AMD Catalyst 8.7 Linux Driver Released

[phoronix]

Phoronix: AMD Catalyst 8.7 Linux Driver Released

Since last month's release of Catalyst 8.6 for Linux we've seen the introduction of the Radeon HD 4850 and 4870 graphics cards and we've been allowed to share with you that CrossFire is coming to Linux along with other yet to be announced features. Today AMD has released the Catalyst 8.7 Linux driver and it doesn't deliver any new ground-breaking features, but it does bring a few improvements. In fact, the new features in this release are just bringing official Ubuntu 8.04 and SuSE Linux Enterprise Desktop 10 SP2 support...

http://www.phoronix.com/vr.php?view=NjYwNQ

[bugmenot]

I recently bought myself a new computer and installed today fglrx for the first time. My Impressions:
After a "successfull-looking" installation and a restart the activated driver was still "radeonhd". I needed to run "aticonfig --initial -f" in order to use fglrx.
After a second restart I got a "white screen of death" after logging in. After removing a few lines in xorg.conf that imply that AIGLX is not available and third restart it worked! :)
Scrolling in Firefox for example is not as smooth as with radeonhd, but compiz works. As soon as radeonhd supports compiz I'll switch back to radeonhd, though.

My System:
Ubuntu 8.04 64 bit
GigaByte GA-MA78G-DS3H (3200 onboard)
3 GB RAM

Edit: Graphics corruption when I shut down and error messages when trying to suspend (it works though). Very bad, for now I'll switch back to radeonhd.

[d2kx]

Support for Mobility Radeon X2000 and X3000 series seems to be added.

[Vighy]

Support for Mobility Radeon X2000 and X3000 series seems to be added.

mhhhhh I will tell you if it's true once I will try it! XD :D :D :D

[Alexander Heß]

Resolved Issues

* An X segmentation fault no longer occurs when launching SUZI/textured video. Further details can be found in topic number 737-35214

Known Issues

* An X segmentation fault may occur when launching SUZI/textured video. Further details can be found in topic number 737-35214


What the…?

Did they fix it or didn't they?

[NeoBrain]

Resolved Issues:
Creating two successive connections no longer results in direct GL context to fail.
I hope that's what disabled my 3D accel in 8.6 when I started a second X server, sounds promising.
TOO BAD however, that they still don't seem to have fixed the damn Wine bug >_<
Well, bug fix releases are better than nothing though, keep up the good work ;-)

[NeoBrain]

What the…?

Did they fix it or didn't they?

Guess they fixed it but forgot to remove the "known issue"...
After the watermark from 8.5 it wouldn't surprise me ;-)

[NeoBrain]

Okay, first impressions (jumped from 8.4 to 8.7):
Everything on first X server works like on 8.4 (2D, 3D, compiz), no clear improvements (can't tell without benchmark, it's fluent); Firefox scrolling is still damn slow with compiz, but we all know that already
On the 2nd X server I still get no 3D accel (it fails to init my AGP)

Ah yeah, nearly forgot to test Wine...
Great, worse than before: Got a corruption for about 5 secs, then the usual problems arised, I quitted Wine and the corruption luckily got fixed... turning off Compiz doesn't change anything.

Guess I stick with that release though, won't get much better than staying with 8.4 ;)

This is on a Sapphire X1600 PRO 512 MB AGP.

[Wise Ferret]

Xv working for me now on x1600.
Interestingly, I can even run full screen 3d apps under compiz without flickering. Video still flickers under compiz.

Keep up the good work!

[Forge]

TOO BAD however, that they still don't seem to have fixed the damn Wine bug >_<

Oh, COME ON AMD!!!

This is the only bug I had in 8.6 that was a show-stopper for my 4850. Windows issues were largely avoidable, though annoying.

I use Wine a LOT. Having it unavailable is a much bigger thing for me than any of the things that were fixed.

I took out my 4850 for my 8800GT last night, it looks like it'll be at least 8.8 before I put it back in. At this rate I might be on RadeonHD sooner.

[seanbarman]

Hi,

Still get graphics corruption!

When I use zattoo player in fullscreen and return to windowed mode. the screen is corrupted, I cant play cedega either.

glxgears is now down to the speed of my 1950 pro.

Thanks AMD I buy a new card and each driver gets slower and slower and more bugs. Guess what you can keep the 4870.

I suppose it's to much to ask, to buy some thing and actually works.

I bet the 3xxx series is forgotten now and all drivers concentrate on 4XXX series thats your usual style, just have to read here to know that.

Thank God you opened up the driver !

Bring on the Radeon HD 3D!

Ill wait for Intels graphics cards!

[Vighy]

Xorg hangs with this new wonderful driver..

(**) fglrx(0): ATI GART size: 255 MB
(II) fglrx(0): [pcie] 261120 kB allocated
(II) fglrx(0): [drm] DRM buffer queue setup: nbufs = 100 bufsize = 65536
(**) fglrx(0): UseFastTLS=1
(==) fglrx(0): BlockSignalsOnLock=1
(II) fglrx(1): === [atiddxPreInit] === begin
(EE) fglrx(1): Quitting secondary screen -- no monitor specified.
(EE) fglrx(1): PreInit failed
(II) fglrx(1): === [atiddxPreInit] === end

Backtrace:
0: /usr/bin/X(xf86SigHandler+0x6a) [0x49263a]
1: /lib/libc.so.6 [0x7f9767b64da0]
2: /usr/lib64/xorg/modules/drivers//fglrx_drv.so(atiddxDriverEntPriv+0xf) [0x7f976592dcef]
3: /usr/lib64/xorg/modules/drivers//fglrx_drv.so(atiddxFreeScreen+0x2e) [0x7f9765932b8e]
4: /usr/bin/X(xf86DeleteScreen+0x7e) [0x4a5ade]
5: /usr/bin/X(InitOutput+0xa1f) [0x46c0bf]
6: /usr/bin/X(main+0x2c6) [0x437256]
7: /lib/libc.so.6(__libc_start_main+0xe6) [0x7f9767b501d6]
8: /usr/bin/X(FontFileCompleteXLFD+0x289) [0x4366f9]

Fatal server error:
Caught signal 11. Server aborting


Obviously I needed to get back to 8.6 Argh! And they said Mobility X2600 are now supported!? when they were not supported, Xorg used to load fine!

This happens with my M76 (Mobility Radeon HD2600)...

[Shred]

Anybody tried it with Fedora 9 yet?

[Fenix-TX]

The same problems with showing tv with tv card like 8.6, i can't resize the window, because if i do that i can't see anything...so backing to 8.5...the only driver that works "fine"

[Marox]

Hi,
using an old Sapphire Radeon 9800 Pro and Xubuntu 8.04.1 i'm quiet surpriced about this new release. Looks like everything was speeded up !

