Catalyst 12.6 final is out

[archibald]

If they don't have the manpower to fix reported bugs than they shouldn't release, simple as that.

Your argument falls down as it would mean that no software can be released if there are any known bugs. It sounds appealing as a way to ensure high quality software.

Companies try to strike a balance - it is better to issue a release and say "it's brilliant as long as you aren't doing X, but don't worry, we're working on it" than to say "we've fixed this problem you're having, but due to an unrelated issue for another customer, we aren't going to give you the fix - please wait patiently".

[andy_0]

@ archibald
That's nice to say but I think TobiSGD got a point. They stopped the monthly released because they want to ensure quality or at least don't have the manpower. Besides, the "known bug" is a major bug which makes the driver useless for, lets estimate, about 80% of all x64 linux distributions as a lot are based on Debian. Keep in mind that the x64 architecture is widely used. Based on the supported cards I would say it is used in more than 50% of all customers.

Let's just assume those numbers are not that impressive. Even with a major issue which affects lets say only 10 percent of their customers they should at LEAST add a "known bug" information to the release notes. There are no information whats so ever and it's still in the system. So the question is: Is that good programming style? Is that good release management? Do they have, at the very least, a good information policy?

The least one may expect is a note on the download description reading "Don't use on x64 Debian-based Linux Distributions". However there is none.

So the questions remains: What's the point in releasing a (really) faulty driver in the first place if they have no time pressure to deliver content (read: drivers to a due date). Fglrx 12-6 does not fix any critical issues which would warrant such behaviour (assuming they had deliverd a release note with the known bug).

[TobiSGD]

Your argument falls down as it would mean that no software can be released if there are any known bugs. It sounds appealing as a way to ensure high quality software.

Companies try to strike a balance - it is better to issue a release and say "it's brilliant as long as you aren't doing X, but don't worry, we're working on it" than to say "we've fixed this problem you're having, but due to an unrelated issue for another customer, we aren't going to give you the fix - please wait patiently".

You are right, I should have said: If they don't have the manpower to fix a bug that makes the driver unusable on a wide range of systems (no one can say that Debian is a rarely used distro), despite the fact that this system critical bug was reported in the beta phase, then they should delay the release until they have fixed that bug.
If it would be a minor glitch nobody would care, but this isn't a minor glitch. Was it a smart move, especially when you look at the negative press AMD has made in the last months regarding their Linux blob, to release a driver with such a heavy bug? I would say definitely not.

[Kano]

Btw. if you use debian wheezy/sid 64 bit and you have got an output with

lspci|grep '01:00.0 VGA'

you can use the packaged fglrx driver (it is in sid) and this deb:

http://kanotix.com/files/fix/fglrx-hack/libpciaccess/libpciaccess0_0.13.1-2kanotix1_amd64.deb

I really hate that this is needed. Why on earth is there no 12-7 beta or so with a proper fix. Also i need kernel 3.5 support...

[archibald]

If it would be a minor glitch nobody would care, but this isn't a minor glitch. Was it a smart move, especially when you look at the negative press AMD has made in the last months regarding their Linux blob, to release a driver with such a heavy bug? I would say definitely not.

Good point - I just hear that phrase quite often (from people who usually haven't thought it through) so I tend to jump in when I see it used.

[Shabang]

I am using ubuntu 12.04 LTS with fglrx 12.6 beta drivers and I am facing with weird problems, such that:

- OpenCL segmentation fault frequently during execution (nothing wrong in my codes, works fine in Intel OpenCL GPU/CPU)
- Ubuntu problems (with windows manager, interruptions, flash player, music player, every kind of application)
- Was surfing on internet using my discrete card 6770M then every flash page (even facebook php profile page) crashes
- Some applictaions doesn't even opens (Rhythmbox, inkscape, gimp, and a huge list...)

I would like to know if the final 12.6 release is better?

[squirrl]

crap

12.6 doesn't detect E450 / 6320 on Windows either
So like WTF?

Luckily 12.7 does detect it and installs. I'm just assuming it works in Linux.

[TobiSGD]

12.6 doesn't detect E450 / 6320 on Windows either
So like WTF?

Luckily 12.7 does detect it and installs. I'm just assuming it works in Linux.

There is no 12.7 for Linux.

[adriankx]

Well thats the usuall AMD trademark they are to late or they can`t provide. In cpu sector they play catch up for years now in driver departement same thing. Plus i think they dont listen or care about bug reports against their drivers. Nvidia i mean on windows always had weaker hardware but very good drivers thats why in most games they performed better. People should leave amd platform until they manage to provide some quality products/hardware. With amd it seems its always half baked half cooked things, drivers etc. Same song over and over "we cant get it right" refers to amd ofc. I really hope intel invests more in GPU sector with Haswell, and like that opensource drivers can lead the way.

[the-me]

Btw. if you use debian wheezy/sid 64 bit and you have got an output with

lspci|grep '01:00.0 VGA'

you can use the packaged fglrx driver (it is in sid) and this deb:

http://kanotix.com/files/fix/fglrx-hack/libpciaccess/libpciaccess0_0.13.1-2kanotix1_amd64.deb

I really hate that this is needed. Why on earth is there no 12-7 beta or so with a proper fix. Also i need kernel 3.5 support...

Hey this ancient, things have changed. Debian has now: 1:12-6+point-1
This includes the following changes:

[ Patrick Matth?i ]
* New upstream point release. AMD gave us the permission, to release this
driver based on internal version 8.982.13.
- X11 crashes on amd64 are fixed.
Closes: #675940
- Drop patch 01-fglrx_kernel_3.4.0, it has been merged by upstream.

[ Andreas Beckmann ]
* 01-fglrx_kernel_3.5.0.diff: New patch from Hans-Frieder Vogt, fix kernel
module build for kernel 3.5.