since the release of opensuse 12.3 I have heavy issues with my radeon 6320 graphics-card (AMD E-450 APU). The default radeon driver crashes the computer when rendering in blender. Apparently I cannot help with a bug report.
After having swapped to AMDs fglrx driver, the multi-monitur setup I use and hibernating is simply a pain. So both options aren’t satisfying in any way.
As I haven’t had any issues with the radeon driver includes in opensuse 12.2 for nine month I think about downgrading my system at least when it comes to the graphics-stack. Is it possible to have the same constellation running than in the previous opensuse release?
So if I had a video card that did not work with openSUSE 12.3, I would either try a newer kernel than 3.7 or consider going back to 12.2, at least until I could buy a new video card. For newer kernels, have a look here:
thank you for your quick reply. I upgraded the kernel to 3.9.something (from the opensuse repo) and uninstalled catalyst (what a pain - but it makes me somewhat proud!). After two hours of work, radeon is back with all its benefits (responsiveness, multi-monitor, etc).
I did a quick test render in blender and could finish the job. I mark this issue as solved now. In case I do get driver problems in the future, I will start a new thread.
I am very happy to hear a newer kernel did the trick. Better open source support for Radeon should be on the upswing due to AMD releasing some of its older stuff to open source. And of course, an added bug or two could have existed with kernel 3.7. Since it is so easy to upgrade to a newer kernel, I try it for most any issues AND it can work the other way, up or down in version depending on the issue at hand.
It is nice to see that his problem with blender appears to be fixed… admittedly, my initial bet would have been on the issue laying elsewhere in the chain – given blender freezing, I was thinking the gallium driver.
well, it seems as if I’ve judged the new kernel a bit prematurely. Just a minute ago, my computer shut down immediately, when I was decoding a printer-ready PDF (with Evince) with loads of graphics. This is exactly the same issue, I had with kernel 3.7 before switching to the fglrx-driver from AMD.
@Tyler_K: I don’t believe this is a flash- or blender-only thing. I first thought so, and could reduce the number of system-crashes by disabling the hw-acceleration. But it also appears when using blender or evince (as it just happened). I guess there the radeon-driver doesn’t cope perfectly well with the graphics-card. There might have been a change from kernel 3.4 (opensuse 12.2) to 3.7 (opensuse 12.3) that simply doesn’t do well with my card.
Can I chang the gallium driver, as you suggested? Or do I better downgrade directly to kernel 3.4?
Is there a possibility your CPU or graphic card chipset is over heating? When was the last complete PC cleaning you had? You should do one at least once a year to remove all dust from heat sinks and fans, make sure all fans work and to reseat all memory modules and cables. Do not dismiss the effect duct can have on any PC six months or older in age.
Yes, you can … and while you can cherry pick some of the bits of the graphics stack, in some cases, new features in a new version of component Y also require use of a new component Z and so forth … so, I’d suggest you just draw in the whole shebang from the appropriate repo from Index of /repositories/X11:/XOrg … it will update all your Xorg and mesa stuff (the Xserver, the ddx, libdrm, mesa, libGL blah blah blah …).
As always, in the case of venturing off the beaten track, or even staying on it for that matter: caveat emptor. lol!
That’s a fair point from another perspective: I don’t know what improvements have come about recently, but I recall that the power management in the OSS drivers for the APUs not being the greatest – to the extent that graphics performance suffers as the adapters have to be clocked way down. Though, that would typically manifest itself as crappy performance, as opposed to outright freezing … (unless things are getting really – hot hot hot )
hmm, okay … evince kinda throws a wrench into my thoughts of there being a 3D/GL problem. So yep, could very well be a regression you’re running up against, given it used to work well for you in prior releases.
I think the truth is somewhere right in the middle. Even though I initially found it a stupid idea to look after the temperature, I did so after some rethinking, and so it is: At idle, I get up to 80°C, under load it’s far beyond 90°C. And yes, the computer is clean - in my opinion.
