The Official ATI Linux Issues Thread

[WhO_KnOwS]

This is the issues list. The old one can be found here. Please use this thread ONLY for posting/discussing issues. If you see an issue is already reported, but you feel it is an important one, just post it, this way I will raise the Issues Impact. Also, if possible check all the issues from the old drivers and report any new info (fixed, improved, same as before...). This will facilitate ATI's job at making a better driver.
Any suggestions about the list should be PM-ed to me.

Also, this here is the official request poll. I tried to list all the most requested stuff in it.

When posting an issue please specify:

• -motherboard (Example: nforce2)
  -gfx card (Example: Radeon 9500 Pro)
  -kernel version (Example: 2.4.22)
  -any special kernel patches (Example: nforce support for 2.4.20)
  -driver version (Example: 3.2.8)
  -where did you get them (Example: Gentoo-portage, www.ati.com, Mandrake power pack)
  -distro and version (Example: Mandrake 9.2)
  -If possible include a screenshot of the issue/bug

ATI is keeping a close eye on the thread, so all the issues on the list are being looked into.

____________________________________
Currently known issues

===============

IMPORTANT
Problem : Bad performance
Applications Affected : All 3D apps
Cards Affected : All
Versions : All Kernels, All XFree, All driver - 3.7.0 in particular
Distributions : All
Impact : High, All Users
Workaround :

none

===============

IMPORTANT
Problem : Running two or more X sessions causes graphical corruption followed by a system freeze
Applications Affected : X11
Cards Affected : All
Versions : All Kernels, All XFree, All drivers
Distributions : All
Impact : High, All Users
Workaround :

For the screen corruption there is currently no workaround. For the colormap corruption run the following command:

fglrx_xgamma -gamma 1.0

To redraw the screen (instead of pulling a window around) you can run:

xfrefresh

Comments :

The system freeze only happens in KDE, kdm and gdm are unaffected, Gnome suffers from some gfx artifacts (fixed by moving windows over them). It can be seen on dark colors (AA fonts for example).
This are in fact two separate issues that ATI is aware of. One is a colormap corruption, the other one is the known screen corruption.

Unfixed

===============

Problem : Artifacts appearing in games
Applications Affected : Enemy Territory, other 3D games, some demos
Cards Affected : All
Versions : All Kernels, All XFree, 3.2.x driver
Distributions : All
Impact : High, All Users
Workaround :

none

Comments :

These artifacts flash randomly, so it is hard to get a screenshot. They happen more often in maps that have foliage (trees). It must be also noted that similar artifacts appear in OpenGL 1.4 demos.

Still present in 3.7.6

===============

Problem : Artifacts appearing in games when FSAA enabled
Applications Affected : UT2003
Cards Affected : Radeon 9x00
Versions : 2.4.22 Kernel, XFree 4.3.0, 3.2.8 driver (possibly older)
Distributions : Fedora, Gentoo (probably all)
Impact : Medium, 3 users (tested, same issue here)
Workaround :

Typing "flush" in the game console, flushes the buffers, which fixes the issue.

Comments :

The higher the AA level, the faster the issue appears.

===============

Problem : AGPGart problems with NForce motherboards
Applications Affected : X11
Cards Affected : All
Versions : Pre-2.6 Kernel without a suitable kernel patch
Distributions : All
Impact : High, All Users
Workaround :

Either switch to 2.6 Kernel, or install a suitable kernel patch. Then enable the apropriate module.

Comments :

Instructions on how to install the kernel patch are availible in the HowTo (link in my sig).

===============

Problem : DRI doesn't work
Applications Affected : All DRI dependant applications
Cards Affected : 8500 (possibly others)
Versions : 2.6 Kernel, XFree 4.3.0 (possibly others)
Distributions : All
Impact : Low, Some 8500 users
Workaround :

none

===============

Problem : Xinerama support causes 3D acceleration to stop working
Applications Affected : Xinerama or all 3D accelerated
Cards Affected : All
Versions : All Kernels, All XFree
Distributions : All
Impact : Medium, All Users
Workaround :

Dissabling Xinerama re-enables 3D acceleration.

