The AMD/ATI Bar & Grill

ChemBro · 2010-08-28 16:29:56

Gillian00 wrote:

same for me :
[gillian@myhost ~]$ glxgears with vsync
1642 frames in 5.0 seconds = 328.219 FPS
1601 frames in 5.0 seconds = 319.964 FPS

This doesn't make any sense. It should be 60 fps or whatever your monitor can handle (or lower like 30/20 fps), not 320 - 330.

dcc24 · 2010-08-28 16:37:17

ChemBro wrote:

This doesn't make any sense. It should be 60 fps or whatever your monitor can handle (or lower like 30/20 fps), not 320 - 330.

True. However, I've seen that behavior with the xf86-video-ati driver where the fps is much higher than the refresh rate, even though vsync is enabled.

Gillian00 · 2010-08-28 16:58:52

huh I always had this kind of result...

I hit 5000 fps with just x11 running.

Vi0L0 · 2010-08-28 17:47:35

Cdh wrote:

With vsync glxgears gives 30 instead of 60 fps. Games with vsync do not run smooth. Without vsync it works normally. Can anyone reproduce?
edit: since catalyst 10.8 I mean. Before it worked normally too.

same here

Vi0L0 · 2010-08-28 17:48:21

IMPORTANT:
lib32-catalyst-utils 10.8-2 package works now with [multilib] repository, so if you wanna make it work you'll need to:
1. add this repository to pacman.conf - add those lines above any other repository:

[multilib]
Include = /etc/pacman.d/mirrorlist

2. then:

pacman -R lib32-libstdc++5

Cuz it looks like lib32-catalyst-utils doesn't need lib32-libstdc++5 package anymore, and (older, it's now updated in AUR) libstdc++5 is in conflict with [multilib] repo:
:: lib32-libstdc++5: requires lib32-gcc
lib32-gcc has been replaced with gcc-multilib package in [multilib] repo
Although it looks like you dont need to replace (ofcourse you can, it's safe) gcc gcc-libs and binutils packages with gcc-multilib gcc-libs-multilib and binutils-multilib cuz lib32-catalyst-utils is working good with both of them - at least for me...
3. and

pacman -Syu

Now you may catch conflicts if you are using some exotic 32-bit libs/applications. Simple removing conflicting packages (mainly libs) isn't always safe cuz those exotic applications could get brake - so before removing make a list of those conflicting packages. Ofcourse you may rebuild removed packages but know that not all of AUR's 32-bit libs/apps have been rewritten to work with new multilib repo. It should be easy to rewrite it by yourself : changing deps + changing libs place from /opt/lib32/usr/lib/ directory to /usr/lib32 directory.
After successful update you should check existing of files in /opt/lib32/usr/lib/ directory. If there are any files(libs) check which package is owner of them with 'pacman -Qo file' command.
Basically after successful update and rebuild /opt/lib32/usr/lib/ directory should not exist.
For more informations visit this thread https://bbs.archlinux.org/viewtopic.php?id=103699

Feedback is as always welcome.

AUR's lib32-catalyst-utils has been updated, repo will be updated soon.

Ps. I wouldn't do this update if it wasn't needed - im just fallowing arch way .

Last edited by Vi0L0 (2010-08-30 13:39:59)

Vi0L0 · 2010-08-28 20:15:21

[catalyst] repo updated:
- catalyst is now supporting 2.6.36 rc2 kernel
- lib32-catalyst-utils (mentioned above)

REQUEST:
can somebody tell me why lib32-libstdc++5 was in deps array of lib32-catalyst-utils? Was/Is it required by any 32-bit app (even if it is - it's still no dep for lib32-catalyst-utils)?

I have also removed libstdc++5 package (it's dep for catalyst) from system and see no difference...

Now i think it's just prehistoric dep that should be removed...

Edit: Bam! In catalyst release notes i can see that catalyst depend on libstdc++, still i dont know why...
ah now i can see:
$ ldd /usr/bin/amdcccle | grep stdc++
libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x00007f00ffb44000)
$ pacman -Qo /usr/lib/libstdc++.so.6
/usr/lib/libstdc++.so.6 is owned by gcc-libs 4.5.1-1

Ok, so it's indeed prehistoric depend and it'll be removed from deps array (now on aur, repo will be updated when time will come ), anyhow
feel free to remove libstdc++5 (and lib32-libstdc++5) package from your system (ofcourse if it's not required by other package).