-boottime reduced by 4 seconds
-my desktop seems to response a bit faster
-firefox scrolls faster (still could be faster) on problematic sites
-all games or gamedemos i tested are working noticeable faster, including:

*Ankh 2
*Penny Arcade EP1
*Penumbra Overdue
*Penumbra BlackPlaque
*Planeshift
*Freeciv

-Both Penumbra parts are working fine now, means no glitches anymore, very improved savegame loadtime !!!
Overdue: from 36sec down to 7sec
Blackplaque: from several minutes down to 12sec
(nearly all CG compile errors are gone now, hpl.logs size reduced from 116kb down to 27kb)
-no new issues for me so far

issues left to solve:

-glitches when starting games in fullscreen (resolution change at start up)
-slow 2D performace of apps(gtk) when running in front of firefox/thundebird
- ... see changelog ;)

So a quite good release, a least for me :)

[yoshi314]

TOO BAD however, that they still don't seem to have fixed the damn Wine bug >_<are you talking about distorted image like this one : http://launchpadlibrarian.net/12254636/Screenshot.jpg , or is it something else?

i haven't been using fglrx for a year or so, and i've tried briefly 8.6 and that was the only 3d-related issue i got. at least startopia started working with wine, yay! :D can't wait till the drivers catch up. mesa seems to crash instantly with wine, though :/

XServer may fail when playing media files with XVideo enabled. Further details can be found in topic number 737-35968and that was the only 2d-related issue i got :] . will have to re-check with new version.

i must say that fglrx has improved a LOT since last release i actually liked, which was 8.35 :] just a few more bugfixes and i'll consider using this driver on a daily basis.


edit: can somebody explain this? :

Resolved Issues
An X segmentation fault no longer occurs when launching SUZI/textured video. Further details can be found in topic number 737-35214

Known Issues
An X segmentation fault may occur when launching SUZI/textured video. Further details can be found in topic number 737-35214
so, is that fixed or what?


fglrx 8.7 - "mindtricks edition" :D

[alexforcefive]

Creating two successive connections no longer results in direct GL context to fail. Further details can be found in topic number 737-35215

... is very strangely worded, but I think I'm having this problem. Hopefully 8.7 will fix it

[dscharrer]

are you talking about distorted image like this one : http://launchpadlibrarian.net/12254636/Screenshot.jpg , or is it something else?

I guess he is talking about this: http://www.phoronix.com/forums/showthread.php?t=10940

Too bad this isn't fixed, since it keeps me stuck on 8.5.


Good to hear though that xv seems to work with an x1600 now. Got to try this once 8.7 is in portage.

[Kano]

That driver is as useless as the 8-6 driver before. It really sucks. The time you invest to fix 2.6.26 issues is really wasted...

a) 2.6.26 support is not integrated, ok it is possible to fetch a patch from git now (still not optimal, as not dkms like - applied only for intrepid packages not via PATCH=... , PATCH_MATCH="^2.6.2[6-9].

b) wine is not working just like 8-6

c) no pointsprites

d) xorg.conf modelines... well at least they don't make the xserver crash, but they still dont work. Like I added 1280x960 @ 85 hz and it uses only 60 Hz - like it would without modelines.

I presonally would remove the driver as it does not fix anything important. The only way is to use 8-5 - then you can at least play games with wine. If you call that Linux support then it must be a real joke.

[roberth]

No video tearing fix? Is it registered by amd at all?

[oyvind]

A couple of issues that have been bothering me seems to be fixed in this release.


Corruption when enabling my external CRT is gone, nice, no need to work-around by re-initializing with xrandr/xrefresh after enabling second display, anymore.


XVideo crasher
An X segmentation fault no longer occurs when launching SUZI/textured video. Further details can be found in topic number 737-35214

I've had problems with X crashing when opening XVideo in Catalyst 8.6 (with any video player app). Hopefully, this is the fix, but what is SUZI ??


From the Phoronix article related to the new release:
There are also a few fixes such as preventing freezes when resuming from OpenGL screensavers ..
Does "resuming" here mean moving the mouse cursor =) ? I've had problems with freezing OpenGL screensavers for as long as I can remember with fglrx (gnome screensaver). Is there any chance this fix will help ?



I'm just crossing my fingers and hope that the awful XVideo crasher is gone. Driver seems to be working just fine, so far.

My hw:
Lenovo Thinkpad Z61m, Radeon Mobility X1400 (chipset 0x7145), 2GB RAM, Intel Core Duo 2GHz, Ubuntu Hardy 32bit.

[jjjj]

For Ubuntu (debs meant for intrepid but currently only works on hardy, due X server problems, see below) download the debs from here:
https://launchpad.net/ubuntu/intrepid/+source/fglrx-installer/2:8.512-0ubuntu1

AMD64 debs can be found here (see the above page for i386 debs)
http://launchpadlibrarian.net/16221108/fglrx-amdcccle_8.512-0ubuntu1_amd64.deb
http://launchpadlibrarian.net/16221107/fglrx-kernel-source_8.512-0ubuntu1_amd64.deb
http://launchpadlibrarian.net/16221109/fglrx-modaliases_8.512-0ubuntu1_amd64.deb
http://launchpadlibrarian.net/16221105/xorg-driver-fglrx_8.512-0ubuntu1_amd64.deb
http://launchpadlibrarian.net/16221106/xorg-driver-fglrx-dev_8.512-0ubuntu1_amd64.deb

These appeared ~29 minutes ago ;)

Seems to work ok, but still some problems with 2.6.26 PAT and fglrx PAT to make peace:

Linux gandalf 2.6.26-2008-07-14 #1 SMP Mon Jul 14 22:21:34 EEST 2008 x86_64 GNU/Linux

[58763.437389] [fglrx] vendor: 1002 device: 9591 count: 1
[58763.438265] [fglrx] ioport: bar 1, base 0x9000, size: 0x100
[58763.438265] [fglrx] Maximum main memory to use for locked dma buffers: 3808 MBytes.
[58763.438279] [fglrx:KCL_enable_pat] *ERROR* Pat entry 2 is already configured
[58763.438339] [fglrx] PAT is disabled!
[58763.438354] [fglrx] module loaded - fglrx 8.51.3 [Jul 3 2008] with 1 minors
[58779.440177] [fglrx] Reserved FB block: Shared offset:0, size:1000000
[58779.440177] [fglrx] Reserved FB block: Unshared offset:ff7f000, size:80000



That driver is as useless as the 8-6 driver before. It really sucks. The time you invest to fix 2.6.26 issues is really wasted...

a) 2.6.26 support is not integrated, ok it is possible to fetch a patch from git now (still not optimal, as not dkms like - applied only for intrepid packages not via PATCH=... , PATCH_MATCH="^2.6.2[6-9].

b) wine is not working just like 8-6

c) no pointsprites

d) xorg.conf modelines... well at least they don't make the xserver crash, but they still dont work. Like I added 1280x960 @ 85 hz and it uses only 60 Hz - like it would without modelines.

I presonally would remove the driver as it does not fix anything important. The only way is to use 8-5 - then you can at least play games with wine. If you call that Linux support then it must be a real joke.

[mancunian]

Right, my Sapphire Radeon X1550-64 Bit is not the world's fastest card, but this really is a case of severe underperfoming. Catalyst 8.7 is like 8.6 still painfully slow under KDE 4.0 and KDE 4.1. That will please a hell of a lot of people who are now making the switch. If run under KDE 4.1 glx-gears gives me an average of 950,000 FPS, compared to an average 2,300,000 under KDE 3.5X or Xfce.

[bridgman]

That driver is as useless as the 8-6 driver before. It really sucks. The time you invest to fix 2.6.26 issues is really wasted...