===============

Problem : 2D acceleration is too slow
Applications Affected : X11
Cards Affected : All
Versions : All Kernels, XFree 4.3.0 (possibly others)
Distributions : All
Impact : Medium, All Users
Workaround :

none

===============

Problem : Possible memory leak
Applications Affected : X11
Cards Affected : All
Versions : All Kernels, All XFree
Distributions : All
Impact : Medium, All Users
Workaround :

none

Comments :

Ctrl + Alt + Backspace causes the memory leak. After the reset, X switches to software rendering. Also, fglrx gets a higher "used by" after every X restart. Looks like something is not unloaded correctly.

===============

Problem : Several TV-Out issues
Applications Affected : X11
Cards Affected : All
Versions : All Kernels, All XFree
Distributions : All
Impact : Medium, All Users
Workaround :

none

===============

Problem : Hardlock running several 3D applications
Applications Affected : All 3D applications
Cards Affected : All
Versions : All Kernels, All XFree, Only 3.2.x drivers
Distributions : All
Impact : Low, Some users
Workaround :

Use pre 3.2.x drivers.

===============

Problem : Crashes when having more than 2 submenus opened
Applications Affected : KDE, possibly other X window managers
Cards Affected : 8500 (other cards not reported so far)
Versions : Any kernel, XFree 4.3.0 (possibly others)
Distributions : All
Impact : Low, 8500 users
Workaround :

none

===============

Problem : Framebuffer causes konsole corruption
Applications Affected : Konsole
Cards Affected : All
Versions : All, using ATI framebuffer
Distributions : All
Impact : High, All Users
Workaround :

Use VESA framebuffer instead.

===============

Problem : Running a 2D qt application reports an error. Result is software rendering in 3D apps
Applications Affected : All 2D qt
Cards Affected : All
Versions : All
Distributions : All
Impact : Low, Some Users
Workaround :

Look here

===============

Problem : No FSAA in dual-head
Applications Affected : All
Cards Affected : All
Versions : All
Distributions : All
Impact : Low, All Users
Workaround :

none

===============

Problem : DRI initialization fails at boot
Applications Affected : X11
Cards Affected : All
Versions : Kernel 2.6-r7 and later, XFree all versions
Distributions : All
Impact : Medium, All Users
Workaround :

On SIS motherboards it can be fixed with a patch. Check the patches thread.

Loading the DRI module at bootup with a bootup script.

===============

Problem : RandR broken
Applications Affected : xrandr, core X11
Cards Affected : All
Versions : All Kernels, XFree 4.3
Distributions : All
Impact : Low, All Users
Workaround :

Disable DGA, in XF86Config-4 find the

Section "Module"

and disable DGA by uncommenting the following file.

Option "omit xfree86-dga"

===============

FIXED IN 3.7.6
Problem : Support for kernel 2.6 is lacking
Applications Affected : X11
Cards Affected : All
Versions : Kernel 2.6
Distributions : All except Gentoo
Impact : Medium, All Users
Workaround :

none

Comments :

ATI suggested that some changes would be done for the next driver.

To compile a working driver, you still need:
- to use the VMALLOC patch
- to correct firegl_public.c : line 3123
#if ( (PAGE_ATTR_FIX == 1) || (LINUX_VERSION_CODE == KERNEL_VERSION(2,4,19)) )
should be replaced with :
#if ( (PAGE_ATTR_FIX == 1) || (LINUX_VERSION_CODE >= KERNEL_VERSION(2,4,19)) )

===============

Problem : Missing support for 16-bit colors
Applications Affected : X11
Cards Affected : All
Versions : All Kernels, All XFree, All drivers
Distributions : All
Impact : Medium, All Users
Workaround :

none

Comments :

This has been an issue since the first release. Hopefully with the community demand high enough it will be added.
No promises from ATI about this.

