AMD Catalyst 12.6 For Linux Disappoints

[GrrlTlak]

http://www.phoronix.com/vr.php?view=MTEzNDA

Yes. A new legacy driver release and the same old non-functionality. I'm impressed.

[RussianNeuroMancer]

I'm stuck with Windows for the same reason; a 1 year old laptop with dropped driver support

http://www.phoronix.com/vr.php?view=MTEzNDA

[GrrlTlak]

Yes. Go and sell your new computer and avoid anything which has a connection to AMD /ATI in the future. I'm stuck with Windows for the same reason; a 1 year old laptop with dropped driver support - funny enough the graphic-chip is still sold openly. But no FUNCTIONAL driver for it was available at ANY time. This company (AMD /ATI) truely is the one and only company i really do hate.

[RussianNeuroMancer]

How you install driver? Generate packages?

Aside from the annoying watermark with 'unsupported driver'

http://phoronix.com/forums/showthrea...871#post272871

It can't handle multi-monitor without restart

May you register bugreport about this?

It can't handle switching between integrated gpu in the apu and the discrete gpu, you have to manually select one

Select where? In CCC?

and reboot and the performance in pts and such is rather disappointing

Upload results to openbenchmarking.org, register bugreport with links to openbenchmarking.org.

Also, I tried a few OpenCL samples that simply don't work with the fglrx.

What samples? What exactly doesn't work? On igpu or on dgpu?

Now, on to the new driver 12.6 'stable', to my surprise it simply does not work. It generates a wrong xorg.conf from which xserver can't start

Register bugreport, attach example of broken xorg.conf and describe how you generate it.

and if I try to hack it 'by hand' from the 12.6 beta it crashes with stable 3.2 and 3.4 kernels.

Crashlog?

Also I tried the open source driver but it seems trinity crossfired with a 7xxxM was too much for it

Version of kernel and open source driver? Linux distribution? (There is no CrossFire support by the way, FOSS driver run only igpu at this moment.)

Note that without the amd blob you can't shut down the dedicated gpu

You can, put

echo "OFF" > /sys/kernel/debug/vgaswitcheroo/switch

to /etc/rc.local.

Don't forget to attach report generated by atigetsysteminfo.sh script to bugreports that you'll register.

[Agross]

Catalyst 12.6 on Trinity

I would like to add something here :

About two month ago I bought the first A10 notebook that came on the market ( where I live, anyway ), an Acer V3 with an A10-4600M APU + dedicated 7670M video card.

From the begging it was all a pain in the ass. No drivers for either linux or windows : Acer shipped it with a CD with a bad 'VGA' driver only (for windows) that didn't work.

Alas, the catalyst 12.6 beta came and it worked ok-ish on windows ( not that I use windows very much, but sine the laptop was unusable in linux I had to ) and I finally got some semblance of a driver for Linux. Aside from the annoying watermark with 'unsupported driver' I find the 'capabilities' of the Linux driver laughable. It can't handle multi-monitor without restart, It can't handle switching between integrated gpu in the apu and the discrete gpu, you have to manually select one and reboot and the performance in pts and such is rather disappointing, from a laptop with a quad core that overclocks up to 3.2 GHz ( it really does work ), 16 GB DDR3 1600 and an SSD.

I have a Phenom X6 with a gf 450 on it as my desktop and it mops the floor with ATI both in native games and in games played through wine. Also, I tried a few OpenCL samples that simply don't work with the fglrx.

Now, on to the new driver 12.6 'stable', to my surprise it simply does not work. It generates a wrong xorg.conf from which xserver can't start and if I try to hack it 'by hand' from the 12.6 beta it crashes with stable 3.2 and 3.4 kernels. I had to revert to 12.6 beta.

Also I tried the open source driver but it seems trinity crossfired with a 7xxxM was too much for it, it is basically unusable. Note that without the amd blob you can't shut down the dedicated gpu so the laptop gets too hot to handle simply by having a xfce with a terminal with vim in it ...

Clearly this is my last AMD purchase with regards to video cards. Unfortunately, since you can't buy an AMD mobile cpu paired with an nVidia gpu, I'll have to pick an Intel notebook or, if ARM64 shapes up, hopefully an ARM-based laptop.

[TobiSGD]

I feel like i should be growing a goatee.

Seriously, though, I have never, ever, heard this complaint before that final releases should include more feature changes from betas. At least not that i can remember off the top of my head.

It's one thing to say that the final product sucks. Complaining about the release process not throwing in extra untested and possibly bug-producing changes at the last minute is a different argument.

Right, beta testing is not for adding extras. No one has said that they should add extras, but why do they make beta tests if they don't intend to fix the bugs the people find before the final release? That is what beta testing is for.

[barkas]

I feel like i should be growing a goatee.

Seriously, though, I have never, ever, heard this complaint before that final releases should include more feature changes from betas. At least not that i can remember off the top of my head.

It's one thing to say that the final product sucks. Complaining about the release process not throwing in extra untested and possibly bug-producing changes at the last minute is a different argument.

The reason for a beta is to fix bugs that you did not find in closed testing. So if you don't, why have a beta at all.

[smitty3268]

May be one shouldn't quote that out of context, so that it makes sense:
Why call it beta if you don't fix the bugs? That's what a beta is for, they say.

I feel like i should be growing a goatee.

Seriously, though, I have never, ever, heard this complaint before that final releases should include more feature changes from betas. At least not that i can remember off the top of my head.

It's one thing to say that the final product sucks. Complaining about the release process not throwing in extra untested and possibly bug-producing changes at the last minute is a different argument.

[TobiSGD]

It's funny, because most people have the opposite complaint. Why call something a beta if you change everything? That's what an alpha is for, they say.

You can't please everyone.

May be one shouldn't quote that out of context, so that it makes sense:

the reason was posted when the beta was out and they did NOTHING. why call it beta first when the final is the same?

Why call it beta if you don't fix the bugs? That's what a beta is for, they say.

[smitty3268]

why call it beta first when the final is the same?

It's funny, because most people have the opposite complaint. Why call something a beta if you change everything? That's what an alpha is for, they say.

You can't please everyone.