Last edited by Vi0L0 (2010-08-28 22:21:58)

DarksideEE7 · 2010-08-29 01:50:13

So I'm having some difficulty getting the proper refresh rate since I reinstalled Arch on my desktop (wanted it on a RAID0 array for speed). I'm using the latest catalyst and lib32-catalyst-utils with my 4870x2. My desktop is sluggish with kwin switching desktops, and when I run glxgears with vsync enabled I get 30FPS. I disabled it and got a high FPS as usual.

This is all coming from a fully working system that I had installed on my external esata drive...........I got 60FPS in glxgears with vsync and I could even play Starcraft2 with good FPS.

So I added your [catalyst] repo and I'm going back to 10.7 for now, hoping that will fix it. Any tips on forcing vsync if this doesn't work? I don't really know what I did different (if it wasn't simply the new catalyst).

EDIT:

Which xorg-server-catalyst-maximize-fix should be installed with Catalyst 10.7 from your repo? I got some errors after just uninstalling my current xorg-server, so I'm removing all xorg packages and starting from scratch to be sure.

I went ahead and forced it....hopefully there will be no conflicts that actually yield a bug. Knowing my luck the last two days I won't get along very far without a crash.

Last edited by DarksideEE7 (2010-08-29 02:31:49)

Vi0L0 · 2010-08-29 02:47:51

DarksideEE7 wrote:

So I'm having some difficulty getting the proper refresh rate since I reinstalled Arch on my desktop (wanted it on a RAID0 array for speed). I'm using the latest catalyst and lib32-catalyst-utils with my 4870x2. My desktop is sluggish with kwin switching desktops, and when I run glxgears with vsync enabled I get 30FPS. I disabled it and got a high FPS as usual.
This is all coming from a fully working system that I had installed on my external esata drive...........I got 60FPS in glxgears with vsync and I could even play Starcraft2 with good FPS.
So I added your [catalyst] repo and I'm going back to 10.7 for now, hoping that will fix it. Any tips on forcing vsync if this doesn't work? I don't really know what I did different (if it wasn't simply the new catalyst).

EDIT:
Which xorg-server-catalyst-maximize-fix should be installed with Catalyst 10.7 from your repo? I got some errors after just uninstalling my current xorg-server, so I'm removing all xorg packages and starting from scratch to be sure.
I went ahead and forced it....hopefully there will be no conflicts that actually yield a bug. Knowing my luck the last two days I won't get along very far without a crash.

Yes, this well known v-sync bug looks like a joke :-) (good job ati!) but v-sync is working fine in wine - at least for me.

Well actually archive catalyst 10.7 repo still exist - change [catalyst] to [catalyst-10.7] in pacman.conf - just know that the lib32-catalyst-utils package from that repo won't work fine with new multilib repo - although i think i will rebild it cuz theres a lot of people here complaining about 10.8... i will do this tomorrow - now gotta sleep (-.-)Zzz

msx · 2010-08-29 02:56:17

Hey Vi0L0 thank you *very* much for your awesome work.

@DarksideEE7: currently I have deactivated V-Sync and everything's working ok. If you're using KDE SC you *must* add --graphicssystem raster to your plasma desktop and yakuake in order to make them usable.
About the FPS, what tool are you running to check them, glxgears?

Last edited by msx (2011-02-07 05:51:11)

DarksideEE7 · 2010-08-29 02:58:28

Vi0L0 wrote:

DarksideEE7 wrote:
So I'm having some difficulty getting the proper refresh rate since I reinstalled Arch on my desktop (wanted it on a RAID0 array for speed). I'm using the latest catalyst and lib32-catalyst-utils with my 4870x2. My desktop is sluggish with kwin switching desktops, and when I run glxgears with vsync enabled I get 30FPS. I disabled it and got a high FPS as usual.
This is all coming from a fully working system that I had installed on my external esata drive...........I got 60FPS in glxgears with vsync and I could even play Starcraft2 with good FPS.
So I added your [catalyst] repo and I'm going back to 10.7 for now, hoping that will fix it. Any tips on forcing vsync if this doesn't work? I don't really know what I did different (if it wasn't simply the new catalyst).