===============

Problem : Radeon 9000 and 9600 no FSAA support
Applications Affected : X11
Cards Affected : All 9000 and 9600
Versions : All Kernels, All XFree, All drivers
Distributions : All
Impact : Medium, All 9000 Users
Workaround :

none

Comments :

Will tell ATI about it. Expect an update soon.

===============
FIXED IN 3.7.6
Problem : WineX crashes and isn't hardware accelerated
Applications Affected : WineX
Cards Affected : All
Versions : All Kernels, All XFree, 3.7.0 drivers
Distributions : All
Impact : High, All users
Workaround :


===============

IMPORTANT
Problem : Dual Head several problems
Applications Affected : X11
Cards Affected : All
Versions : All Kernels, All XFree, All drivers
Distributions : All
Impact : High, All users
Workaround :

none

Comments :

-Xv doesn't work in Dual Head
-"sync gl buffer swaps to vertical refresh" only looks at first head
-Xinerama doesn't work with Dual Head
-Artifacts appearing on the second monitor with 3.7.0 drivers

===============

Problem : Radeon 9600 XT gets recognised as a Radeon 9600 Pro
Applications Affected : X11
Cards Affected : 9600 XT
Versions : All Kernels, All XFree, All drivers
Distributions : All
Impact : Low, 9600 XT user(s)
Workaround :

Getting a ChipID for 9600 XT might help. No luck so far

===============

Problem : Mobility Radeons cause system freeze
Applications Affected : X11
Cards Affected : Mobility Radeons
Versions : All Kernels, All XFree, Newer drivers
Distributions : All
Impact : High, Mobility users
Workaround :

none

===============

Problem : Mouse cursor in KDE not dissapearing on top
Applications Affected : KDE
Cards Affected : All
Versions : All Kernels, All XFree, All Drivers
Distributions : All
Impact : Low, All users
Workaround :

See here

===============

Problem : Driver does not recognize the DVI port
Applications Affected : X
Cards Affected : All
Versions : All Kernels, All XFree, All Drivers
Distributions : Gentoo, probably all
Impact : High, Some users
Workaround :

none

===============

Wishlist (listing "Other" answers from the poll)

Bold entries are new

• -Support for cards older than 8500
  -HW accelerated XRENDER IMPORTANT
  -Extended control panel (profiler, setup of XF86Config, Windows-like driver control)
  *Introduce the option of changing AA, Vsync and AF on the fly
  *Enable to keep gamma settings between reboots (even if the module doesn't get unloaded)
  *ATI says they will do something with the control panel in the future
  -Allow the usage of Xinerama and DRI at the same time
  -Buffer swaps synchronizations looking at the second head also (not only first one)
  -A kernel patch that would allow Radeon Framebuffer to work correctly
  -TV-Out on R200 cards IMPORTANT (still missing in 3.7.0)

[gentooalex]

Speed is a very important issue for me! I can't play UT at 14 fps!! Make the drivers FASTER . Performance should be addressed before you start adding on the finishing touches. PLEASE ADD SPEED TO THE TOP OF THE LIST!!!! Please DON'T CHANGE the setup to make it easy, its great. the way it is, spend time on other things.

Who_Knows: Can you make it so I can edit the list?

[WhO_KnOwS]

Who_Knows: Can you make it so I can edit the list?

Only I (and the Mods) can modify the list. I don't see any good coming out of more people modifying the list. If you have a suggestion about it, PM me. And no chatting please.

[gentooalex]

Originally posted by WhO_KnOwS
Only I (and the Mods) can modify the list. I don't see any good coming out of more people modifying the list. If you have a suggestion about it, PM me. And no chatting please.

Most users have complained that games run too slow to play well. I think you should add speed to the top of the list.

[WhO_KnOwS]

Updated. I'm thinking about removing some of the wishes from the Wishlist, as they are in the poll --> the community has direct control of it.