I'll check, but my recollection was that we announced support for Ubuntu 8.04, not for 8.10 (or for 2.6.26 kernel).

Are we even claiming 2.6.26 support with the 8.7 release ?

[Saist]

I'll check, but my recollection was that we announced support for Ubuntu 8.04, not for 8.10 (or for 2.6.26 kernel).

Are we even claiming 2.6.26 support with the 8.7 release ?

https://a248.e.akamai.net/f/674/9206/0/www2.ati.com/drivers/linux/catalyst_87_linux.html

I see no mention of 2.6.26 support in the readme. The only mention of 2.6 is limited to:

# XOrg 6.8, 6.9, 7.0, 7.1, 7.2 or 7.3
# Linux Kernel 2.6 and above
# glibc version 2.2 or 2.3
# POSIX Shared Memory (/dev/shm) support is required for 3D applications

[Kano]

Well last thing to add:

Even when you patch kernel 2.6.26 support it still does not run with intrepid - because Xserver 1.4.99.905 which is used there is not supported due to undefined symbol miZeroLineScreenIndex. So what did you think you get ;)

@jjjj

The only way to use those debs _are_ on hardy or older, because intrepid is not supported - although the target exists.

[bridgman]

I didn't think we supported Intrepid (8.10) yet, just Hardy (8.04).

[Redeeman]

does it support wine yet?

[Kano]

Nope you don't do that. But intrepid's OSS driver for X700 SE (RV410) absolutely suck too. Everything flickers whenever GNOME calls xrandr or when you directly use it. The VGA port also does not detect DCC (at max DVI->VGA adapter). Using that card with a new Xserver is absolutely no joy and only increases the hate against such a crap.

[oyvind]

A couple of issues that have been bothering me seems to be fixed in this release.

<snip>

I'm just crossing my fingers and hope that the awful XVideo crasher is gone. Driver seems to be working just fine, so far.

My hw:
Lenovo Thinkpad Z61m, Radeon Mobility X1400 (chipset 0x7145), 2GB RAM, Intel Core Duo 2GHz, Ubuntu Hardy 32bit.


Dammit, X just crashed with XVideo (Xine this time) and Catalyst 8.7.
Arrrgh!

Here's some kind of backtrace (X log):
0: /usr/bin/X(xf86SigHandler+0x7e) [0x80c780e]
1: [0xb7f72420]
2: /usr/lib/xorg/modules//glesx.so [0xb4c0f5de]
3: /usr/lib/xorg/modules//glesx.so [0xb4be64bc]
4: /usr/lib/xorg/modules//glesx.so [0xb4be6c6a]
5: /usr/lib/xorg/modules//glesx.so [0xb4b6369b]
6: /usr/lib/xorg/modules//glesx.so [0xb4b5575e]
7: /usr/lib/xorg/modules//glesx.so [0xb4b533a4]
8: /usr/lib/xorg/modules//glesx.so(esutUpdateSurfData+0x2e) [0xb4b525de]
9: /usr/lib/xorg/modules//amdxmm.so [0xb4b014f0]
10: /usr/bin/X [0x80da805]
11: /usr/lib/xorg/modules/extensions//libextmod.so(XvdiPutImage+0x178) [0xb7bdd608]
12: /usr/lib/xorg/modules/extensions//libextmod.so [0xb7be0476]
13: /usr/bin/X [0x81506ee]
14: /usr/bin/X(Dispatch+0x2cf) [0x808d8df]
15: /usr/bin/X(main+0x48b) [0x807471b]
16: /lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xe0) [0xb7d06450]
17: /usr/bin/X(FontFileCompleteXLFD+0x201) [0x8073a91]

Fatal server error:
Caught signal 11. Server aborting

(II) AIGLX: Suspending AIGLX clients for VT switch

Oh my.
:(

[jeffro-tull]

man, I'm so glad radeonhd does everything I want it to on my mobility x1300. From the looks of things so far, it's another month, another buildup of hope, and another let-down from fglrx.

[reavertm]

xorg.conf modelines... well at least they don't make the xserver crash, but they still dont work. Like I added 1280x960 @ 85 hz and it uses only 60 Hz - like it would without modelines.


Actually ModeLines don't work with xf86-video-ati but they work with ATI proprietary pretty well.
I'm using 8.6 with such ModeLine with no problems (exactly the same resolution - 1280x960@85) - check that

ModeLine "1280x960" 149.25 1280 1372 1510 1740 960 963 967 1011

(ModeLine generated by xf86-video-radeonhd driver :P - the only that worked like a charm with my setup before they added atombios support - yep - radeonhd after adding initial atombios support refuses to fetch EDID from my AOC 9k+ CRT connected to x1950 PRO AGP card via DVI<->Dsub connector)

[jjjj]

Well last thing to add:

Even when you patch kernel 2.6.26 support it still does not run with intrepid - because Xserver 1.4.99.905 which is used there is not supported due to undefined symbol miZeroLineScreenIndex. So what did you think you get ;)

@jjjj

The only way to use those debs _are_ on hardy or older, because intrepid is not supported - although the target exists.

Ok, so I'm running Hardy with 2.6.26 anyways, so it worked for me, sorry ;)

[bridgman]

(ModeLine generated by xf86-video-radeonhd driver :P - the only that worked like a charm with my setup before they added atombios support - yep - radeonhd after adding initial atombios support refuses to fetch EDID from my AOC 9k+ CRT connected to x1950 PRO AGP card via DVI<->Dsub connector)

The atombios support is in a separate branch right now -- were you running from master or from the atombios support branch when you had the EDID problems ?

[Danyul]

does it support wine yet?

Forget wine!!! Does it support linux?!?

HAHA :p

[sundown]

Oh man, I don't even want to test that one. I'm staying on 8.5 :)

[reavertm]

The atombios support is in a separate branch right now -- were you running from master or from the atombios support branch when you had the EDID problems ?

It was master branch of radeonhd - I used to check it out from time to time. It was few days ago - this driver just for the first time suddenly refused to detect (even grab EDID) from my setup. Xorg.0.log shew some atombios entries, so it looks to me that some atombios code was being included in master branch. Anyway, I see there are frequent commits to this branch all the time so I guess it should work again shortly. For now I've switched to ATI proprietary (8.6, though it's fan control is driving me nuts - still waiting for the Gentoo ebuild for 8.7 to arrive) and using Modeline to set my favourite mode.

What's the best way to contact radeonhd team and give them some feedback about this? Looks like my configuration (Sapphire x1950 PRO AGP, AOC 9k+ CRT monitor, connected by DVI<->DSub cable) is quite hard to detect as radeonhd driver used to be the ONLY that was able to work without Modelines, "Driver" section Options except "norandr" and with no "Monitor" section at all. ATI proprietary doesn't work automagically here - and xf86-video-ati rejects Modelines - so probalby my feedback would help solving some issues.

cheers

[poofyyoda]

I think that this release is reasonable. Wine works perfect for me now, the corruption is fixed. Thats all I actually wanted for this release since 8.6 allowed me to actually 'run' some games in wine, 8.7 allows me to run them without virtual desktop on.

