The AMD/ATI Bar & Grill

mouse256 · 2011-11-02 15:07:31

The opensource driver works rather well on my mobility HD 4570 (dell studio 1555). Only issue I had with it is that powersave did not reduce the number of pci lanes used, hence resulting in a way higher battery usage. And that's exactly the reason why I still use the catalyst driver.
Don't know if this has been fixed in the meanwhile, I should test it again...

koshon · 2011-11-03 15:44:15

damacas wrote:

I upgraded from 11.8 to 11.10 and found that X crashed every time I resized konsole, in KDE 4.7.2. Downgraded to 11.8 and all is good again.

Yepp, i also have crashes on 11.10 and i haven't this on 11.9. I probably would have use the 86-video-ati, but on my core i3, when i watching video i have load CPU more than 50% . On catalyst i have 10-15%.

Last edited by koshon (2011-11-03 15:44:44)

Unia · 2011-11-03 15:48:46

Can anyone other van Vi0L0 tell me how it is on Gnome Shell? On different sources I read that it finally works, although Vi0L0 claims it doesn't. (and since he's almost always correct on these things...)

Last edited by Unia (2011-11-03 15:49:08)

achilleas.k · 2011-11-03 16:06:36

Unia wrote:

Can anyone other van Vi0L0 tell me how it is on Gnome Shell? On different sources I read that it finally works, although Vi0L0 claims it doesn't. (and since he's almost always correct on these things...)

Haven't tested it extensively, but for me any Catalyst, even 11.10 is more or less unusable (tearing and flickering).
Used it for about 30 mins before reverting back to Radeon.

VGA: HD5730
Lenovo Ideapad Y560

EDIT: The above experience relates to Gnome Shell only

Last edited by achilleas.k (2011-11-03 16:10:42)

Weegee · 2011-11-03 17:32:46

Unia wrote:

Can anyone other van Vi0L0 tell me how it is on Gnome Shell? On different sources I read that it finally works, although Vi0L0 claims it doesn't. (and since he's almost always correct on these things...)

Tried it today.

It was indeed better than I expected, however you shouldn't consider it to be stable. You still have crashes, graphical bugs now and then, tooltips are very likely to be unreadable (however this isn't always the case) and the performance of games is still horrible when GNOME Shell runs in the background. That's why I'll stick with Xfce.

nemolivier · 2011-11-03 23:57:21

nemolivier wrote:

logomancer wrote:
nemolivier wrote:
Ok, so I upgraded from 11.9 (running as well as possible) to 11.10 from catalyst repo, I use the xorg110 repo
[…]
But :
— when I open my session (on « Gnome ») I don’t get gnome-shell
— fglrxinfo give me :
sudo fglrxinfo 1 ↵
 X Error of failed request:  BadRequest (invalid request code or no such operation)
   Major opcode of failed request:  136 (GLX)
   Minor opcode of failed request:  19 (X_GLXQueryServerString)
   Serial number of failed request:  15
   Current serial number in output stream:  15
[…]
Any idear where I can search for the problem ?
Thank’s
[EDIT] wrong version number
I'm getting the same problem too. I reboot, and dmesg tells me that the CMMQS handle is invalid for fglrx. I'll post an example when I'm at that machine again.
This seems to be an issue for me with 11.9 as well. I had to downgrade to 11.8 to get it to work again.
OK… after looking at my Xorg log (sorry for not doing this before), I’ve found this :
Kernel Module Version Information:
[    19.946] (II) fglrx(0):     Name: fglrx
[    19.946] (II) fglrx(0):     Version: 8.89.4
[    19.946] (II) fglrx(0):     Date: Sep 14 2011
[    19.946] (II) fglrx(0):     Desc: ATI FireGL DRM kernel module
[    19.946] (WW) fglrx(0): Kernel Module version does *not* match driver.
[    19.946] (EE) fglrx(0): incompatible kernel module detected - HW accelerated OpenGL will not work
[    19.946] (II) fglrx(0): [uki] removed 1 reserved context for kernel
And in my dmesg :
dmesg| grep fglrx    
[    0.919888] fglrx: module license 'Proprietary. (C) 2002 - ATI Technologies, Starnberg, GERMANY' taints kernel.
[    0.998516] [fglrx] Maximum main memory to use for locked dma buffers: 3801 MBytes.
[    0.998763] [fglrx]   vendor: 1002 device: 954f count: 1
[    0.999296] [fglrx] ioport: bar 4, base 0xae00, size: 0x100
[    0.999627] [fglrx] Kernel PAT support is enabled
[    0.999646] [fglrx] module loaded - fglrx 8.89.4 [Sep 14 2011] with 1 minors
[   18.615765] fglrx_pci 0000:01:00.0: irq 46 for MSI/MSI-X
[   18.616312] [fglrx] Firegl kernel thread PID: 1402
[   18.616380] [fglrx] Firegl kernel thread PID: 1403
[   18.616445] [fglrx] Firegl kernel thread PID: 1404
[   18.616594] [fglrx] IRQ 46 Enabled
[   19.946530] [fglrx:firegl_cmmqs_BIOSControl] *ERROR* CMMQS BIOS Control: CMMQS handle is not valid.
[   19.946534] [fglrx:firegl_bios_control] *ERROR* CMMQS BIOS Control is failed: firegl_bios_control
[   21.907488] [fglrx:fireglAsyncioIntEnableMsgHandler] *ERROR* interrupt source ff000066 is not supported on this hardware (return code = 1)
[ 3624.252845] [fglrx:fireglAsyncioIntEnableMsgHandler] *ERROR* interrupt source ff000066 is not supported on this hardware (return code = 1)
[26021.280155] [fglrx:fireglAsyncioIntEnableMsgHandler] *ERROR* interrupt source ff000066 is not supported on this hardware (return code = 1)
[26327.647875] [fglrx:fireglAsyncioIntEnableMsgHandler] *ERROR* interrupt source ff000066 is not supported on this hardware (return code = 1)
So it look’s like the kernel module don’t match the driver. I only have one fglrx module. What driver it-is ? I’ve just update with pacman, and re-compile after the first fail.
What can-I check now ?

