xorg ati driver corruption and overheating

[JerichoKru]

The card in question is an ATi Mobility 4570 on an Asus K50AB.

I'm really confused by this. I first noticed it on Arch: the open source driver causes small horizontal lines that seem to rotate across the screen and temperatures IDLE at 90+C for the card.

If I use catalyst (or windows) I have no issues. Idle temps are the high 60s to low 70s.

I figured it was an issue with Arch, so I moved onto Fedora 15. I didn't even get the chance to install it as the issues started once KMS did. On Arch, I attempted going with KMS disabled and even compiled a Kernel without KMS support. All this did was kill off all acceleration. Except with the kernel without KMS...It looked like I got accel but, I was essentially looking at a very slow slideshow.

If I want Gnome-Shell to work I need accel. The Arch forums might was well be a ghost town as I was, for the most part, the only poster.

I'm hoping someone in the Fedora community might have some idea.

[Fenrin]

Here (http://www.x.org/wiki/RadeonFeature) is a description of "KMS Power Management Options". Maybe it helps.

[JerichoKru]

Here (http://www.x.org/wiki/RadeonFeature) is a description of "KMS Power Management Options". Maybe it helps.

I forgot to mention that I tried that. Even forcing low power mode still has the corruption but, slightly lower temps (if you call high 80sC much lower).


From my attempts to troubleshoot with Arch, I'm guessing its something wrong with the xorg-ati driver, its components, and/or its interaction with Xorg.

Here's my original post in the Arch forums that's still unsolved:
My laptop is an ASUS K50AB with an HD 4570 (mobility). Under Windows or using catalyst, my videocard behaves properly, running with reasonable temperatures even under load. However, since I'm using Gnome 3/Shell I have to use an open source driver (unless something changed recently).

radeon-pci-0100
Adapter: PCI adapter
temp1: +87.0°C

With it running at this temperature and above, I get small bits of "static" appearing all over the screen and I'm slowly killing my 4570.

Also, before you suggest it, I have cleaned out the vents and even replaced the thermal paste.

As you can see here, the other temp sensors are withing safe limits.

acpitz-virtual-0
Adapter: Virtual device
temp1: +65.0°C (crit = +105.0°C)

k10temp-pci-00c3
Adapter: PCI adapter
temp1: +72.8°C (high = +70.0°C)
(crit = +115.0°C, hyst = +110.0°C)

EDIT: If I start without KMS, no overheating (and no acceleration either since Gnome Shell defaults to its gnome2 appearance)

Currently, the temp is at 73C in Windows.

Here's the full thread if anything in there will be of any help: https://bbs.archlinux.org/viewtopic.php?pid=910456

[JerichoKru]

I went and net-installed F15 with a VGA setting instead of KMS. Now, it boots to a black screen.

Not sure what's wrong here since I can't get into console or anything to read the logs. I did see something along the lines of {read} access denied It flew by too fast to read it.


I even tried using KMS, despite my issue this thread was created for, black screen.


I'd try to use a live cd to get into the filesystem....but having BTRFS as the root filesystem makes this an issue.

Any ideas?

[hen770]

I'd try to use a live cd to get into the filesystem....but having BTRFS as the root filesystem makes this an issue.

Any ideas?


Why is that an issue?

Thanks.
Hanan.

[JerichoKru]

Why is that an issue?

Thanks.
Hanan.

Most live cds don't support btrfs

[JerichoKru]

I DL'd a live version of F15 to get a look at the logs...except I can't because they are empty.

The only thing that actually has anything in it is "dracut". "lastlog" has a file size but I am unable to open it in a readable way.

I'm stumped.

[JerichoKru]

I have included a video that (I hope) explains the issue a bit better.

http://www.youtube.com/watch?v=s0-kYZSHRG8

[JerichoKru]

Someone pointed this out to me: https://bugs.freedesktop.org/show_bug.cgi?id=31000

Importance: medium minor

If it is this bug, then I definitely wouldn't class it like that if it has the potential to kill hardware.