EDIT:

Actually, after trying to play some more decent games in wine, I have huge lagging, which I haven't had since many driver releases ago.
This is only experienced when using wine, native apps run perfectly. I hope AMD starts to look at its wine support some time soon..

[metala]

Four months after upgrading the kernel and now I have driver that really works, :)

[chrisr]

Dammit, X just crashed with XVideo (Xine this time) and Catalyst 8.7.

I've found that fglrx/XVideo crash the X server whenever xine reaches the end of a media file during normal playback. The work-around is either to use XShm instead of XVideo, or to quit xine before it reaches the end. This has been true with every fglrx module I've tested so far (up to 8.6, haven't tested 8.7 yet) for the past 12 months.

[marcobrancalion]

Nv now works, in viewing videos.

But, the sloppiness in 2D acceleration (even in stupid things like normale konsole windows of KDE) forced me to revert back to the 8.5 release.


As always, the work of guys at AMD's is INSULTING, for ati cards owners.
I'm firm with my intention to have absolutely an Nvidia card in my next laptop (i hope soon).
Even a shitty Nvidia card will work better with linux than my 'wonderful' Ati X1400.

[oyvind]

I've found that fglrx/XVideo crash the X server whenever xine reaches the end of a media file during normal playback. The work-around is either to use XShm instead of XVideo, or to quit xine before it reaches the end. This has been true with every fglrx module I've tested so far (up to 8.6, haven't tested 8.7 yet) for the past 12 months.

Yup, you are right, sir. I've set Xine (v 0.99.6cvs) to OpenGL-output, for now. This should be easy to reproduce, ATI guys. Just play something with Xine using xv output module, skip to near the end of the video and let it finish. Then watch your beautiful console appear, as X dies and restarts. My GPU is an X1400 Radeon Mobility. Ubuntu Hardy 32bit.

[yoshi314]

As always, the work of guys at AMD's is INSULTING, for ati cards owners.they're doing much better, but there are still brown-paper-bag bugs in the driver. i'd rather see fglrx come out every 3 months if it would guarantee better driver quality.

[sundown]

Yup, you are right, sir. I've set Xine (v 0.99.6cvs) to OpenGL-output, for now. This should be easy to reproduce, ATI guys. Just play something with Xine using xv output module, skip to near the end of the video and let it finish. Then watch your beautiful console appear, as X dies and restarts. My GPU is an X1400 Radeon Mobility. Ubuntu Hardy 32bit.

It's funny because I think there is a different behaviour under KDE. When the end of a video is reached, the KDE crash handler appears without X dying. But then again, I do this with Kaffeine which uses Xine.

[bridgman]

It was master branch of radeonhd - I used to check it out from time to time. It was few days ago - this driver just for the first time suddenly refused to detect (even grab EDID) from my setup. Xorg.0.log shew some atombios entries, so it looks to me that some atombios code was being included in master branch.

Ahh. There has been *some* atombios usage for perhaps 10 months; the difference is that the atombios-support branch uses atombios for most of the modesetting operations while master only uses atombios for a couple of initialization functions. If you weren't using the atombios-support branch I doubt your problems had anything to do with atombios.

What's the best way to contact radeonhd team and give them some feedback about this? Looks like my configuration (Sapphire x1950 PRO AGP, AOC 9k+ CRT monitor, connected by DVI<->DSub cable) is quite hard to detect as radeonhd driver used to be the ONLY that was able to work without Modelines, "Driver" section Options except "norandr" and with no "Monitor" section at all. ATI proprietary doesn't work automagically here - and xf86-video-ati rejects Modelines - so probalby my feedback would help solving some issues.

Probably best to subscribe to the radeonhd mailing list (info at http://lists.opensuse.org/radeonhd and post a message there.

[NeoBrain]

are you talking about distorted image like this one : http://launchpadlibrarian.net/12254636/Screenshot.jpg , or is it something else?


Nope, I'm talking about that one:
http://bugs.winehq.org/show_bug.cgi?id=13335
Prevents me from playing anything demanding on Wine, really sucks...
IIRC the problem has been existing since around 8.44, if not even 8.42 (i.e. with the new code base).
I guess that only computers with lower memory (I got 512 MB RAM) experience this problem as the explanation somewhere in the bug report suggests...
However, the corruption thing (which I also had) seems to be fixed.

[shreg]

Does anyone have tried it with a dual desktop settings?

If yes, is the video or opengl workings fine on both screen without compiz ? and with compiz?

thanks

[Porter]

Yup, you are right, sir. I've set Xine (v 0.99.6cvs) to OpenGL-output, for now. This should be easy to reproduce, ATI guys. Just play something with Xine using xv output module, skip to near the end of the video and let it finish. Then watch your beautiful console appear, as X dies and restarts. My GPU is an X1400 Radeon Mobility. Ubuntu Hardy 32bit.

There is a similar problem with RealPlayer 11 for Linux, which I believe is based on the Helix DNA Client. I have no clue whether Helix may also exhibit the problem.

[SavageX]

Good release for me so far (Radeon 3850, x86_64 Ubuntu 8.04) - now playing videos in Kaffeine won't crash the X server anymore for me, can you believe it!

Sorry to hear others aren't that lucky, though.

[alexcorscadden]

Sadly, the context creation issue (the successive connection bug) is still present. If one of the AMD folks wants to contact me, I have source for a small test app that I submitted to someone at AMD/ATI a few months back that very quickly (2 seconds to compile, less than a second to run) proves the existence of this bug.

Stability seems to be better now. I can call glXCreateNewContext without risking a segfault deep inside libGL.

[sundown]

Good release for me so far (Radeon 3850, x86_64 Ubuntu 8.04) - now playing videos in Kaffeine won't crash the X server anymore for me, can you believe it!

Sorry to hear others aren't that lucky, though.

Wow, that sounds encouraging coming from you, having in mind that what I only do with this driver is to play Nexuiz :D

[SavageX]

Wow, that sounds encouraging coming from you, having in mind that what I only do with this driver is to play Nexuiz :D

Well, one downside: I can't get Source-engine based games to work in wine anymore. With 8.6 I could just launch with "-dxlevel 81" and have it mostly running.

Given that I only have the Half-Life 2 demo/DM/Lost-Coast thingies this isn't a big loss for me, though ;)

[Forge]

Are we even claiming 2.6.26 support with the 8.7 release ?

I dunno, but you better not be claiming Wine support.

[Kano]

There is no direct 2.6.26 support, but you find the patch here:

http://phorogit.com/index.php?p=fglrx-packaging.git&dl=plain&h=a94205d6bb86dbf154c3e519b6da869c529cafe4

[Death Knight]

Thanks a lot Kano

[Vighy]

Kano 4 President (Of the Phoronix Forum) :D :D :D :D

[Vighy]

There is no direct 2.6.26 support, but you find the patch here:

http://phorogit.com/index.php?p=fglrx-packaging.git&dl=plain&h=a94205d6bb86dbf154c3e519b6da869c529cafe4

ehm sed -i -e "s/fglrx-installer-8.512/common/g" <that_patch>

would work better :D

[Kano]

Depends of the dir you are in and which -p level you use.