Anyone have idears on this ?

Thank’s,
Olivier.

sublimepua · 2011-11-04 02:24:13

Got it. It really needed Xorg 1.10. I noticed that catalyst was much newer than the wiki indicates. Someone should clean that up

DonVla · 2011-11-04 06:35:37

Hi all,

I'm about to return to ATI graphics and buy a new HD5750 card.
However, one of the major reasons I stopped using ATI cards was the non-existent support for suspend-to-disk/-ram (hibernating) by the proprietary driver.
How does this work nowadays?
Does the "normal" kernel suspend work out-of-the-box with the fglrx module?
Is much tweaking needed?

Thank you in advance,
Vlad

kritter · 2011-11-04 07:43:47

I'm using suspend-to-ram almost in an two-hourly manner. This works out of the box with the fglrx driver. When using suspend-to-disk with the TuxOnIce patch, I had to increase the pages variable to a higher number, can't recall it correctly. Or do you mean the hybrid suspend?

reflexing · 2011-11-04 07:47:11

DonVla wrote:

Hi all,
I'm about to return to ATI graphics and buy a new HD5750 card.
However, one of the major reasons I stopped using ATI cards was the non-existent support for suspend-to-disk/-ram (hibernating) by the proprietary driver.
How does this work nowadays?
Does the "normal" kernel suspend work out-of-the-box with the fglrx module?
Is much tweaking needed?
Thank you in advance,
Vlad

I've checked it now on my HD5770, both STR and STD works (never used it before

What doesn't work is MPEG4/AVC video playback with mplayer-vaapi and xv video output (accelerated vaapi playback actually didn't work for me for a long time, it worked maybe on 20% of videos).

Can you guys share your video playback experience? I'm using KDE 4.7.3 with OpenGL backend.

DonVla · 2011-11-04 09:26:20

kritter wrote:

I'm using suspend-to-ram almost in an two-hourly manner. This works out of the box with the fglrx driver. When using suspend-to-disk with the TuxOnIce patch, I had to increase the pages variable to a higher number, can't recall it correctly. Or do you mean the hybrid suspend?

Hello,

I meant the standard "echo disk (mem) > /sys/power/state" kernel suspend functionality - no uswsusp or tuxonice.
Does this one work properly?

ChemBro · 2011-11-04 10:27:04

All I know is, that suspend-to-ram (haven't tried the other one) never worked for me with Catalyst. And I don't use a "tuxonice" thing or whatever it is, only "out-of-the-box".

kritter · 2011-11-04 14:08:42

I meant the standard "echo disk (mem) > /sys/power/state" kernel suspend functionality - no uswsusp or tuxonice.
Does this one work properly?

echo mem > /sys/power/state works like a charm. No broken screen, no artifacts, everything's just fine

DonVla · 2011-11-04 17:03:19

kritter wrote:

I meant the standard "echo disk (mem) > /sys/power/state" kernel suspend functionality - no uswsusp or tuxonice.
Does this one work properly?
echo mem > /sys/power/state works like a charm. No broken screen, no artifacts, everything's just fine

That sounds great. Thank you.
What card do you have? And what about "echo disk > /sys/power/state"?

