You nailed it on the head there. AMD has duplicated their Windows release schedule on Linux, and it simply doesn't work for that platform.
Originally Posted by Kano
There are way to many bugs that pop up on linux, and then take 3 months to get fixed - it would be better to just release the driver every 3 months to begin with.
And likewise, new kernel/x support is always 3 months late as well - if AMD didn't have to release a new version every single month they could focus on working ahead to get support out for new versions around when the software is actually released rather than forcing everyone to wait for it.
I've noticed that not only does it take AMD 2-3 months (or more) to fix their damned bugs, but by the time they do, they've managed to insert more bugs somehow even though I'll be god damned if I can figure out what new features it has now that it didn't before.
Originally Posted by smitty3268
I'm starting to get the rhythm of catalyst releases. 11.6 was a good driver. 11.11 was a good driver. Update every 5-6 months, after having received early adopter feedback that the driver is mostly functional.
11.12 is indeed a flaming crapfest, doubly so on newer X versions. Managed to independently run headfirst into every bug mentioned in this thread (minus gnome shell and unity ones. I loathe that interface so much I've switched to kde). fglrx works OK if you use the combination of Ubuntu 11.04, Firefox and gnome2 -- but stray from that combination and horrible things happen. Sub-1 frame per second performance with full screen flash video and kwin (known problem for years and years) caused me to find even MORE fun bugs running kde+compiz. Game hangs galore, mouse cursor warping, cursor disappearing under Wine plus numerous other papercuts all contribute to feeling 11.12 is a giant regression from 11.11.
Has fglrx improved? Compared to the completely non-funcitonal-crash-with-more-than-512-megs version of 2005, technically yes. While it's still mostly unusable it does not fatally malfunction, under a subset of real world conditions. Small comfort to the end user, however, when the end result is "does not work at all" compared to "mostly does not work if you want more than basic VGA framebuffer."
This is not a case of green envy either. I have several production and home machines, all running various Linux distributions and versions. One has ATI hardware (2x5830s), the rest have NV cards ranging from 7xxx to 4xx SKUs. I spend at least 10x as much time tweaking the one ATI box as all the NV machines put together, and with worse results. That is not hyperbole, I futz with the ATI box monthly and I literally can't remember the last time I did anything with the graphics driver on any of the NV boxes.
IMO fglrx is still an "avoid at all costs" if the end goal is a functional Linux desktop as opposed to a hobby and challenge.
I agree and would like to put it this way, too: Avoid AMD /ATI if you like your computer /desktop to be functional. I look forward to the end of life of my current hardware so i can replace it with something better than AMD.
Originally Posted by v8envy
Anybody knows if the xv video bug will be fixed in the next release? The next release ( 12.01 ? ) when will be released?
Today i tested fglrx with kernel 3.3, there was basically no problem. Then i thought i test gl2benchmark again. Well there was a problem with driver 11-12 - test 1 fails with hd 5670, but not with hd 4550. Then i thought i used that hd 5670 with 11-12 to test xbmc-xvba and had only artefacts. Now i reverted to 11-11 and it finally worked as expected. I really hope that 12-1 will work again with gl2benchmark + xvba...
I find these tails of disaster interesting as I have had no problems (except 1, see below) with FGLRX. I have NEVER seen this tearing evryone talks about the players Xine and mplayer working OK. I have three machines One has an NVidia GTX 560, the second AMD HD 5870 and the third a 4870. The NVidia and the 4870 using closed src driver, the 5870 using the current Radeon driver in debian testing. None of the three systems gives my trouble. My own prefernce is for the open source driver the performce of this being exxcellent for most things I do. I do do some 3D programming building a terrain editior this is mostly done on the NVidia. This machine has a Samsung 305T 30 inch 2560x1600 monitor on it. I have tried FGLRX on it a number of times and its is best to say its pathetic being a like a snail on horse tranqulisers, taking around a 1/4 of a second to redraw the screen........ Way to slow. Interstingly the AMD cards work OK on this monitor under Windows.
I have never sudeeded in getting the NVidia open srce driver to work (not a topic for this thread of course).
It is intreging how FGLRX seems to work OK on some systems and not on others!.
Catalyst 11.12 works fine for me too on CentOS 6.2, with a Radeon HD 3200 IGP. I can't complain about tearing and considering it's only a IGP, can't complain about 3D or 2D performance either :cool:
Sometimes I use the Radeon driver instead, but it doesn't matter so much in my case.
I don't have any problems with the Catalyst with my Radeon HD 6870 (desktop) and E-350/Radeon HD 6310 (notebook). However I aware there are known bugs with the 5xxx series in the 11.12 driver release.