[Vighy]

Depends of the dir you are in and which -p level you use.

yep, but including it in a ebuild, -p0 is used by default and they patch from the root.

however, since 8.7 crashes Xorg for me ( http://www.phoronix.com/forums/showthread.php?t=11570 ) I back-ported the patch for 8.6... (just trivial changes)

may it be interesting for anyone?

[NIghtwulf]

Hi,

again a release without support for the HD3870X2 :(

Since I need the card for blender and ogre works, I wouldn't need the second core to be active. So it would be enough if the driver could make use of the first core. Since the HD3870 with just one core does run with the driver, that shouldn't be that big problem or am I wrong with that?

I can use radeonhd as the only alternative at the moment. But that one does support only the very basic 2D functions in 1.2.1 (anyways good work!).

I have to say, that this is my first ATI card for years. Most important reason for buying no NVidia again is, that ATI started to support OSS and relasing card specs.

So if somebody knows a workaround to get fglrx up and running with one core I would be very glad!

Greets,

Nightwulf

[Vighy]

Hi,

again a release without support for the HD3870X2 :(

Since I need the card for blender and ogre works, I wouldn't need the second core to be active. So it would be enough if the driver could make use of the first core. Since the HD3870 with just one core does run with the driver, that shouldn't be that big problem or am I wrong with that?

I can use radeonhd as the only alternative at the moment. But that one does support only the very basic 2D functions in 1.2.1 (anyways good work!).

I have to say, that this is my first ATI card for years. Most important reason for buying no NVidia again is, that ATI started to support OSS and relasing card specs.

So if somebody knows a workaround to get fglrx up and running with one core I would be very glad!

Greets,

Nightwulf

Guys, you were told that HD3870X2 would have not been supported under linux...

I know it was not clear to everybody.... but Bridgman already said it was supposed to work just under winblows.

[Almindor]

They finally fixed Savage 2 shaders, but they broke ETQW (crashes now) which sucks big time.

otherwise it seems to be working ok so far for me, no big difference in performance.

[NIghtwulf]

Hi,

Guys, you were told that HD3870X2 would have not been supported under linux...

I know it was not clear to everybody.... but Bridgman already said it was supposed to work just under winblows.

who told that? I'm new here and the Releas Notes in the official drivers say *currently* not supported. ATI did never tell their customers on any official channel, that the X2 cards will be only supported in Windows.

So that's no excuse nor do I believe they don't want to support it at all.

Nightwulf

[grantek]

Hi All,

This is my first post, so I just have to get this out of the way: THANK YOU AMD :) - my system is broken and doesn't play games, but I really appreciate the effort being put into Linux support, and my new system is all-AMD because of that fact.

Having said that, I need some help. I just bought a new AMD 9750 system with a HD4850, and I'm mainly trying to get ET:QW and Wine/Cedega going. I get a segfault on ET:QW, I haven't tested Wine on 8.7 (I was having unrelated problems while using 8.6, but I was also getting corruption). Reading the thread for the 8.6 driver, it seems like an mtrr issue.

Full system is:
AMD Phenom 9750
Gigabyte GA-MA790X-DS4
Gigabyte 512MB Radeon HD4850
1x 2GB Kingston 800MHz RAM module
Ubuntu 8.04 x86_64,running either -rt or -generic kernel (the X driver is currently compiled for -rt)

lspci -v is:
00:00.0 Host bridge: ATI Technologies Inc RD780 Northbridge only dual slot PCI-e_GFX and HT1 K8 part
Subsystem: ATI Technologies Inc RD780 Northbridge only dual slot PCI-e_GFX and HT1 K8 part
Flags: bus master, 66MHz, medium devsel, latency 32
Memory at <ignored> (64-bit, non-prefetchable)
Capabilities: [c4] HyperTransport: Slave or Primary Interface
Capabilities: [40] HyperTransport: Retry Mode
Capabilities: [54] HyperTransport: UnitID Clumping
Capabilities: [9c] HyperTransport: #1a

00:02.0 PCI bridge: ATI Technologies Inc RD790 PCI to PCI bridge (external gfx0 port A) (prog-if 00 [Normal decode])
Flags: bus master, fast devsel, latency 0
Bus: primary=00, secondary=01, subordinate=01, sec-latency=0
I/O behind bridge: 0000d000-0000dfff
Memory behind bridge: fdf00000-fdffffff
Prefetchable memory behind bridge: 00000000d0000000-00000000dfffffff
Capabilities: [50] Power Management version 3
Capabilities: [58] Express Root Port (Slot-) IRQ 0
Capabilities: [a0] Message Signalled Interrupts: Mask- 64bit- Queue=0/0 Enable+
Capabilities: [b0] Subsystem: ATI Technologies Inc Unknown device 5958
Capabilities: [b8] HyperTransport: MSI Mapping

00:0a.0 PCI bridge: ATI Technologies Inc RD790 PCI to PCI bridge (PCI express gpp port F) (prog-if 00 [Normal decode])
Flags: bus master, fast devsel, latency 0
Bus: primary=00, secondary=02, subordinate=02, sec-latency=0
I/O behind bridge: 0000e000-0000efff
Memory behind bridge: fde00000-fdefffff
Prefetchable memory behind bridge: 00000000fdd00000-00000000fddfffff
Capabilities: [50] Power Management version 3
Capabilities: [58] Express Root Port (Slot-) IRQ 0
Capabilities: [a0] Message Signalled Interrupts: Mask- 64bit- Queue=0/0 Enable+
Capabilities: [b0] Subsystem: ATI Technologies Inc Unknown device 5958
Capabilities: [b8] HyperTransport: MSI Mapping

00:12.0 SATA controller: ATI Technologies Inc SB600 Non-Raid-5 SATA (prog-if 01 [AHCI 1.0])
Subsystem: Giga-byte Technology Unknown device b005
Flags: bus master, 66MHz, medium devsel, latency 32, IRQ 22
I/O ports at ff00 [size=8]
I/O ports at fe00 [size=4]
I/O ports at fd00 [size=8]
I/O ports at fc00 [size=4]
I/O ports at fb00 [size=16]
Memory at fe02f000 (32-bit, non-prefetchable) [size=1K]
Capabilities: [60] Power Management version 2

<USB controllers snipped because the post was too long>

00:14.0 SMBus: ATI Technologies Inc SBx00 SMBus Controller (rev 14)
Subsystem: Giga-byte Technology Unknown device 4385
Flags: 66MHz, medium devsel
I/O ports at 0b00 [size=16]
Capabilities: [b0] HyperTransport: MSI Mapping

00:14.1 IDE interface: ATI Technologies Inc SB600 IDE (prog-if 8a [Master SecP PriP])
Subsystem: Giga-byte Technology Unknown device 5002
Flags: bus master, 66MHz, medium devsel, latency 32, IRQ 16
I/O ports at 01f0 [size=8]
I/O ports at 03f4 [size=1]
I/O ports at 0170 [size=8]
I/O ports at 0374 [size=1]
I/O ports at f900 [size=16]

00:14.2 Audio device: ATI Technologies Inc SBx00 Azalia
Subsystem: Giga-byte Technology Unknown device a002
Flags: bus master, slow devsel, latency 32, IRQ 16
Memory at fe024000 (64-bit, non-prefetchable) [size=16K]
Capabilities: [50] Power Management version 2