EDIT:
Which xorg-server-catalyst-maximize-fix should be installed with Catalyst 10.7 from your repo? I got some errors after just uninstalling my current xorg-server, so I'm removing all xorg packages and starting from scratch to be sure.
I went ahead and forced it....hopefully there will be no conflicts that actually yield a bug. Knowing my luck the last two days I won't get along very far without a crash.
Yes, this well known v-sync bug looks like a joke :-) (good job ati!) but v-sync is working fine in wine - at least for me.
Well actually archive catalyst 10.7 repo still exist - change [catalyst] to [catalyst-10.7] in pacman.conf - just know that the lib32-catalyst-utils package from that repo won't work fine with new multilib repo - although i think i will rebild it cuz theres a lot of people here complaining about 10.8... i will do this tomorrow - now gotta sleep (-.-)Zzz

Yea that sucks. So you're saying it's a known bug to get 30Hz while even on the desktop? If that's true then it's just pitiful. C'mon ATI............

So I just said screw it and I'm wiping my whole system since I started installing last night in the middle of them changing to [multilib], so I had a bunch of missing packages while building my base system for WINE and other very important packages for my system. Everything was really out of order and scattered, and I don't like that for a base install. I like to have everything in perfect order and know exactly what depends on what, so now that [multilib] is in much better order I'm rebuilding my system.

What could I do to get around the problem with the multilib repo? I mean I could just edit the depends and conflicts section of the PKGBUILD for lib32-catalyst-utils and compile using gcc-multilibs, right?
If there's something I can do to help let me know. I know maintaining ATI's mess is probably a lot of work so if I would be glad to help out......I have a little development experience and I'm currently enrolled in a c++ college course for EE, but I'm no C pro mind you.

Last edited by DarksideEE7 (2010-08-29 02:59:38)

Vi0L0 · 2010-08-29 03:25:41

Archive repositories section of Catalyst wiki has been updated with catalyst-10.7 repo infos.

DarksideEE7 wrote:

Yea that sucks. So you're saying it's a known bug to get 30Hz while even on the desktop? If that's true then it's just pitiful. C'mon ATI............

Well it looks like true - users from other forums also complain about that. And i really doubt that i screwed something in PKGBUILD since original ati's archives of 10.7 and 10.8 contain same files - i mean the list of files is same, but there's a lot of differences in files. Before building i'm always checking differences between original ati's archives with basic 'diff -rq extracted_archive1 extracted_archive2 | sort' command

DarksideEE7 wrote:

What could I do to get around the problem with the multilib repo? I mean I could just edit the depends and conflicts section of the PKGBUILD for lib32-catalyst-utils and compile using gcc-multilibs, right?

Well i've just updated [catalyst-10.7] repo with new lib32-catalyst-utils-10.7-2 package . And the answer for you question is - yes, basically yes, but there's no compilation in this package - just moving files from archive to more appropriate place . Plus since AUR's lib32-catalyst-utils (10.8) is working with multilib - it's pretty easy to copy-paste it's PKGBUILD and just change pkgver and md5sums arrays and then build a 10.7 package.

DarksideEE7 wrote:

If there's something I can do to help let me know. I know maintaining ATI's mess is probably a lot of work so if I would be glad to help out......I have a little development experience and I'm currently enrolled in a c++ college course for EE, but I'm no C pro mind you.