DonVla · 2011-11-04 19:49:40

Hello,

Do I get it right, the latest fglrx module and catalyst-utils do not work with "xorg-server<1.11.0".
Does this mean I have to downgrade xorg?

PS: If I understand the wiki catalyst 11.10 should work with xserver 1.11: https://wiki.archlinux.org/index.php/AT … repository

Last edited by DonVla (2011-11-04 20:32:43)

ChemBro · 2011-11-04 22:13:04

DonVla wrote:

PS: If I understand the wiki catalyst 11.10 should work with xserver 1.11: https://wiki.archlinux.org/index.php/AT … repository

No, that's only a wikipage with not enough data. You can add it, if you want.

Vi0L0 · 2011-11-05 14:28:54

ChemBro wrote:

DonVla wrote:
PS: If I understand the wiki catalyst 11.10 should work with xserver 1.11: https://wiki.archlinux.org/index.php/AT … repository
No, that's only a wikipage with not enough data. You can add it, if you want.

I've added info on the top of Installation section, it should clarify this case.

Turris · 2011-11-05 15:14:51

Vi0L0 wrote:

ChemBro wrote:
DonVla wrote:
PS: If I understand the wiki catalyst 11.10 should work with xserver 1.11: https://wiki.archlinux.org/index.php/AT … repository
No, that's only a wikipage with not enough data. You can add it, if you want.
I've added info on the top of Installation section, it should clarify this case.

I also added a heading concerning this in the discussion-page btw: https://wiki.archlinux.org/index.php/Ta … structions .

On a different note, catalyst 11.10 caused a problem for me, related to Xinerama. I have a dual-head setup, and on my screen # 2 a big grey box would be visible where I closed Firefox. This didn't happen when I closed Firefox on screen # 1. (xfce4)terminal also displayed some erratic behaviour on screen # 2, not on # 1.

I solved this by following the instructions here (https://wiki.archlinux.org/index.php/AT … inerama.29), thus ditching Xinerama.

nathan28 · 2011-11-06 04:45:09

Just got an AMD HD6310 in an Asus laptop. Suspend to RAM is totally borked with the Catalyst drivers as far as I can tell. Tried every BIOS combination, echo mem > /sys/power/state, adding vga=0, vbetools, uswsusp, etc. In each case I get a blank backlit screen and lose the keyboard entirely. Alt + sys rq + combos don't work, the power key won't work (even when depressed for a long time) and I have to yank AC power and the battery before I can bring anything up. This looks common based on teh interwebz.

pm-suspend log:

Running hook /usr/lib/pm-utils/sleep.d/98video-quirk-db-handler suspend suspend:
ATI Catalyst driver detected, not using quirks.

I saw one forum post saying that if you disable the framebuffer, exit X and/or runlevel 5, remove the flgrx module, you can then safely suspend-to-RAM. Didn't try that, b/c if I have to close X apps and telinit 3 what's the point?

Suspend to disk works fine, almost made me want to install fbsplash.

Otherwise Catalyst support for the card is really good. Dcompmgr and WINE will get tearing and refresh issues/artifacts... [sarcasm]bugs? In WINE? Really?[/sarcasm] Native 3D apps seem good. I haven't tried anything like HDMI or HD video or external monitors. Anyway, I will be quite happy when/if AMD gives us s2ram support.

ico · 2011-11-06 08:59:17

Using an AMD card for the first time with Arch Linux. (HD 6950 + xf86-video-ati)

Was wondering if anyone is getting this message at boot up?

[drm:radeon_dp_i2c_aux_ch] *ERROR* aux i2c too many retries, giving up
[drm:radeon_dp_i2c_aux_ch] *ERROR* aux i2c too many retries, giving up
[drm:radeon_dp_i2c_aux_ch] *ERROR* aux i2c too many retries, giving up
[drm:radeon_dp_i2c_aux_ch] *ERROR* aux i2c too many retries, giving up

Kernel is 3.0.7-1.

Edit:

Everything seems to be working fine though.

Last edited by ico (2011-11-06 09:00:06)

achilleas.k · 2011-11-07 13:42:39

I just read that Gnome shell now works with software rendering.