00:14.3 ISA bridge: ATI Technologies Inc SB600 PCI to LPC Bridge
Subsystem: Giga-byte Technology Unknown device 5001
Flags: bus master, 66MHz, medium devsel, latency 0

00:14.4 PCI bridge: ATI Technologies Inc SBx00 PCI to PCI Bridge (prog-if 01 [Subtractive decode])
Flags: bus master, VGA palette snoop, 66MHz, medium devsel, latency 64
Bus: primary=00, secondary=03, subordinate=03, sec-latency=64
I/O behind bridge: 0000c000-0000cfff
Memory behind bridge: fdc00000-fdcfffff
Prefetchable memory behind bridge: fdb00000-fdbfffff

00:18.0 Host bridge: Advanced Micro Devices [AMD] Family 10h [Opteron, Athlon64, Sempron] HyperTransport Configuration
Flags: fast devsel
Capabilities: [80] HyperTransport: Host or Secondary Interface

00:18.1 Host bridge: Advanced Micro Devices [AMD] Family 10h [Opteron, Athlon64, Sempron] Address Map
Flags: fast devsel

00:18.2 Host bridge: Advanced Micro Devices [AMD] Family 10h [Opteron, Athlon64, Sempron] DRAM Controller
Flags: fast devsel

00:18.3 Host bridge: Advanced Micro Devices [AMD] Family 10h [Opteron, Athlon64, Sempron] Miscellaneous Control
Flags: fast devsel
Capabilities: [f0] #0f [0010]

00:18.4 Host bridge: Advanced Micro Devices [AMD] Family 10h [Opteron, Athlon64, Sempron] Link Control
Flags: fast devsel

01:00.0 VGA compatible controller: ATI Technologies Inc Unknown device 9442 (prog-if 00 [VGA controller])
Subsystem: ATI Technologies Inc Unknown device 0502
Flags: bus master, fast devsel, latency 0, IRQ 18
Memory at d0000000 (64-bit, prefetchable) [size=256M]
Memory at fdfe0000 (64-bit, non-prefetchable) [size=64K]
I/O ports at de00 [size=256]
[virtual] Expansion ROM at fdf00000 [disabled] [size=128K]
Capabilities: [50] Power Management version 3
Capabilities: [58] Express Legacy Endpoint IRQ 0
Capabilities: [a0] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable-

01:00.1 Audio device: ATI Technologies Inc Unknown device aa30
Subsystem: ATI Technologies Inc Unknown device aa30
Flags: bus master, fast devsel, latency 0, IRQ 19
Memory at fdffc000 (64-bit, non-prefetchable) [size=16K]
Capabilities: [50] Power Management version 3
Capabilities: [58] Express Legacy Endpoint IRQ 0
Capabilities: [a0] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable-

02:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 01)
Subsystem: Giga-byte Technology Unknown device e000
Flags: bus master, fast devsel, latency 0, IRQ 509
I/O ports at ee00 [size=256]
Memory at fdeff000 (64-bit, non-prefetchable) [size=4K]
[virtual] Expansion ROM at fdd00000 [disabled] [size=128K]
Capabilities: [40] Power Management version 2
Capabilities: [48] Vital Product Data
Capabilities: [50] Message Signalled Interrupts: Mask- 64bit+ Queue=0/1 Enable+
Capabilities: [60] Express Endpoint IRQ 0
Capabilities: [84] Vendor Specific Information

03:0e.0 FireWire (IEEE 1394): Texas Instruments TSB43AB23 IEEE-1394a-2000 Controller (PHY/Link) (prog-if 10 [OHCI])
Subsystem: Giga-byte Technology Unknown device 1000
Flags: bus master, medium devsel, latency 32, IRQ 22
Memory at fdcff000 (32-bit, non-prefetchable) [size=2K]
Memory at fdcf8000 (32-bit, non-prefetchable) [size=16K]
Capabilities: [44] Power Management version 2


/proc/mtrr is:
reg00: base=0x00000000 ( 0MB), size=15730688MB: write-back, count=1
reg01: base=0x7ff00000 (2047MB), size=15728641MB: uncachable, count=1

If I try to disable reg00 with the system booted, or in the startup before X, the system crashes. If I add a new line at base 0x00000000 with a 2GB size, the count goes up and the size reported in /proc/mtrr doesn't change (removing it works, and just decrements the count). As far as I'm aware, I should be removing reg01 and adding an uncachable block where the 4850's memory is, but I don't know where that is. I saw some references to 0xD0000000, but the only reference to that in lspci is a range of 0xD0000000-0xDFFFFFFF, which is 256MB.

I also have no clue why my /proc/mtrr ranges are so huge by default. If anything, it looks like it should work because all but 1MB of the system memory is write-back, and from there on (where the graphics RAM would be) it's uncachable. I don't know what happens when there's an overlap in the memory ranges.

Any ideas?

[Vighy]

Hi,



who told that? I'm new here and the Releas Notes in the official drivers say *currently* not supported. ATI did never tell their customers on any official channel, that the X2 cards will be only supported in Windows.

So that's no excuse nor do I believe they don't want to support it at all.

Nightwulf

Bridgman said it here: http://www.phoronix.com/forums/showpost.php?p=37822&postcount=8

I'm sorry.

[pchristy]

There is no direct 2.6.26 support, but you find the patch here:

http://phorogit.com/index.php?p=fglrx-packaging.git&dl=plain&h=a94205d6bb86dbf154c3e519b6da869c529cafe4

Hi, I'm new to this forum and fairly desperate for a fix! I'm running Slamd-64 (64-bit port of Slackware) on kernel 2.6.26. None of the ATI drivers will compile with this kernel, and I can't wait a month for the next release!

I normally install from the xxx.run file that I download from the ATI website. However, I can't figure out how to apply this patch to the "run" file. I suspect the file has to be unpacked, patched and then re-packed, but how?

An idiots guide would be appreciated!

--
Pete

[gsmd]

ok, on gigabyte's 780 mobo i don't get the black screen anymore, the system jest hangs! quit outsourcing to india, seriously!

[bridgman]

Those are *really* wierd looking mtrrs... don't know if they're the problem though. Can you pastebin a copy of your xorg log somewhere so we can see if the driver is complaining about anything ?

[mega_mike]

My Ubuntu 8.04 system in a Dell Inspiron 6400 with an ATI Mobility x1400 still has the same issue I have had since the initial catalyst driver release. When the hell is ATI going to address stability in their driver damnit! :mad:

Performance seems to be slightly better but my system still grinds to a halt after roughly 6 hours of usage (compiz enabled). I can't restart the X server at that point without it hard locking my machine (the only way to reboot is to issue direct kernel commands alt+prntscn+s then al+printscn+b).

[grantek]

Those are *really* wierd looking mtrrs... don't know if they're the problem though. Can you pastebin a copy of your xorg log somewhere so we can see if the driver is complaining about anything ?
pastebin.com/me640986 - dmesg
pastebin.com/m57b42621 - Xorg.0.log
pastebin.com/m56e4491c - xorg.conf - pretty standard I think

I'm mildly worried about the memory ranges in dmesg, if the card wants a single block from 0xd0000000-0xefffffff and dmesg shows 0xe0000000-0xefffffff is unavailable, it could be artificially limiting me to 256MB.