Three things I did so far:
Installed lm-sensors and wrote a little bash-script to track the temperature of the CPU
#!/bin/bash
while true
do
date >> ~/temperature_history.txt
date
sensors >> ~/temperature_history.txt
echo "************************************************************************************************" >> ~/temperature_history.txt
sleep 2
done
This writes the CPU temperature every two seconds in a file - in hope that I can get a quite valuable log the next time the system crashes to fill a proper bug report.
I upgraded the whole Xorg and Mesa stuff as you proposed Tyler_K. No problems at all. System boots, runs, goes to sleep, as expected. Temperature is still very hot (up to 95°C, whereas 90°C is critical according to some of my googling), but haven’t got any crash so far
I tried hard to thrash the computer. Ran blender over an hour, watched videos (what a hard work!), did some other graphics-stuff. So far the system runs stable for a couple of hours. This is far better than yesterday, when the system crashed twice in about two hours.
It would be interesting to see if this shortens the battery life. I’ll have a look in the next couple of days and try to have a guess.
Thank you so far for you help!
Simon
PS When I got the computer about a year ago, I had trouble with it, and somewhere in the opensuse documentation I found the advice to switch of the C6-states directly in BIOS. I’ve never switched it back. Can this be part of the trouble?
Hm… well so far I have one little glitch. In the KDE-system-settings I have only the resolutions “1024x768”, “800x600” and “640x480” when using two screens. The default resolution when starting X is actually the preferred one (1680x1024 and 1366x768), but returning from a full-screen application resets the resolution to “1024x768” only.
Had a look into the X-settings. All configuration files are commented out. I added a few lines with options from “cvt”, but this doesn’t seem to change anything.
I reenabled C6-states in BIOS, and cannot see any difference. I think to remember, that I had problems with suspend2ram/disk in earlier kernel-releases and someone had the idea of disabling it. Maybe this issue has been fixed in the meantime.
Yes, I’m following the other threat. Maybe it’s a coincidence, that two threats about the same subject have been started with just so little time in between. It’s interesting to see all these information falling together. Hopefully we can get some stuff sorted out and help improving suse/linux till the next release.
I don’t know what you mean by “the missing warning at boot”
KDE-settings still list just the three resolutions possible.
Xrandr returns the expecting resolutions for both displays.
You should be able to manually change it with xrandr then if that is the case, however, just so as I’m clear, what is the terminal output from
xrandr
In addition, what is the output of
grep radeon /etc/modprobe.d/*
… checking to see whether you still have the requisite boot feature edits left on from your “Adventures in Fglrxland”
BTW: Temperatures are now safely under 70°C when idle (about 10% load by KDE, IMAP, Xorg, Firefox etc).
I suppose that’s a measure of improvement
Well, I got a short message like “vga=… is depreciated”. This is now of course not the the case anymore. Therefore I have the plymouth splash screen back!
xrandr gives the following output:
Screen 0: minimum 320 x 200, current 1366 x 768, maximum 8192 x 8192
LVDS connected primary 1366x768+0+0 (normal left inverted right x axis y axis) 270mm x 150mm
1366x768 60.0*+
1368x768_32.00 32.0
1280x720 59.9
1152x768 59.8
1024x768 59.9
800x600 59.9
848x480 59.7
720x480 59.7
640x480 59.4
HDMI-0 disconnected (normal left inverted right x axis y axis)
VGA-0 disconnected (normal left inverted right x axis y axis)
Right now, I just have no external monitor I could connect, but the output of xrandx is correct. Funny enough, KDE-settings now show the correct display resolutions. Today in the morning - with an external monitor connected - it didn’t.
Now, from time to time on boot I have a kernel panic, saying that CPU(0) is locked. Will there be a log-file that I can copy-paste or do I have to jot it down the next time it occurs?
Please don’t remind me of fglrx. I actually got this computer with fglrx preinstalled. On a presentation at university I had to restart the X-server to get the external monitor running - how embarrassing! Then the computer froze at random when using the trackpad.
I then opted for a fresh install as I wanted to have a new harddisk as well. This was over a year ago and since then lived happily with opensuse 12.1 and 12.2. Just till the heating problem in 12.3 struck me.
