The AMD/ATI Bar & Grill

Vi0L0 · 2010-11-27 11:30:11

McFlow wrote:

Edit: Still I am a bit confused which is the "best" choice to rebuild the module after a kernel update?

It depends... hook/daemon are more userfriendly - they will build fglrx module automagically so no user ingerention is required.
With generator you will have to type "catalyst_build_module some_new_kernel_version" manually after kernel's update (but before booting new kernel otherwise you will see tty login screen) - so its not such userfriendly but has got other features:
- it builds more appropriate module because of using makepkg.conf (although it's hard to tell are there any differences in funcionality because of that, i mean it should be better in theory cuz of using makepkg flags - but i can not see any difference )
- it's more secure because whole module building operation can be done by not priviliged user.

Choose what you want - it's that easy.

Anyhow: if you don't care about automatic re-compilation, you are using AUR and you were just rebuilding 'catalyst' package with every kernel update - i think it's better to use generator because it's easier to use, it's building same module as 'catalyst' and it's taking much less hdd space - after one generator installation (it's only 4MiB fat) you can remove ati's installation stuff (~120MiB)

Cdh · 2010-11-27 12:17:48

There is a bug with privilegue dropping in catalyst-generator

 ~ % LC_ALL=C sudo catalyst_build_module all
Please specify your (unprivileged) user name:
chris

Building catalyst-2.6.36-ARCH package for 2.6.36-ARCH kernel ...
--------
chown: invalid user: `dreemtul:video'
su: user dreemtul does not exist

Building catalyst-2.6.36-pf package for 2.6.36-pf kernel ...
--------
chown: invalid user: `dreemtul:video'
su: user dreemtul does not exist

Vi0L0 wrote:

its always good to look at winehq tips: http://appdb.winehq.org/objectManager.p … n&iId=2890

I tried with backbuffer too but then only everything game is upside down and it still freezes like with fbo.

Vi0L0 wrote:

i thought that my pc will blow away after starting black ops under linux, but no - its almost fine on highest details (hd4850)

Yes, half life 2 runs very performant too - until it freezes.

I also have also done "cp /etc/ati/amdpcsdb.default /etc/ati/amdpcsd" with no X running so now I am using standard config from ati driver...

Vi0L0 · 2010-11-27 15:17:25

Cdh wrote:

There is a bug with privilegue dropping in catalyst-generator

Thanks, should be fixed in catalyst-generator 10.11-2

+ thanks to dickeny catalyst-generator is now using sudo (if it's present) to install generated package, so if sudo is present you will have to type your user password instead of root's pass.

snack · 2010-11-27 20:23:57

I installed the new catalyst-10.11-2 from [catalyst] and the excessive power consumption with kernel 2.6.36 seems to be gone away.
Thanks Vi0l0!

Cdh · 2010-11-28 10:32:03

Ok, this is a proper recording: http://dl.dropbox.com/u/13721560/hl2-3.mkv
It happens every time i look along the canal.

I now use a geforce 8800gt and there it works without any problem. Seems to run a little bit smoother with nvidia too. I think amd has finally convinced me to not use catalyst anymore. They really tried hard with making games finally completely unplayable and such.

dcc24 · 2010-11-28 19:24:25

Hi guys. After a long long time, I'm back to using catalyst due to massive heating problems with the radeon driver. I'm using catalyst-utils-10.11-1 and xorg-server-1.9.2.-2. The problem is, whenever I try to view a fullscreen flash video, the entire display gets corrupted. 99% of the screen becomes black, there is only a 5 - 10px high corrupted (though visible) area on the very top section of the screen. The system is still responsive though, I can - without any visual help - open up a terminal and execute commands. Even killing X (with "sudo killall X") doesn't help, tty's are corrupted as well. Only a hard reboot fixes it.

Any idea what might be wrong?

jaydoc · 2010-11-29 19:52:58

Hi

I am building a new rig, with an ASUS mobo that has an intergrated Radeon 4250 graphics card. What is the latest on this card as far as driver support under Arch is concerned..? Thanks.

Revolt · 2010-11-30 13:16:39

Ever since the kernel 2.6.36 upgrade I could no longer resume from hibernate (the first time always works, the second leaves a black screen). Tried every possible quirk to no success. Even tried using tuxonice instead of uswsusp.

Reverted back to xf86-video-ati (don't really need the extra 3D and opensource powersave works ok).

Wondering if anyone else experienced this problem and/or if it is a known bug.

LeCrayonVert · 2010-11-30 13:32:32

Revolt wrote:

Ever since the kernel 2.6.36 upgrade I could no longer resume from hibernate (the first time always works, the second leaves a black screen). Tried every possible quirk to no success. Even tried using tuxonice instead of uswsusp.
Reverted back to xf86-video-ati (don't really need the extra 3D and opensource powersave works ok).
Wondering if anyone else experienced this problem and/or if it is a known bug.

