Open-Source 2D, 3D For ATI Radeon HD 5000 Series GPUs

[etnlWings]

Good, I will wait for 3 years and THEN buy an Evergreen!

If you can't make use of this code within a couple of months, you can't be trying too hard.

[markg85]

oh wow, didn't expect this awesome news anytime soon!
gotta give this a try sometime soon.

Does anyone know how this runs with KDE 4.5?
And how is the opengl support.... 2.1? 3.? or peehaps even 4.?

[nanonyme]

"Lower than 2.1", probably assuming you mean actual working functionality.

[bridgman]

oh wow, didn't expect this awesome news anytime soon!
gotta give this a try sometime soon.

Does anyone know how this runs with KDE 4.5?
And how is the opengl support.... 2.1? 3.? or peehaps even 4.?

The GPU hangs seems to happen mostly when 2D and 3D drawing are both being used, so you probably don't want to be running a compositor yet.

The code push is basically the existing radeon and classic mesa HW drivers with changes and additions to let the code run on Evergreen hardware, so you should see basically the same functionality that you get on 6xx and 7xx today (plus some bugs).

Roughly GL 2.05, in other words

[rohcQaH]

Is powermanagement implemented yet?

I've got it running, some visual glitches (that were mentioned already), quick X crashes when the compositor starts running.


Unfortunately, my computer roars like I'm running furmark, so I'll be back to fglrx in a bit

[bridgman]

http://www.overclock.net/linux-unix/731469-how-power-saving-radeon-driver.html

Note the new instructions starting with 2.6.35. There is also a "mid" profile now IIRC.

[V!NCENT]

This has made /. and in the comment 'section' (lol) I read that Macs, for which Apple appearantly makes their own drivers for, only get to enjoy 40-50 of the performance compared to when Windows is ran on the same machine.

Now I remember Bridgman saying he still expected to get 70% the performance of fglrx with the floss drivers. So not only does Linux has or will get faster graphics than the proprietary driver stack of OS X, but also has or will get a faster floss graphics driver.

That means floss is already (or will be) murdering the insanely expensive proprietary Apple computers.

Fscking awesome!

Apple has, in the mean time, responded to this issue and is about to improve the situation. Hah! Poor Steam users!

[bridgman]

Precise and Correct is fine.

Precise and Incorrect, not so much

If you drop the precision a bit your chances of being correct go up.

[Kano]

@bridgman

Can you talk to fglrx devs that they should initialize the gfx chip completely and do NOT rely on the default register states it has directly after boot? Currently it is impossible to switch from loaded radeon (with kms) to fglrx.

It is possible to do so when you want to switch from nouveau to nvidia. All you need is to run a script inside screen that stops X, uses this hack:

for x in $(ls /sys/class/vtconsole/*/bind 2>/dev/null); do echo 0 > $x; done

and then unload the oss modules:

this works for nouveau:

rmmod nouveau ttm drm_kms_helper drm

this unloads radeon, but fglrx will crash

rmmod radeon ttm drm_kms_helper drm

But when you think of live systems it may NOT crash when you switch drivers!

Isn't it funny when reverse engeneered drivers behave better than those written from open specs when you just want to load the other driver?

I hope somebody will try that tricky code... My nvidia script uses it when called inside screen.

[bridgman]

Hi Kano;

Is there a real-world use case for switching between fglrx and radeon other than testing ? I understand the desire for fglrx-to-fglrx and radeon-to-radeon (basically "update the driver without rebooting just like Windows") but not sure how to argue for switching between open source and proprietary without reboot.

If your measurement of "betterness" is "programs the same registers the same way" then presumably a reverse-engineered driver is going to win every time. In fairness though, that's not a common way to determine a "good driver".

"It works just like fglrx so it must be the best" ?