Ok

Last edited by Vi0L0 (2010-08-29 03:52:21)

DarksideEE7 · 2010-08-29 03:54:09

Vi0L0 wrote:

DarksideEE7 wrote:
Yea that sucks. So you're saying it's a known bug to get 30Hz while even on the desktop? If that's true then it's just pitiful. C'mon ATI............
Well it looks like true - users from other forums also complain about that. And i really doubt that i screwed something in PKGBUILD since original ati's archives of 10.7 and 10.8 contain same files - i mean the list of files is same, but there's a lot of differences in files. Before building i'm always checking differences between original ati's archives with basic 'diff -rq extracted_archive1 extracted_archive2 | sort' command
DarksideEE7 wrote:
What could I do to get around the problem with the multilib repo? I mean I could just edit the depends and conflicts section of the PKGBUILD for lib32-catalyst-utils and compile using gcc-multilibs, right?
Well i've just updated [catalyst-10.7] repo with new lib32-catalyst-utils-10.7-2 package . And the answer for you question is - yes, basically yes, but there's no compilation in this package - just moving files from archive to more appropriate place . Plus since AUR's lib32-catalyst-utils (10.8) is working with multilib - it's pretty easy to copy-paste it's PKGBUILD and just change pkgver and md5sums arrays and then build a 10.7 package.
DarksideEE7 wrote:
If there's something I can do to help let me know. I know maintaining ATI's mess is probably a lot of work so if I would be glad to help out......I have a little development experience and I'm currently enrolled in a c++ college course for EE, but I'm no C pro mind you.
Ok

Awesome. I just finished installing the base system, setting up sshd, nfs, etc. and kde. Now it's time to install catalyst-10.7 and xorg-server-maximize-fix. So what are the differences between xorg-server-backclear and xorg-server-catalyst-maximize-fix? I see that the wiki says:

[catalyst] repository contains xorg-server-backclear (patched with backclear patch) and xorg-server-1.8-catalyst-maximize-fix (patched with fedora patch) packages. Both this patches are fixing problems with maximizing/resizing but they are doing it in different way - you may choose which patched xserver is best for you.

.....any insight on that?

Last edited by DarksideEE7 (2010-08-29 03:54:36)

Vi0L0 · 2010-08-29 04:02:01

DarksideEE7 wrote:

So what are the differences between xorg-server-backclear and xorg-server-catalyst-maximize-fix? I see that the wiki says:
[catalyst] repository contains xorg-server-backclear (patched with backclear patch) and xorg-server-1.8-catalyst-maximize-fix (patched with fedora patch) packages. Both this patches are fixing problems with maximizing/resizing but they are doing it in different way - you may choose which patched xserver is best for you.
.....any insight on that?

Umm... no. You just need to choose you destiny . Install first, check it, remove it, install second, compare. I personally preferred xorg-server-backclear cuz it was generating less "trashes". ok, now i rly need to go asleep, gn

Last edited by Vi0L0 (2010-08-29 04:02:49)

DarksideEE7 · 2010-08-29 05:01:19

Vi0L0 wrote:

DarksideEE7 wrote:
So what are the differences between xorg-server-backclear and xorg-server-catalyst-maximize-fix? I see that the wiki says:
[catalyst] repository contains xorg-server-backclear (patched with backclear patch) and xorg-server-1.8-catalyst-maximize-fix (patched with fedora patch) packages. Both this patches are fixing problems with maximizing/resizing but they are doing it in different way - you may choose which patched xserver is best for you.
.....any insight on that?
Umm... no. You just need to choose you destiny . Install first, check it, remove it, install second, compare. I personally preferred xorg-server-backclear cuz it was generating less "trashes". ok, now i rly need to go asleep, gn

All right.........well I suppose I'll take the red pill )

Gillian00 · 2010-08-29 08:47:45

msx wrote:

Hey Vi0L0 thank you *very* much for your awesome work.
@DarksideEE7: currently I have deactivated V-Sync and everything working ok. If you're using KDE SC you *must* add --graphicssystem raster to your plasma desktop and yakuake in order to make them usable.
About the FPS, what tool are you running to check them, glxgears?

how do you do that ??

thanks for your work Vi0L0 :-)

EDIT: Ok, I did it with this: https://bbs.archlinux.org/viewtopic.php?id=81677 I don't understand why can with use the option OpenGl instead of raster ? wouldn't it be still faster ?