[benwick]

Here is an ebuild file I made for Gentoo users who wish to use Kernel 2.6.26. I took it from the ebuild in the previous release, changed the download link and version number, and added the patch Kano posted before (with the paths fixed). I can't guarantee it's perfect but it seems to work.

If you're wondering why it's 8.502 instead of 8.512, it's only so that my custom ebuild will be overridden by the real ebuild when Gentoo gets the real ebuild into portage.

http://www.undergroundcircus.net/ati-drivers-8.502.ebuild
http://www.undergroundcircus.net/kernel-2.6.26.patch

Download the above files. Put ati-drivers-8.502.ebuild in /usr/portage/x11-drivers/ati-drivers . Put the kernel patch file into /usr/portage/x11-drivers/ati-drivers/files AND into /usr/portage/x11-drivers/ati-drivers/files/8.476 (because I say so is why).

cd /usr/portage/x11-drivers/ati-drivers
ebuild ati-drivers-8.502.ebuild manifest
emerge ati-drivers

Voila! If you have problems I can provide only very limited support.

[dscharrer]

Initially Catalyst 8.7 didn't change anything regarding screen corruption for me on Gentoo amd64, xorg 7.3 (xorg-server 1.4.2) with a
Radeon HD 4850.

However, after adding all X packages to package.keywords (I had most of them there, but seemed to have forgotten a few libraries / headers) and unmerging all X packages and then re-emerging the "xorg-x11" "xorg-server" packages I don't get any corruption anymore.

I tested this with wine and Kaffeine (xine).

[reavertm]

http://bugs.gentoo.org/show_bug.cgi?id=232609

[RealNC]

This is my first post, so I just have to get this out of the way: THANK YOU AMD :) - my system is broken and doesn't play games

Am I the only one who went LOL at that one? :p

[grantek]

Am I the only one who went LOL at that one? :p
:p I just hate it when people bite the hand that's trying very hard to feed them.

I basically have no usable 3d though. compiz reports texture_from_pixmap is not present, ET:QW segfaults when loading a map, doom3 segfaults before showing the main menu, Cedega games seem to die whenever trying to do something 3d. Quake4 does get into the game, but all the textures are black and even on "low quality" performance is horrible.

If bridgman doesn't have any ideas given my logs on page 7, my next step is to try a clean install of something (maybe Ubuntu 64/32, maybe something else), to see if it's anything to do with my OS. :(

[RealNC]

I can't even resize a window in Compiz with fglrx. Every time I try, everything freezes for about 6 to 8 seconds. Oh yes, that's a very "pro" driver indeed :P

[bridgman]

:p I just hate it when people bite the hand that's trying very hard to feed them.

I basically have no usable 3d though. compiz reports texture_from_pixmap is not present, ET:QW segfaults when loading a map, doom3 segfaults before showing the main menu, Cedega games seem to die whenever trying to do something 3d. Quake4 does get into the game, but all the textures are black and even on "low quality" performance is horrible.

If bridgman doesn't have any ideas given my logs on page 7, my next step is to try a clean install of something (maybe Ubuntu 64/32, maybe something else), to see if it's anything to do with my OS. :(

I don't see anything obviously wrong with any of the files, but those mtrr's still seem really wrong. I would probably try a fresh install of Ubuntu 32-bit, get the 8.7 download, then use the installer to build a package for Ubuntu and try that.

[rob2687]

With 8.5 onwards I was getting the corrupted screen on Gnome System Monitor and some apps had messed up colours. There was also the screen corruption when running Wine.

I removed every trace of the current drivers and installed 8.7. Now only Wine gives the screen corruption.

[mirak63]

hi
is video tearing fixed ?

[RealNC]

hi
is video tearing fixed ?

Nope. Still there and kicking. :p

[grantek]

I don't see anything obviously wrong with any of the files, but those mtrr's still seem really wrong. I would probably try a fresh install of Ubuntu 32-bit, get the 8.7 download, then use the installer to build a package for Ubuntu and try that.
OK, I installed Ubuntu 8.04 32-bit on a new partition, didn't update any packages (although I would have got the latest build-essential etc. packages as part of the driver install process), built the fglrx packages, and they seemed to install OK. When I rebooted, compiz was enabled and seemed to work fine, but I disabled it before doing too much testing.
Loaded up ET:QW, and it got into the game OK. There were some moments of video tearing, but the main problem was occasionally the screen would freeze, alternating between two frames (half-torn as well) with the game still running fine. Pressing esc to get to the menu then resuming the game occasionally fixed it, but after a while it got to the point where even that wasn't working. What did work was looking directly up or down - looking horizontal caused the graphics to freeze, but I could run around with no problems looking at the ground or sky.

I also tried installing TrackMania Nations Forever in the latest Wine, but got full-screen corruption when loading the laucher window. I checked the mtrrs, and they were perfectly sane - one at base 0x0, size 2047MiB, one at base 2047MiB, size 1MiB. I tried adding an uncachable range at 0xd000000 size 256MiB, then one more at 0xe000000 size 256MiB (I have a 512MB card, but lspci etc. shows a range from 0xd000000-0xdfffffff), but neither step helped.

What was odd, was what happened when I rebooted into my normal 64-bit system - I checked the mtrrs, and they were perfectly normal, just like in the 32-bit install (before I had something like 15-terabyte ranges that were mostly overlapping). It didn't seem to have any effect on my problems in the 64-bit install though. I don't remember if I tried it with a cold boot or just a warm reboot, but I definitely don't understand enough about how the mtrrs are detected.

Things I'm thinking I should try:
- play with user limits (someone mentioned it helped wine screen corruption)
- take a closer look at adding a card-specific non-cachable mtrr
- other apps, both in and out of Wine/Cedega
- uninstalling fglrx, updating the OS, reinstalling fglrx
- fresh install of Ubuntu 64-bit on the spare partition
- fresh install of CentOS, since Red Hat is supposed to be supported

Any other ideas? If anyone wants some specific testing done, let me know :)

If I can get my problems sorted in Ubuntu32 I'll be much happier with a dual-boot system with two Ubuntus than Ubuntu and Windows :p

[legume]

compiz was enabled and seemed to work fine, but I disabled it before doing too much testing.

Loaded up ET:QW, and it got into the game OK. There were some moments of video tearing, but the main problem was occasionally the screen would freeze, alternating between two frames

This probably won't help at all, but -

make sure echo $LIBGL_ALWAYS_INDIRECT doesn't say 1 after disabling compiz as if it does it can really hurt game performance.

Once I got ETQW into a state where it would display corrupt graphics on loading some maps - the solution was to delete ~/.etqwcl, so you could try back up / delete that.

I don't think you are alone in having problems with wine games - nexuiz may be a better test.

[reavertm]

looking horizontal caused the graphics to freeze, but I could run around with no problems looking at the ground or sky.

same problem hereon ati x1950 pro agp with 8.6 and 8.7 catalyst

[mega_mike]

same problem hereon ati x1950 pro agp with 8.6 and 8.7 catalyst

Hey, I have a Mobility x1400 and I was wondering if this issue affects your system too:

After about 6 hours of usage your cpu usage spikes to around 50% (i have a dual core) with gnome-system monitor and xorg processes being marked as culprits. Memory usage is at 27%

Trying to restart the x server results in a system hard lock once this behavior is exhibited.