Have you got the same issue with suspend to ram ?

Revolt · 2010-12-01 02:17:42

Both suspend to ram and suspend to disk. On the 2nd resume either the screen stays off (as in no backlight) or it is turned on but entirely black.

foutrelis · 2010-12-01 06:55:42

jaydoc wrote:

I am building a new rig, with an ASUS mobo that has an intergrated Radeon 4250 graphics card. What is the latest on this card as far as driver support under Arch is concerned..? Thanks.

Radeon HD 4200 here, on both desktop and laptop.

The Catalyst driver works quite well, and Vi0l0 has done a great job with the unofficial [catalyst] repo.

There's also the open source xf86-video-ati driver which also works well. I have used it in the past, but switched to Catalyst for better power management (maybe this is fixed now?) and hardware video acceleration (with xvba-video and mplayer-vaapi).

o1911 · 2010-12-01 13:06:03

Revolt wrote:

Ever since the kernel 2.6.36 upgrade I could no longer resume from hibernate (the first time always works, the second leaves a black screen). Tried every possible quirk to no success. Even tried using tuxonice instead of uswsusp.
Reverted back to xf86-video-ati (don't really need the extra 3D and opensource powersave works ok).
Wondering if anyone else experienced this problem and/or if it is a known bug.

I have noticed this also. The first hibernate or suspend works as expected, but the second leaves a blank screen. ssh still worked so I rebooted that way. It is mighty tiring how something like this can be an ongoing problem. Slowly but steadily the AMD driver is getting better, though.

LeCrayonVert · 2010-12-02 13:37:40

I think this hibernate/suspend issue is more likely related to the kernel upgrade. Have you tried 10.11 driver with the previous kernel (2.6.35) ?

o1911 · 2010-12-02 13:41:30

LeCrayonVert wrote:

I think this hibernate/suspend issue is more likely related to the kernel upgrade. Have you tried 10.11 driver with the previous kernel (2.6.35) ?

I have not, and I've since done a pacman -Scc so I cannot be bothered at this point It isn't a large concern to me, but it is nice when things "just work".

Revolt · 2010-12-02 22:30:50

LeCrayonVert wrote:

I think this hibernate/suspend issue is more likely related to the kernel upgrade. Have you tried 10.11 driver with the previous kernel (2.6.35) ?

Yep it worked with the previous kernel. But it is some kind of incompatibility between the two of them since suspend works with xf86-video-ati.

EDIT: Double confirmation. Reverted back to the previous kernel since the powersaving on the open source drivers seemed to be reset on resume. Suspension/Hibernation works fine on 2.6.35.

Last edited by Revolt (2010-12-03 18:41:59)

mouse256 · 2010-12-04 17:10:45

I'm also experiencing issues with standby. If I close my lid, the screen goes to black (but the backlight stays on). Opening the lid does not react, the system is not usable anymore.

LeCrayonVert · 2010-12-04 17:44:19

Revolt wrote:

LeCrayonVert wrote:
I think this hibernate/suspend issue is more likely related to the kernel upgrade. Have you tried 10.11 driver with the previous kernel (2.6.35) ?
Yep it worked with the previous kernel. But it is some kind of incompatibility between the two of them since suspend works with xf86-video-ati.
EDIT: Double confirmation. Reverted back to the previous kernel since the powersaving on the open source drivers seemed to be reset on resume. Suspension/Hibernation works fine on 2.6.35.

Have you tried also 10.10 with 2.6.36 ?

Revolt · 2010-12-05 01:15:43

Nope I didn't. And don't have the time to experiment at the moment

snack · 2010-12-05 10:27:07

My problems with suspend to ram began far before kernel 2.5.36. I can suspend closing the lid or with pm-suspend; everything works but the laptop is subsequently rebooted instead of being resumed. I'm not sure when this problem appeared (I don't use suspend to ram very often) but I'm sure that 2.6.35 with catalyst 10.10 and 10.9 didn't work. I hoped that 2.6.36 or 10.11 would fix this, but I was dreaming...
BTW, my laptop is a Sony Vaio VGN-SR21M with a Mobility HD3470.

txus · 2010-12-05 11:32:23

Yeah, I have similar issues with suspend (also with a Vaio VGN-SR55X), but I don't think they are catalyst-related, because it also happens with the radeon driver.

I also do not suspend that often, so I don't know what broke it, but it has been like that at least since August. I have tried both fglrx and radeon, and the result is the same: after suspend, the laptop just doesn't wake up, and the hdd led keeps blinking very fast (plus the hdd does some strange clicks from time to time).

I also tried other distros (fedora and ubuntu, they are quickest to set up) to see if the behaviour is the same. Results:
Arch somewhere around March: WORKED, as far as I remember
Arch at least from August on: laptop doesn't resume
Fedora 14: same issues
Ubuntu 10.10: same issues
Ubuntu 10.04: WORKS!