[kihaji]

Issue and a few questions:

Issue - I still want Xinerama + DVI support. I think this is a key selling point. The number of dual monitor people is increasing, and this is a big factor.

Question 1: With the release of the next Xfree soon, are these drivers tested against the latest release candidate? If not, what is the possibility of getting an early update to fix any small compatibilty issues with it that may arise?

Question 2: Possibility of getting a formal bugzilla set up? I'm not certain how the current beta testing goes, but I believe that a public bugzilla, where everyone can see and search the bugs out there, can help immensely. People may find a quick fix to help others without an update from perusing them, or give insight to dev's on how to fix them, and beta testers can go through and test the issues with the newest drivers to see if they do in fact fix them.

Question 3: Posibility of taking a kernel style of updating, IE have a testing branch open to everyone and a stable branch, every 2 months they can take the last good test release, polish it up a bit, and release it, while those that are bored can use the beta versions and give a lot more feedback.


Basically, I don't mind that the development is closed, I even understand it and support it, but I do wish the testing would be more open.

[gentooalex]

Originally posted by kihaji
Issue and a few questions:

Issue - I still want Xinerama + DVI support. I think this is a key selling point. The number of dual monitor people is increasing, and this is a big factor.

Question 1: With the release of the next Xfree soon, are these drivers tested against the latest release candidate? If not, what is the possibility of getting an early update to fix any small compatibilty issues with it that may arise?

Question 2: Possibility of getting a formal bugzilla set up? I'm not certain how the current beta testing goes, but I believe that a public bugzilla, where everyone can see and search the bugs out there, can help immensely. People may find a quick fix to help others without an update from perusing them, or give insight to dev's on how to fix them, and beta testers can go through and test the issues with the newest drivers to see if they do in fact fix them.

Question 3: Posibility of taking a kernel style of updating, IE have a testing branch open to everyone and a stable branch, every 2 months they can take the last good test release, polish it up a bit, and release it, while those that are bored can use the beta versions and give a lot more feedback.


Basically, I don't mind that the development is closed, I even understand it and support it, but I do wish the testing would be more open.

I like all the ideas kihaji mentioned and think it would be great to make the testing process more open. THANK YOU ATI!!!!

[FroMan]

This isn't really a problem, since 2.6 isn't officialy out. ATI promised that with 2.6 officialy out, the support will get a lot better.


http://kernel.org/ begs to differ since 2003-12-18.

[panickme]

Should ATI cater to all the early adopters out there or continue to work for what the majority of Linux users will likely be running for at least the next 6 - 8 months? Case in point: the 2.6 kernel. Is it absolutely vital to have the driver functioning perfectly with the new kernel right at release? I don't think so. I don't plan to move to the 2.6 series kernels for at least the next 3 or 4 months depending on what issues I see cropping up on forums and what not. The new XFree versions are a little more important but that has a more frequent release schedule. You just compound the frustration of yourself and others when you put forth all the varied demands that I see being made so far. What's most important to one person may be of zero importance to everyone else. Does ATI spend the developer time to make that one person happy or not?

As for me, I just want Mindrover to work instead of segfaulting.

BTW, does the Nvidia driver function on the 2.6 kernels? Not the newest one that doesn't work at all but the release before it?

[lordvader]

I tried installing the new drivers, and every game, except Enemy Territory hard locks on start-up. I can, however, alt-ctrl-backspace back to the console, so it's not the end of the world. I though it was an issue with the ati drivers, but when I reverted to the DRI drivers, exactly the same thing happened.
It used to work, but I can't figure out what I may have done that will cause this to happen. I'm using kernel 2.4.23-rc3(or 2), the CVS version of XFree86 4.4, and mandrake 9.1. This setup did work at one stage, but now, it doesn't.
The only thing I can think of was my attempt to upgrade to the 2.6 kernel, which resulted in upgrading the module utilities. The 2.6 kernel didn't work, but my old kernel still worked fine. Is it possible that the new module utils case the modules to load somewhat differently?

