A Big Comparison Of The AMD Catalyst, Mesa & Gallium3D Drive

[Kano]

Of course there a commerial games, would like to see a castor 3d benchmark *g*

[liam]

I bought an ATI 4870 in the hope of having nice open-source stack in few months, when ATI opened-up 3 years ago.
I want to play recent games like Quake Wars, