv6 2120es, My laptop get 92 degrees celsius...

[Morgan00]

overheating...

Hi,

I have a laptop studio xps 1647 with HD4670 mobility.
With the recent drop of HD2000/3000/4000 ati cards by catalyst, I tested the last open source ati drivers with xorg 1.12 and kernel 3.4. (from xorg-edgers ppa)
I was initially surprised by the speed of my 3D desktop gnome and the fluidity of my HD movies with VLC! Really better than Catalyst. But then I realized that my graphics card was very hot. There are over 10 degrees Celsius more on idle than with the Catalyst drivers. And it rises very quickly when I use the 3D.

Yet I configured my xorg.conf with this:

http://www.thinkwiki.org/wiki/Additional_options_for_the_radeon_driver

And I adopted the "low profile" like this:

https://wiki.archlinux.org/index.php/ATI#With_KMS_enabled

But that does not change anything. Any ideas ??

How is your experience with radeon drivers on a laptop ???

[asfalatus]

I'm surprised low profile didn't make a difference.

It did. But it was not very effective. Not like fglrx. As I said, with fglrx case remains cool. With opensource drivers case was hot permanently.

[bridgman]

I tried to use "dynpm" or "profile" without success (also forcing low power profile did'nt help). Laptop case was hot.

I'm surprised low profile didn't make a difference. My recollection was that "low" was available on pretty much all systems.

[Psycho Game]

Okay, thank you for you're reply.
Hopefully some other people who have the same problem with the open-source driver will leave a comment on this topic.
If this is the case, I will file a bug report on this problem at the driver developers page, if it is not already filed.
At this point I hope some other persons also are willing to leave a reply when having the same problems, while in the meantime I will try to find a bug report on this.
If it already exists I will put a link on the forum, so you can also subscribe yourself to the bug report.

Greetings Jasper

[asfalatus]

Greetings,

I have Dell XPS 16 1640 with same card (ATi Mobility Radeon HD 4670). When I am using fglrx everything is fine. When I used opensource radeon driver my GPU started to overheat, even if fan was constantly running. I tried to use "dynpm" or "profile" without success (also forcing low power profile did'nt help). Laptop case was hot.

With fglrx everything remains fairly cool. So it isn't HW or case design problem.

[Psycho Game]

Dear Kano,

I'm currently unable to test the case you pointed out.
At the moment i'm running Gnome Shell 3 so I'm affected by the well known bug in the proprietary driver.
Of course I will test this case as soon as the Catalyst 11.9 driver comes out.

Also thank you bridgman for pointing out exactly how the open-source drivers function.
That explains why the computer after a restart defaults to the high profile, because that's his boot standard.
But what's still strange is that the graphics processor overheats on the high profile, which it shouldn't do is what I believe.
I base this on the fact that the hardware is designed to be capable of running at these processor and mem. clock frequencies according to the card specs.
So I think Kano may be right that it may be badly designed.
My laptop is manufactured by MSI, and also when using GPU-z under windows the information states that the card is manufactured by MSI.
Another thing I can think of is that the fan control is not working properly. Is this normally done through monitoring thermal sensors?
Maybe someone can share a light on how fan control is realized in the open-source drivers.
How can I check if the thermal sensors are giving correct readings?

Greetings Jasper

Ps. I will report back what the outcome of the test case is which was pointed out by Kano.

[bridgman]

The proprietary drivers cycle the clocks and voltages up and down depending on the amount of activity on your display, and would normally only run at full power when doing things like 3D gaming. The open drivers, on the other hand, currently default to whatever clock/voltage settings were established by the VBIOS and stay there unless you explicitly enable one of the power management schemes, referred to as "dynpm" or "profile".

The dynpm option is closest to what the proprietary drivers do, but that option only works on a subset of systems. The profile option is the most commonly used, but you need to manually choose a power level from the options in your BIOS. If there is a "mid" option that works pretty well for most people.

(see KMS Power Management section of RadeonFeature page for more details) : http://www.x.org/wiki/RadeonFeature

Note that power settings are controlled by writing to a system device, not via the X config file.