I'm got an nforce2 mobo, and a radeon 9100.

Cheers.

[spyke]

I dont find these drivers all that bad, they are a good improvement over the last for what I do.

Improvements:
Framebuffer console issues are gone.

Kernel errors that flood about buffer issues are gone.

glxgears now reaches about 2500fps where before it wouldnt break 2000.

I no longer get weird curruption redrawing issues on my desktop. (2D mode) (and yes it does feel somewhar faster)

UT runs happily over 100fps. (not like it didnt before, but)



Now, seriously, these drivers have improvements.
Another thing I keep wondering is why does it seem all the gentoo users are the ones having issues? Are you guys using the most overkill GCC flags? No idea, but you guys really should set your compiler flags and not go overkill with the optimizations. I get tired of seeing gentoo users constantly whining and crying about things not working, when things work perfect on my slackware 9.1 install.

Heck, a friend came up to me the other day and asked why he couldnt install UT.

It seems that the tail program that gentoo uses in the latest coreutils doesnt like the + flag, which is what the shell script in many loki installers call. It sees it as a file, when it should do what +# is supposed to do.

tail +6 (+6 no such file or directory).

And then people wonder why I complain so much.

[Lauwenmark]

(Already submitted in the old thread, resubmitting for completeness)
System used:
AMD Athlon XP 2400+
1.5GB RAM DDR333
Motherboard ABIT NF7-S (NForce 2)
ATI Radeon 9600Pro 128MB
Linux Debian Unstable
Linux Kernel 2.6.0 (standard kernel.org version)

Issues I spotted so far with the new 3.7 drivers:

Kernel warning messages: Debug: sleeping function called from invalid context at mm/slab.c:1856, as reported earlier with 3.2.8 drivers. Looks like none of the changes made in the kernel side part of the driver corrected this.

CTRL+ALT+BackSpace results in memory leak: Same problem as with 3.2.8 drivers.

CTRL+ALT+BackSpace results in switching to software rendering mode: Same problem as with 3.2.8 drivers.

FireGL Control doesn't work: fireglcontrol application always gives the following when launched:

Xlib: sequence lost (0x10000 > 0xa) in reply type 0x0!
X Error: 0 0
Major opcode: 0
Minor opcode: 0
Resource id: 0x80dd720

and never goes further.

Source code building of FireGL Panel Sources doesn't work 'out of the box': Manual edition of the Makefile is needed, both for defining environment vars and to correct some hard-coded library sub-paths. Maybe asking for a plainly standard 'configure' script isn't too much asking ?

I didn't yet encounter the screen corruption problem like with the 3.2.8 drivers - I'll do further tests to see if that got corrected.

Minor irritations:

I also notice that the driver isn't yet available in a tgz format, allowing easier installation in non-RPMs environments.

I note that patching of the kernel-side code is still needed to allow the driver to run on AMD Athlon systems.

I remark that installation script doesn't create a libGL.so symlink to libGL.so.1.2, which is the recommended behaviour.

Kernel 2.6.x doesn't seem to be officially supported with this release (maybe I'm wrong on this). Kernel 2.6 is now officially out for a couple of weeks, and preversions have been available for much longer. (I want to underline what's said in the Currently Known Issues List: ATI promised that with 2.6 officialy out, the support will get a lot better.). (I've seen here and there that it was logical that ATI didn't spend resources on 2.6 support as not a lot of people were using it... May I underline the fact that porting a driver from 2.4 to 2.6 kernel - although non-trivial - doesn't require massive rewriting of the code ? The porting job mostly involves changes in the module interface, not in the inner workings of the driver).

I'm glad that ATI added extended support for OpenGL extensions; but I'd rather have preferred a bigger interest in bug-fixing and stability issues. I would be rather curious to see a full list of improvements made since the last release. I'm also curious about the way some bugs will be handled: shall people have to wait two months for their correction, or can we expect revisions as soon as fixes are ready ?

