AMD Catalyst 13.12 GPU Driver For Linux Released

[drozd]

Actually there is a big change in comparison to beta driver. Latest beta driver had (13.11 Beta V9.4):
Linux kernel 2.6 or above (up to 3.12)
and 13.12 driver has:
Linux kernel 2.6 or above (up to 3.11)

Source:

http://support.amd.com/en-us/kb-articles/Pages/Latest-LINUX-Beta-Driver.aspx

http://support.amd.com/en-us/kb-articles/Pages/AMDCatalyst13-12LINReleaseNotes.aspx

Haven't tried it yet on 3.12 kernel and I think I will stay on 13.11 Beta V9.4 driver this time.

(my first post! I just registered to post this, but I was lurking here for a while. )

[drozd]

There is a big change in system requirements:
13.11 Beta V9.4 - Linux kernel 2.6 or above (up to 3.12)
13.12 - Linux kernel 2.6 or above (up to 3.11)

Sources:

http://support.amd.com/en-us/kb-articles/Pages/Latest-LINUX-Beta-Driver.aspx

http://support.amd.com/en-us/kb-articles/Pages/AMDCatalyst13-12LINReleaseNotes.aspx

[AJSB]

Well, the "regression" in kernel version support is not that bad....after all, Valve discovered some audio problems with more recent kernels....

...and yes, before add new features, it's better that AMD solves many of the notorious bug in Catalyst, so, i like their approach.

[tomtomme]

The rate at which the bugs are solved is however too slow. Out of curiosity I went through all release notes (from AMD site or from planet3dnow) of this year and counted the resolved bugs and added features. See:

http://pastebin.com/hinyhVK1

They had 10 driver releases including beta and resolved 148 issues according to release notes. But then you have to throw out the duplicates, because they SOMETIMES seem to announce fixes multiple times: first for the multiple betas and again in the stable release.

So the REAL count is thus reduced to 116, that makes solving 1 bug or 1 feature (like kernel or x-server version) every 3 days... is that much? I am not a developer at all, but to me it seems, that there is only 1 guy left for bugfixing the linux-catalyst.
For my HD 7950 the bugs drove me (some months ago) to the radeon oss driver. That is also buggy and even less feature complete, but not so bad that it would me get back to catalyst...

[polo]

same issue on Ubuntu 12.04.3 with 3.11 kernel but on 3.8 work well

[grigi]

So the REAL count is thus reduced to 116, that makes solving 1 bug or 1 feature (like kernel or x-server version) every 3 days... is that much? I am not a developer at all, but to me it seems, that there is only 1 guy left for bugfixing the linux-catalyst.

Hmm, my take on that would be that only the bugs that get logged by the internal testing team is ones that would get recorded in the release notes. Bugs probably get fixed on a scheduling basis, meaning they
Now if a team of say 7 developers were working on this, and they spend a small amount of their time doing formal bugfixes, then it would make sense.
Often things get fixed that doesn't appear in the release notes, so I suspect that like any reasonable developer, they fix a bug when they come across it.

Working on a large project myself, you really just need one idiot that is not rooted out by the management (because they either talk too well, play the toady, or whatever), to introduce hundreds of really difficult to find bugs. It works best if this person knows they mess up, and go on the attack whenever it seems that an issue of theirs is uncovered. Management that doesn't actually code, see this guy always berating people about bad code, so they often think this person has the highest standards (when it is the opposite, really).
Ok, I should stop this before it turns into a proper tirade... :-P

[entropy]

[384194]: Tear-Free Desktop sets V-Sync to 30Hz instead of 60Hz
[376115]: Display corruption when using rotation

Cool! It took them several months, but they fixed it!

[LN17]

From where do you get, that they have only one guy who fixes the bugs? As I see Marek, Alex and other guys fixes hundreds defects a a year on the FOSS driver side. Is this is just assumption?

[AJSB]

I think people are sometimes too harsh with AMD and not so with NVIDIA...."Na'vi People" in flash anyone ?

How much time did it took for Nvidia to fix it ?

[Forage]

I got this error in make.log on Xubuntu 13.10 with these drivers:

DKMS make.log for fglrx-13.251 for kernel 3.11.0-14-generic (x86_64)
Thu Dec 19 03:06:45 EST 2013
AMD kernel module generator version 2.1
doing Makefile based build for kernel 2.6.x and higher
rm -rf *.c *.h *.o *.ko *.a .??* *.symvers
make -C /lib/modules/3.11.0-14-generic/build SUBDIRS=/var/lib/dkms/fglrx/13.251/build/2.6.x modules
make[1]: Entering directory `/usr/src/linux-headers-3.11.0-14-generic'
  CC [M]  /var/lib/dkms/fglrx/13.251/build/2.6.x/firegl_public.o
  CC [M]  /var/lib/dkms/fglrx/13.251/build/2.6.x/kcl_acpi.o
/var/lib/dkms/fglrx/13.251/build/2.6.x/kcl_acpi.c: In function ?KCL_ACPI_ParseTable?:
/var/lib/dkms/fglrx/13.251/build/2.6.x/kcl_acpi.c:999:5: warning: passing argument 1 of ?(acpi_status (*)(u32,  void *, void *))handler? makes integer from pointer without a cast [enabled by default]
     ((acpi_table_handler)handler)(hdr);
     ^
/var/lib/dkms/fglrx/13.251/build/2.6.x/kcl_acpi.c:999:5: note: expected ?u32? but argument is of type ?struct acpi_table_header *?
/var/lib/dkms/fglrx/13.251/build/2.6.x/kcl_acpi.c:999:5: error: too few arguments to function ?(acpi_status (*)(u32,  void *, void *))handler?
make[2]: *** [/var/lib/dkms/fglrx/13.251/build/2.6.x/kcl_acpi.o] Error 1
make[1]: *** [_module_/var/lib/dkms/fglrx/13.251/build/2.6.x] Error 2
make[1]: Leaving directory `/usr/src/linux-headers-3.11.0-14-generic'
make: *** [kmod_build] Error 2
build failed with return value 2

13.11 9.4 beta drivers seem to be fine though

Confirmed, got me the same error when trying to install the build packages in Ubuntu GNOME 13.10. Reverted to 13.11 9.4 beta again.