And here is the weirdest thing: after booting Ubuntu 10.04, when I REBOOT (not turn off & turn on, then it doesn't work) to Arch, suspend works again!

I'll probably start a thread, but seeing you have similar issues I just wanted to share it here also.

Fuzzo · 2010-12-07 08:40:24

When I start blender-svn on Arch64 with my HD 4350, I get a black screen and corrupted (mostly invisible) console fonts.

When I start it with -noglsl, I still get a black screen on my X server, but the console still works and I can restart X.

Using catalyst-daemon 10.11-2 from [catalyst], though the problem was there with the AUR catalyst packages as well.

Does anyone have any ideas?

Kariddi · 2010-12-07 17:29:26

Hmmm, I have an alarming problem with these latest drivers on a 5850 ...

What happens is that sometimes , when I stay in the pure linux console (CTRL+ALT+FX) the video card overheats, but the fan doesn't rise its speed. I noticed it today and yesterday. I used the pure linux console to clean the system from unused packages and then I rebooted the system from the console (X was running in background) , as I rebooted the the video card fan started to spin at almost full speed and a lot of heat was coming out from behind my PC. I stayed only a little while in the console before I rebooted, but who knows what would have happened if I stayed more!

Beware that there could be this problem ... I'm still not sure of the exact actions I have to perform to reproduce this, so if this doesn't happen for you maybe there's something specific I do that triggers the problem.
I also use a kernel26-ck . Maybe this is related?

mouse256 · 2010-12-07 17:38:34

The suspend not working seems to be unrelated to catalyst, it's a kernel bug: https://bbs.archlinux.org/viewtopic.php?id=109620
Adding the following to your boot line fixed it for me: (dell studio 1555 with ati 4570)

intel_idle.max_cstate=0

Revolt · 2010-12-08 11:42:47

mouse256 wrote:

The suspend not working seems to be unrelated to catalyst, it's a kernel bug: https://bbs.archlinux.org/viewtopic.php?id=109620
Adding the following to your boot line fixed it for me: (dell studio 1555 with ati 4570)
intel_idle.max_cstate=0

Didn't work for me (that issue seems to be related to 2.6.35 and only have problems with 2.6.36).

mouse256 · 2010-12-08 11:59:21

Revolt: You seem indeed to have a different problem as I did.

You: first suspend works, 2nd leave the screen turned off
Me: first suspend already has the problem: screen is on but completely black.

Arch Linux

#2676 2010-11-27 11:30:11

Re: The AMD/ATI Bar & Grill

#2677 2010-11-27 12:17:48

Re: The AMD/ATI Bar & Grill

#2678 2010-11-27 15:17:25

Re: The AMD/ATI Bar & Grill

#2679 2010-11-27 20:23:57

Re: The AMD/ATI Bar & Grill

#2680 2010-11-28 10:32:03

Re: The AMD/ATI Bar & Grill

#2681 2010-11-28 19:24:25

Re: The AMD/ATI Bar & Grill

#2682 2010-11-29 19:52:58

Re: The AMD/ATI Bar & Grill

#2683 2010-11-30 13:16:39

Re: The AMD/ATI Bar & Grill

#2684 2010-11-30 13:32:32

Re: The AMD/ATI Bar & Grill

#2685 2010-12-01 02:17:42

Re: The AMD/ATI Bar & Grill

#2686 2010-12-01 06:55:42

Re: The AMD/ATI Bar & Grill

#2687 2010-12-01 13:06:03

Re: The AMD/ATI Bar & Grill

#2688 2010-12-02 13:37:40

Re: The AMD/ATI Bar & Grill

#2689 2010-12-02 13:41:30

Re: The AMD/ATI Bar & Grill

#2690 2010-12-02 22:30:50

Re: The AMD/ATI Bar & Grill

#2691 2010-12-04 17:10:45

Re: The AMD/ATI Bar & Grill

#2692 2010-12-04 17:44:19

Re: The AMD/ATI Bar & Grill

#2693 2010-12-05 01:15:43

Re: The AMD/ATI Bar & Grill

#2694 2010-12-05 10:27:07

Re: The AMD/ATI Bar & Grill

#2695 2010-12-05 11:32:23

Re: The AMD/ATI Bar & Grill

#2696 2010-12-07 08:40:24

Re: The AMD/ATI Bar & Grill

#2697 2010-12-07 17:29:26

Re: The AMD/ATI Bar & Grill

#2698 2010-12-07 17:38:34

Re: The AMD/ATI Bar & Grill

#2699 2010-12-08 11:42:47

Re: The AMD/ATI Bar & Grill

#2700 2010-12-08 11:59:21

Re: The AMD/ATI Bar & Grill

Board footer