[lwr]

Documentation.
Drivers should come with instructions on kernel configuration (what should be compiled in, what to compile as modules, what can break the driver e.g. framebuffer). Options in XFree config should be documented too.

ChangeLog
It could be included in drivers package too. I know there is one on ATI's page but having it in the package won't certainly hurt

Tarball
Make it unsupported. Make it unofficial. But make it available This way non-rpm distros (e.g Debian, Gentoo, Slackware) can provide their own installation scripts making it much less painfull to install flgrx drivers. Look at this forum to see how many people have problems with installation.

2.6.0
It's a must. Simple and plain. New kernel is much more responsive and feels generally faster than 2.4 and thus is more appropriate as a gaming platform.

[wedgey__]

Another thing I keep wondering is why does it seem all the gentoo users are the ones having issues? Are you guys using the most overkill GCC flags? No idea, but you guys really should set your compiler flags and not go overkill with the optimizations. I get tired of seeing gentoo users constantly whining and crying about things not working, when things work perfect on my slackware 9.1 install.

I doubt that Gentoo gcc flags have any effect on the driver directly. Unless you edit the kernel makefile, the driver always gets compiled using the standard kernel CFLAGS, which I think we can assume are pretty safe.

If you go nuts with your CFLAGS when compiling X, that may cause problems.

[WhO_KnOwS]

I updated the list and pointed out some important issues.
If you feel other issues already listed should become "important" please post the issue along with your system specs here.
Also, if you have any suggestions about the list, feel free to PM me.

[schoett]

There appears to be no change from driver 3.2.8 regarding my startup problems with the 2.6 kernel:

- with external AGPgart (kernel modules agpgart and sis-agp), the system wedges solid (blank screen, only SysRq keys work, but disk I/O (IDE) never recovers),

- with internal AGPgart, DRI initialisation fails at boot, but works after Ctrl-Alt-Backspace.

Even the external AGPgart can be made to work by first running the card with internal AGPgart and then changing the config, unloading and reloading the modules without reboot.

System: Medion 8008 (ALDI PC, March 2003)
Host bridge: Silicon Integrated Systems [SiS] SiS 645xx (rev 03)
PCI bridge: Silicon Integrated Systems [SiS] SG86C202 (prog-if 00 [Normal decode])
IDE interface: Silicon Integrated Systems [SiS] 5513 [IDE] (prog-if 80 [Master])
Graphics Card: Radeon 9600tx 128mb
VGA compatible controller: ATI Technologies Inc Radeon R300 NF [Radeon 9700] (prog-if 00 [VGA])
Display controller: ATI Technologies Inc Radeon R300 [Radeon 9700] (Secondary)
Kernel Version: 2.6.0
Driver Version: 3.7.0, installed via gentoo-portage
Distro: Gentoo up-to-date, XFree86 4.3.0-r3

[Biazz]

Again, my issues list:

-Xinerama doesn't work. At least not with dual head, 2 separate drivers. You say in your issues list that it doesn't work with DRI. Well, for my setup it doesn't work at all. period.
-Xv doesn't work in dual head
-I still can't enable overscan on TV out, or define my own modeline
-When enabling 'sync gl buffer swaps to vertical refresh' this is always done to the refresh of the 1st head, even though the gl window is on the second head
-2D is still completely unaccellerated (no Xrender extension used) This makes X noticably slow, even on a fast system (I'm using 2.8GHz p4, 1GB of pc1066rambus)
-module still doesn't unload nicely: [fglrx:drm_ioremapfree] *ERROR* [mappings] Attempt to free NULL pointer, loading/unloading repeatedly crashes the system
-less stable than previous release.. 1 crash upto now.. but only 1 day running. Old driver never crashed the system, unless the module was unloaded/loaded repeatedly
-Extremely unproffesional distribution, mainly two reasons:
1 little documentation (none to be precise) If I remember correctly, in the old RPM there was at least some sort of readme included in /usr/share/doc/fglrx
2 confusing warnings during install, because the RPM was packaged with owner set to ptidbit.ptidbit or something.
3 stupid version number. The X release it's for is in the version number, the driver version is in the release number. Obviously, this is not how it should be done; this will give problems with dependency calculations and updates.

I voted 'other' for my main issue (aside from stability, that shouldn't even be an option since stability in a OS such as linux always comes first):

My setup just doesn't work as dual head (1 CRT, 1 PAL-TV). I've got no xinerama, no overlays, no adjustment of TV settings.. no nothing

regards

regards
Bas

Motherboard: Asus P4T-533C
Graphics Card: Club3D Radeon 9700pro 128mb
Kernel Version: 2.4.20-27.9 (redhat stock kernel)
Driver Version: fglrx-glc22-4.3.0-3.7.0
Got them from ati
Distro: Red Hat Linux release 9 (Shrike)

[WhO_KnOwS]

Ok. I'll add Dual Head as a new Issue.
Oh, about the documentation. I think they are writing a new readme to go with the driver. Let's just wait and see.

[LinuxATIUser]

I saw the XRENDER impl. under "less requested features", i would really like this implemented with HW accel. as most of us these days are running X with XFT and antialiased fonts.

There was a test showing that ATI's driver really sucked at rendering antialiased fonts (both DRI and XIG were infinately faster).

[spellbinder]

i still got no shiny water in nwn
is this just me??

[WhO_KnOwS]

Xender pointed out...

[gentooalex]

UT2K3 is really slow compared to the other drivers(3.2.8). The Anatalus level is especially slow with frame rates at 9-15 for me, and I have a pretty fast system.

[Diffie]

I wonder if the AA artifacts issue is related to one of the problems listed here with Radeon 9000 and no FSAA?

For the record i was contacted by ATI person who will take a look at the AA artifacts issue.

[cybowolf]

Personally, I'de like to see a fix to the "Trace/breakpoint trap" issue that occurs with the 9800 XT. Thankfully, with the help of the awesome folks over on the Gentoo boards, I was able to get things working by telling the driver in my XF86Config that my card was a 9800 instead of a 9800 XT by changing the CodeID. This makes me think that it has to be a pretty easy fix, but it sure did make for a lot of headaches trying to work the problem out.

For those interested, details can be seen here:
http://forums.gentoo.org/viewtopic.php?t=118669

-cybowolf

[gentooalex]

Postal 2 goes really slow. (like 3fps !!!) Is this a known issue?

[skirst]

From what I can see, it is already supported, despite the version string saying 1.3. Every extension required by 1.5 is supported by this driver release. Keep in mind, the new shading language extensions are not required by OpenGL 1.5.

[gentooalex]

Originally posted by lwr

Tarball
Make it unsupported. Make it unofficial. But make it available This way non-rpm distros (e.g Debian, Gentoo, Slackware) can provide their own installation scripts making it much less painfull to install flgrx drivers. Look at this forum to see how many people have problems with installation.

Adding rpm2targz to an installation script isn't painful at all.

[trasher]

neither are
$alien --to-deb fglrx*
$dpkg -i --force-overwrite fglrx*

[Lauwenmark]

Adding rpm2targz to an installation script isn't painful at all.

... but it does require preliminary installation of rpm2targz.

Originally posted by trasher
neither are
$alien --to-deb fglrx*
$dpkg -i --force-overwrite fglrx*

... unless you do worry about conflicting dependency problems.

The goal should be to give easy access to a clean installation procedure for most people. With that idea in mind, solutions like alien/forced dpkg install or rpm2targz appear mostly counterintuitive, complex and non-standard to me.

[trasher]

That's true though...
And if you do like I and force-overwrite other files (libMesa)
you'll have to reinstall it every time you upgrade the system.

But since many apps must have libmesa I can't think of another
way of solving it. Maybe if the fglrx driver comes with everything
the mesa package does you could make a dummy packages for
mesa.