Last edited by Gillian00 (2010-08-29 11:59:26)

spinmc · 2010-08-29 14:06:10

Hello, to everyone.
I use Arch from a long time.
A very big problem is a memory leak that affects the X process (in "top") whenever I use the catalyst proprietary drivers.
With KDE 4.4 and 4.5 "free -m" tells me that it eats about 1, 3 GB of ram in less than an hour.
With XFCE it arrives to 700 MB in 2 hours with no programs loaded.
The "top" command tells me that the X process uses about 4~5 % of memory. These are very huge values.
When I try to use the open source drivers everything comes back in normal values (but I *must* to use catalyst for dynamic power management).
I saw this problem since catalyst 10.4.
Have you found a solution for this?!? Is this a problem related to arch x86_64 only?
My graphic card is a mobility radeon HD 4570 on a dell studio 1555.
Thank you so much for your help.

Vi0L0 · 2010-08-29 21:02:41

[catalyst-10.7] repo update: catalyst package should now support all released kernels. But whats more important: directories in this package aren't conflicting with xvba-video and libva packages now.
IMPORTANT NOTE:
If you are using libva and xvba-video packages then you need to reinstall them, and type this as root:
rm /usr/lib/xorg/modules/dri/*_drv_video.so

I've spent some time trying to fix vsync regression in 10.8. I tried many configurations, checked PKGBUILD many times, i tried to replace files or options, i asked on irc - NOTHING... absolutely no idea why is that happening.

Summary:
I'm using catalyst for 1.5 year, never had such bug before.
BUG:
- if v-sync is enabled - opengl applications/games are generating only half of frames/second (fps), ie. if screen sync = 60 - then fps = 30, if screen sync = 75 - then fps = 38;
- i've checked it on gnome 2.30.2, kde 4.4.5/4.5 with compositing on/off - it occurs always;
- problem is touching every native opengl application/game _AND_ compositing - i'm mainly using kde composition manager (kwin) and even when vsync in kwin is disabled and v-sync in amdcccle is NOT set as "Always On" [v-sync is set as (forgive translation): "Enabled until application disable it"] it looks like catalyst 10.8 is somehow forcing this buggy v-sync to be enabled - and it's really really annoying laggy... plus i need to switch v-sync on/off whenever i want to watch_video_through_vaapi:gl/use_my_compositing_desktop;
NOTE:
- v-sync is working fine for wine games (wine in version 1.3.1)

So yes - it's my bugreport that i'm gonna post on ati's bugzilla. Ok, bugreport is here, feel free to sign over this .

spinmc wrote:

Hello, to everyone.
I use Arch from a long time.
A very big problem is a memory leak that affects the X process (in "top") whenever I use the catalyst proprietary drivers.
With KDE 4.4 and 4.5 "free -m" tells me that it eats about 1, 3 GB of ram in less than an hour.
With XFCE it arrives to 700 MB in 2 hours with no programs loaded.
The "top" command tells me that the X process uses about 4~5 % of memory. These are very huge values.
When I try to use the open source drivers everything comes back in normal values (but I *must* to use catalyst for dynamic power management).
I saw this problem since catalyst 10.4.
Have you found a solution for this?!? Is this a problem related to arch x86_64 only?
My graphic card is a mobility radeon HD 4570 on a dell studio 1555.
Thank you so much for your help.

wow! For me memory leaks disappeared and i don't even know why - now X is taking less than 100M. I found bugreport related to your problem here: https://bugs.launchpad.net/ubuntu/+sour … bug/391628, unfortunately looks like theres no solution .

Last edited by Vi0L0 (2010-08-29 21:42:58)

dcc24 · 2010-08-29 21:14:03

Memory leaks are always present with catalyst, although not as severe as spinmc reports.

I've just made a quick test. The only open apps are: Firefox with 5 tabs, tilda, pidgin. There are also some non-X background daemons, but they are not relevant.

With Catalyst 10.7 => RAM usage is 545MB after 10 min of idle time. 250MB of this is from X.
With xf86-video-ati 6.13.1 =>RAM usage is 390MB after 10 min of idle time. 29MB of this is from X.

So there is a memory leak that is related to catalyst.

spinmc · 2010-08-29 22:54:42

Thank you for your answers and for your work in packaging this driver.
Yes, I know it's related to catalyst, so I will hope that new drivers could solve finally this problem.
Thank you again.

