OpenSUSE Says Farewell To RadeonHD Driver

[d4ddi0]

I used radeonhd for a long time, and never had a problem. Like opensuse, it was when all my other machines had kms drivers which meant I didn't have to fight with FB settings.(yes, I still spend a fair amount of time in the console). Other than the different naming conventions for outputs, the switch was a non-event for me.

As part of the rise of usable open source ATI/AMD drivers, I am grateful for its existence and contribution.

[quintesse]

Used radeonhd mostly for the HDMI Audio support. Using radeon for quite some time now but never have been able to get the audio to work again. (The patches have been ported from radeonhd, but I can't find any info on how to configure it for the radeon driver, possibly needs a .34 kernel)

[marek]

More than Canonical ever invested into driver development.

Yeah last time I checked they just changed a skin.

I wonder if the classic mesa driver is going to be superseded by the Galleon3D driver in a similar way

The target is Mesa 7.9 if we fix all known regressions until then.

[mcgreg]

Yes, infact you need a current .34-rc Kernel to make hdmi-audio work. Works just fine here on my rv730.

Used radeonhd mostly for the HDMI Audio support. Using radeon for quite some time now but never have been able to get the audio to work again. (The patches have been ported from radeonhd, but I can't find any info on how to configure it for the radeon driver, possibly needs a .34 kernel)

[libv]

To those who trash radeonhd and claim it should never have existed: your statements can be likened to denying that concentration camps ever existed in WWII. While the gravity of events is vastly different, both of those statements ignore massive amounts of facts and evidence, and your statements are a shameful defamation of history.

For those with some grasp of reality: it was bound to happen as there were no developers really tasked to work on radeonhd any more (while redhat and ati have developers tasked to work on the competing driver, and documentation no longer is made available as freely as before, making it difficult for others to work on this hardware).

We at SUSE wanted to do what was best for the free software desktop, and it's a real shame how politics and shortsighted egotripping wasted a lot of resources and destroyed many of the good and honest advancements and goals of this project.

Luckily, quite a few of ideas from and around radeonhd have been picked up, even though those who picked them up refuse to admit that, and radeonhd has changed free software graphics drivers for ever, beyond some basic driver development paradigms.

As one big for instance: free documentation was proposed by egbert, emmes and me, even though we never thought AMD would go with it. But it stuck, and several other companies followed suit because of that. Sadly though, in recent times, ATI has become a significantly less good citizen.

[mcgreg]

Seriously, even if I don't object to most of your arguments, the comparison you've made with:

To those who trash radeonhd and claim it should never have existed: your statements can be likened to denying that concentration camps ever existed in WWII. While the gravity of events is vastly different, both of those statements ignore massive amounts of facts and evidence, and your statements are a shameful defamation of history.

is really really more than just inadequate.


Besides that, I think most open source ati radeon gfx driver users are more than just thankful for what you did. It's not like we've forgot what you did. And just because some of "us" think now it was inevitable that this happened, doesn't mean we have no respect for you guys or for what you did.

[rohcQaH]

way to pull a godwin on this one.

way to pull a hypocrite as well, still ignoring the facts you don't like and spreading the same stupid accusations, despite very reasonable arguments presented in previous discussions on that topic.

If you're intentionally trying to look like an idiot, you're certainly doing a good job right now. Please keep wondering why you can't seem to make any friends in the community.

[libv]

way to pull a godwin on this one.

way to pull a hypocrite as well, still ignoring the facts you don't like and spreading the same stupid accusations, despite very reasonable arguments presented in previous discussions on that topic.

If you're intentionally trying to look like an idiot, you're certainly doing a good job right now. Please keep wondering why you can't seem to make any friends in the community.

Ok, let's throw in some facts then. One major accusation made against radeonhd is the fact that it was a standalone driver.

Here are some reasons as to why it was a standalone X.org driver:
* until _very_ recently, an X.org driver handled: modesetting (including cursor), 2d acceleration, xvideo, dri initialisation.
With R500, modesetting was completely different, and hw video overlay (xvideo) was barely existent on this hardware. Render support in ati only existed up to R200. R300 render and textured video support was only added in Q1 2008, more than 6 months later. So, only XAA and base EXA acceleration would need to be ported over, with a massive amount of code for hw init and modesetting being vastly different.
* -ati was in a state of flux, as randr1.2 support was recently added, introducing a lot of bugs. Stuffing more in there would only lead to more problems, a fact still seen today (http://cgit.freedesktop.org/xorg/dri...?id=6aedd57f6c). By starting a new driver, the r100-r400 driver would've been allowed to stabilise, without new bugs being introduced all the time.
* AMD did not allow SUSE to communicate with the outside world. This prohibited us to communicate with avivo developers to get their license changed, this also made it impossible for us to commit to -ati directly. With -ati moving that much still, merging in code 3 months later was pretty much a nogo.
* clean code and good structure is only really possible with a driver written from scratch.

Can you provide any technical points that are in favour of another conclusion? I cannot, not then, not today.

Heck, i believe that it was the second half of 2007 that mach64 and r128 was moved out of the -ati blob because there was little to no code shared.

I can only see non-technical reasons for creating a driver inside the (remainder of) -ati:
* branding, which kind of makes one wonder about the mach64 and r128 move.
* by building one big binary, everyone would be installing something resembling a cuckoo's egg next to radeonhd anyway.

The latter was underlined by the fact that, when code was taken from radeonhd to create the competing driver, some RHD symbols were kept, making it impossible to for instance run radeon on r100-r400 hw and radeonhd on r500+ hw at the same time. The ABIs of those RHD symbols were of course different which led to some interesting bug reports.

Feel free to try some name calling again now, and i'd be most amused if you'd do so while talking about technical excellence and stuff.

[nanonyme]

Technically that wasn't an applicance of Godwin's law either, really. This would require the debater to compare their opponent to Hitler or Nazis. Someone who claims concentration camps didn't exist isn't necessarily a Nazi, however, they definitely would be somewhat so to say historically challenged. I don't consider calling someone stupid in a roundabout way really counts as an invocation of Godwin's law.
Carry on.

[smitty3268]

libv, you seem awfully sensitive about all this.

Admittedly I haven't been involved in all the behind the scenes stuff, but from what I've seen here no one is really saying radeonhd wasn't useful. Lots of good code came from it and got copied back into the main driver, they're just saying that once -ati gained the support it needed then continuing development on radeonhd was wasteful. And it seems like the main reason that even happened in the first place was because radeonhd didn't want to use atombios and everyone else thought that was stupid.

Once KMS was being used, is there even a big difference in the code anymore?