Does this mean we can use Catalyst drivers and turn on software rendering to avoid tears but still have proprietary 3D acceleration available (not to mention, power management)?

I think I wouldn't mind slower, software rendering for good power and heat management on my laptop.

buntolo · 2011-11-07 15:51:50

Berseker wrote:

AFAIK vesa drivers have just to be installed
$ yaourt -Ss xf86-video-vesa
xorg110/xf86-video-vesa 2.3.0-5 [0,01 M] (xorg-drivers xorg)
X.org vesa video driver
extra/xf86-video-vesa 2.3.0-7 [0,01 M] (xorg-drivers xorg)
X.org vesa video driver
they do not need to be loaded as a kernel module, they are automatically openend by xorg in absence of other drivers

And should I see them loaded as modules?

sublimepua · 2011-11-09 00:09:33

I've got the new A6-3400m apu with the integrated 6520g, and sleep does not work for me at all. pm-utils, tuxonice, and the echo mem command all give my the same situation: the screen goes black with the backlight on and the fans kick up. a white line flickers across the screen until I hold down the power button to cut power. When I hit the power button again, I am left with a black screen. If I hold down the power button, it doesn't shut off. I have to remove the battery and charger to get it to boot back into the bios. On windows, sleep works fine. Rebooting and shutting down works fine too. Suspending to ram and hibernation both do not work. I am using the catalyst driver because the opensource one doesn't support the card yet.

sublimepua · 2011-11-09 00:17:09

pm-suspend just started working. I didn't update anything. This is creepy. Arch forum gods, thank you for your assistance.

goofy03 · 2011-11-09 01:41:20

Hi

Today i discover, that a use old version of flashplugin (flashplugin-prerelease 10.3.162.29-1 on archlinux). Then i have upgraded to standard extra/flashplugin 11.0.1.152-1 but now, no more fullscreen HD videos (test on youtube).
Very very slow i can calculate the framerate with my eyes (1 or 2) . And closing the fullscreen mode is a nightmare. I try beta version too but it's the same thing.

Am I alone with this problem ? any advice ? There is no vdpau lib or something similar for ATI to install ?

I use last version of catlyst total on a HD5870 mobility

Thanks for informations or advice.

Arch Linux

#3326 2011-11-02 15:07:31

Re: The AMD/ATI Bar & Grill

#3327 2011-11-03 15:44:15

Re: The AMD/ATI Bar & Grill

#3328 2011-11-03 15:48:46

Re: The AMD/ATI Bar & Grill

#3329 2011-11-03 16:06:36

Re: The AMD/ATI Bar & Grill

#3330 2011-11-03 17:32:46

Re: The AMD/ATI Bar & Grill

#3331 2011-11-03 23:57:21

Re: The AMD/ATI Bar & Grill

#3332 2011-11-04 02:24:13

Re: The AMD/ATI Bar & Grill

#3333 2011-11-04 06:35:37

Re: The AMD/ATI Bar & Grill

#3334 2011-11-04 07:43:47

Re: The AMD/ATI Bar & Grill

#3335 2011-11-04 07:47:11

Re: The AMD/ATI Bar & Grill

#3336 2011-11-04 09:26:20

Re: The AMD/ATI Bar & Grill

#3337 2011-11-04 10:27:04

Re: The AMD/ATI Bar & Grill

#3338 2011-11-04 14:08:42

Re: The AMD/ATI Bar & Grill

#3339 2011-11-04 17:03:19

Re: The AMD/ATI Bar & Grill

#3340 2011-11-04 19:49:40

Re: The AMD/ATI Bar & Grill

#3341 2011-11-04 22:13:04

Re: The AMD/ATI Bar & Grill

#3342 2011-11-05 14:28:54

Re: The AMD/ATI Bar & Grill

#3343 2011-11-05 15:14:51

Re: The AMD/ATI Bar & Grill

#3344 2011-11-06 04:45:09

Re: The AMD/ATI Bar & Grill

#3345 2011-11-06 08:59:17

Re: The AMD/ATI Bar & Grill

#3346 2011-11-07 13:42:39

Re: The AMD/ATI Bar & Grill

#3347 2011-11-07 15:51:50

Re: The AMD/ATI Bar & Grill

#3348 2011-11-09 00:09:33

Re: The AMD/ATI Bar & Grill

#3349 2011-11-09 00:17:09

Re: The AMD/ATI Bar & Grill

#3350 2011-11-09 01:41:20

Re: The AMD/ATI Bar & Grill

Board footer