Vi0L0 · 2010-08-30 20:30:33

wooohooo! I've got great news for you!
GPU acceleration in newest chromium works absolutely amazingly great!

Even Microsoft's Demos for IE9 (to show ie9 acceleration like this one) works with highest framerate!

But... whats more important here in normal desktop using is that with it and v-sync enabled in amdcccle flash videos are tear-free now! No tearing/flickering at all with compositing enabled and at full screen! (tested with flashplugin and nspluginwrapper from multilib repo, and with both catalyst 10.7 and 10.8)

I'm gonna post you my PKGBUILD for newest chromium-dev in minutes so you may build it,
Edit: PKGBUILD with other needed files is here. Note that you need to download ~124M fat sources archive file to build it.
And maybe i'll also upload a package for x86_64 (within a ~10 minutes).
Edit: ok, here's the package, it's only ~16M fat, but i'm not promising it will work - maybe linker linked some specific for my system libs.

I'm running chromium-dev with those arguments:

chromium-dev --enable-gpu-plugin --enable-accelerated-compositing --enable-accelerated-2d-canvas

You can run it with webgl support:

chromium-dev --enable-gpu-plugin --enable-accelerated-compositing --enable-accelerated-2d-canvas --enable-webgl

and watch ie. this demo ;P

Ofcourse you can also add these arguments to Exec line of /usr/share/applications/chromium-dev.desktop file, here's example with extensions enabled:

Exec=chromium-dev %U --enable-gpu-plugin --enable-accelerated-compositing --enable-accelerated-2d-canvas --enable-webgl --enable-extensions

Important Edit: Ive found that running it with --enable-accelerated-2d-canvas argument could cause to system crash while viewing some pages like http://images.google.pl so feel free to remove it (although performance on some pages like those ms tests will drop)

Btw i've builded this chromium-dev package when i was using/testing open source drivers - but it should work good on catalyst, at least it's working good for me.
Ps. oss drivers are good, better than when i was using them before (0,5year ago) but i was catching some problems with flash videos when watching them on fullscreen (laggy after some time), also this gpu acceleration doesn't work on them.

cheers!

Last edited by Vi0L0 (2010-08-31 01:30:29)

monshin · 2010-08-31 01:22:44

@Vi0L0: Chromium-dev works great here, and it's FAST! Thank you for the effort.

Vi0L0 · 2010-08-31 01:40:04

monshin wrote:

@Vi0L0: Chromium-dev works great here, and it's FAST! Thank you for the effort.

Indeed, chromium was always fastest browser on my machine, but now it's just "wow" .
Too bad that we probably won't see any kind of gpu acceleration in firefox 4, i prefer ff because of it's add-ons.
Edit: Well... at least there's adblock and notScripts extension for chromium.

Last edited by Vi0L0 (2010-08-31 02:04:35)

codycarey · 2010-08-31 01:40:19

dcc24 wrote:

Memory leaks are always present with catalyst, although not as severe as spinmc reports.
I've just made a quick test. The only open apps are: Firefox with 5 tabs, tilda, pidgin. There are also some non-X background daemons, but they are not relevant.
With Catalyst 10.7 => RAM usage is 545MB after 10 min of idle time. 250MB of this is from X.
With xf86-video-ati 6.13.1 =>RAM usage is 390MB after 10 min of idle time. 29MB of this is from X.
So there is a memory leak that is related to catalyst.

Psh, step back so a man can show you what a REAL memory leak looks like...

  cody@tritium~ $ uptime
 21:37:11 up  7:50,  6 users,  load average: 0.64, 0.48, 0.37

  cody@tritium~ $ sudo pmap -d 9613
9613:   /usr/bin/X :0 vt7 -nolisten tcp -auth /var/run/xauth/A:0-LhSUta
Address           Kbytes Mode  Offset           Device    Mapping
0000000000400000    1796 r-x-- 0000000000000000 008:00003 Xorg
00000000007c1000      72 rw--- 00000000001c1000 008:00003 Xorg
00000000007d3000      64 rw--- 0000000000000000 000:00000   [ anon ]
0000000002428000 1305088 rw--- 0000000000000000 000:00000   [ anon ]
< snip >
mapped: 1550116K    writeable/private: 1338788K    shared: 93400K

Last edited by codycarey (2010-08-31 01:47:08)

unideal · 2010-08-31 05:55:57

Hello fellas!