[grantek]

make sure echo $LIBGL_ALWAYS_INDIRECT doesn't say 1 after disabling compiz as if it does it can really hurt game performance.

Once I got ETQW into a state where it would display corrupt graphics on loading some maps - the solution was to delete ~/.etqwcl, so you could try back up / delete that.

I don't think you are alone in having problems with wine games - nexuiz may be a better test.
$LIBGL_ALWAYS_INDIRECT wasn't set, and changing the stack ulimit to unlimited didn't help ETQW or wine. After reading the thread again, I believe bridgman now that the mtrrs aren't a problem with 2GB memory (the graphics card sits out at past the 3GB mark, and I believe if it isn't explicitly set it defaults to uncachable).

I downloaded Nexuiz and played it for a while with no problems - great fun, I hadn't bothered with it before since I already have Quake 3 [Team] Arena, and tend to play Unreal Tournament as the "old game" at the LAN parties I go to :)

Playing ETQW again, it seemed to be a deterministic thing, depending on where I am in the map and/or what I'm looking at (ie. there's specific zones on the map where the screen freezing happens). I didn't have too much time to think about what was happening in the graphics, but it didn't happen so much indoors, or when looking out over vast spaces. The only theory I have is that it's to do with the number of polygons being rendered / processed - I don't know how modern games do polygon-culling on the Z axis, but I know they use lower-quality models at long distances, so the "vast open spaces" situation is probably similar to indoors.

Next up I'll try some similar games with Linux clients (Quake 4, Doom 3), and try patching the system to the latest kernel/libraries to see if that breaks anything.

[enzobelmont]

There is no direct 2.6.26 support, but you find the patch here:

http://phorogit.com/index.php?p=fglrx-packaging.git&dl=plain&h=a94205d6bb86dbf154c3e519b6da869c529cafe4

applied Kano solution but now my screen blinks every 90 seconds...

any idea?

PS: using gentoo ebuild and patch from post on 7th page

[grantek]

Ah ha! - I updated my 32-bit install to the latest kernel/libraries etc., and all was fine, so I figured I was looking in the wrong place.

I went back to the start, trying to figure out what was different about my setup, and found it - a workaround from my previous hardware. I had an nVidia card and was suffering the black window bug, so I'd added a configuration option somewhere to force indirect rendering. I disabled that, and now compiz works in my 64-bit system (I don't know how I lived without it :))

I still get segfaults trying to play my native linux games, but I've got my 32-bit install for that. Seems to always get a laugh when I tell people about my "dual-boot setup for gaming":
"oh, so you boot into winxp for games"
"nah, I have Ubuntu as my main system, and boot into Ubuntu when I want to play Quake Wars :)"

[Inkaine]

ok, on gigabyte's 780 mobo i don't get the black screen anymore, the system jest hangs! quit outsourcing to india, seriously!

Well, apparently maybe you should go to India? What you report could be caused by ANYTHING... I also got "gigabyte's 780 mobo" (there are 4 different one's atm, I got the GA-M78G-DS3H) - and 8.7 has been the best release driver for me so far. On my Radeon 3200 I used to have the reported crashes at the ends of videos until the 8.5 release. With 8.6 however I had segmentation faults trying to run opengl, so I had been using 8.5 until recently. Now switching to the latest release both are working for me. More stable (and IMO faster) 2D and working 3D. Thank you, AMD!

Now if just Wine worked ... I had the iPixel error, some people had been reporting, until 8.6. This is now gone but I get the "famous" screen corruption. Strangely I got that with a fullscreen opengl app also. Switching to window mode fixed it there (even an immediate in game switch). I wonder if that worked with wine too? Anyone knows how to prevent windows games from starting up fullscreen?

[mirak63]

why isn't video tearing with xv not a top priority ?

[roberth]

why isn't video tearing with xv not a top priority ?

I ask my self the same question.

[grantek]

I still get segfaults trying to play my native linux games, but I've got my 32-bit install for that.
Some more interesting stuff I've found when I've had time to meta-play games over the past week or so:
- Quake Wars is unplayable at any resolution even in the 32-bit Ubuntu install, due to the graphics-freezing bug when the view gets busy.
- Quake Wars in my 32-bit OS causes the "checkerboard of doom" - full-screen graphics corruption, offsetting bits of the screen everywhere, when I try to change the screen resolution.
- Nexuiz compiled for x84_64 runs brilliantly on my normal 64-bit workstation. I've had fullscreen corruption when trying to change resolution, but I've also changed resolution and not had it. Nexuiz for i386 on my x86_64 system experiences heavy screen glitching (screen flashing black) and crashing.

Anyone know if any of this is useful/new information the unofficial bugzilla? I haven't looked at it much, I'll need to do a thorough search and see what I can add :)

[kusuriya]

I was wondering if I could get a few brave souls running 8.7 that have a problem with the wine "Checkerboard of Doom" to try something and see if they get video corruption off it. I was noticing when I enabled

Option "DesktopSetup" "horizontal"
Option "UseFastTLS" "1"
Option "OpenGLOverlay" "1"
Option "TexturedVideo" "On"
Option "Textured2D" "on"
Option "TexturedXRender" "on"
Option "BackingStore" "on"
Option "XAANoOffscreenPixmaps" "true"

as such even my normal 2d desktop (such as just moving a terminal window) was having the same corruption issues I got in wine/checkerboard of doom... just to a bit of a lesser degree. If it has already been tried and pointed out sorry, but if not I would really like to see if its the same crossed the board.

[grantek]

I was wondering if I could get a few brave souls running 8.7 that have a problem with the wine "Checkerboard of Doom" to try something and see if they get video corruption off it.
Whoah, yep, I see what you mean, and can confirm it happens on my setup. It's not the same as the checkerboarding I've seen (where sections of the display seem permanently displaced by regular amounts), but moving windows around, there's heaps of corruption with bits of windows not being updated, going invisible, and random bits of windows being blatted out across random other bits of the screen. I didn't see any problems with compiz-fusion enabled.

[kusuriya]

Whoah, yep, I see what you mean, and can confirm it happens on my setup. It's not the same as the checkerboarding I've seen (where sections of the display seem permanently displaced by regular amounts), but moving windows around, there's heaps of corruption with bits of windows not being updated, going invisible, and random bits of windows being blatted out across random other bits of the screen. I didn't see any problems with compiz-fusion enabled.


Yeah I noticed on my setup that the back buffer was also displaying correctly (screenshots looked fine) and I really havent bothered with compiz yet makes me wonder
if that is an extension of the same bug or just a whole different one.

[grantek]

Yeah, from a lay perspective it seems like a memory management bug...

[kusuriya]

Well honestly ive found in my past debugging experience, the more ways you can reproduce the same bug, the easier it is to find the source. The trick is figureing out is if its the same bug or a new one. *sigh* I just want my wine XD

[Michael]

8.8 is now out: http://www.phoronix.com/vr.php?view=12767

[Michael]

This thread is being closed, please discuss 8.8 in: http://phoronix.com/forums/showthread.php?t=12183