Do you know if ATI FirePro 2460 plays well when using Arch? It has Linux support mentioned on it's manual, but there seems to be lacking user discussions about that matter. How are the drivers of ATI/AMD workstation cards?

http://www.amd.com/us/products/workstat … 60-mv.aspx

My other option is ASUS 9600GT SILENT, but it has a lot of unneeded features in my eyes. I'm looking for a card to support >= 2 monitors with DVI (or here, DP -> DVI-I passive). FirePro would also run cooler since it only uses fraction of the watts of 9600GT. I'll be using it for my mATX desktop PC running Arch x86_64 at home, so pretty much lightweight usage: programming, web browsing, some multimedia (but Core i7 will decode those), simple photo editing etc.

mDuo13 · 2010-08-31 07:17:51

Upgraded to Catalyst 10.8. Dunno what I was thinking - perhaps hoping for some reason that ATI would fix bugs rather than introduce new ones. I guess that was naive of me. Blackboxes bug didn't go away (fixed it by reverting to old 2D engine as previously mentioned) but now Starcraft 2 won't run under WINE. Getting messages like this:

 II.exe: xcb_io.c:506: _XReply: Assertion `!dpy->xcb->reply_data' failed.

(It seems cut-off; I've seen a couple variations, with different amounts of the to Starcraft II.exe trimmed from the start)

Anyway, Googling for the error message didn't turn up anything helpful, but I thought I'd ask around. Anyone else seeing this problem? Anyone else running SC2 successfully under Catalyst 10.8? If so, what card do you have? (Mine's a 5770.)

Arch Linux

#2276 2010-08-28 16:29:56

Re: The AMD/ATI Bar & Grill

#2277 2010-08-28 16:37:17

Re: The AMD/ATI Bar & Grill

#2278 2010-08-28 16:58:52

Re: The AMD/ATI Bar & Grill

#2279 2010-08-28 17:47:35

Re: The AMD/ATI Bar & Grill

#2280 2010-08-28 17:48:21

Re: The AMD/ATI Bar & Grill

#2281 2010-08-28 20:15:21

Re: The AMD/ATI Bar & Grill

#2282 2010-08-29 01:50:13

Re: The AMD/ATI Bar & Grill

#2283 2010-08-29 02:47:51

Re: The AMD/ATI Bar & Grill

#2284 2010-08-29 02:56:17

Re: The AMD/ATI Bar & Grill

#2285 2010-08-29 02:58:28

Re: The AMD/ATI Bar & Grill

#2286 2010-08-29 03:25:41

Re: The AMD/ATI Bar & Grill

#2287 2010-08-29 03:54:09

Re: The AMD/ATI Bar & Grill

#2288 2010-08-29 04:02:01

Re: The AMD/ATI Bar & Grill

#2289 2010-08-29 05:01:19

Re: The AMD/ATI Bar & Grill

#2290 2010-08-29 08:47:45

Re: The AMD/ATI Bar & Grill

#2291 2010-08-29 14:06:10

Re: The AMD/ATI Bar & Grill

#2292 2010-08-29 21:02:41

Re: The AMD/ATI Bar & Grill

#2293 2010-08-29 21:14:03

Re: The AMD/ATI Bar & Grill

#2294 2010-08-29 22:54:42

Re: The AMD/ATI Bar & Grill

#2295 2010-08-30 20:30:33

Re: The AMD/ATI Bar & Grill

#2296 2010-08-31 01:22:44

Re: The AMD/ATI Bar & Grill

#2297 2010-08-31 01:40:04

Re: The AMD/ATI Bar & Grill

#2298 2010-08-31 01:40:19

Re: The AMD/ATI Bar & Grill

#2299 2010-08-31 05:55:57

Re: The AMD/ATI Bar & Grill

#2300 2010-08-31 07:17:51

Re: The AMD/ATI Bar & Grill

Board footer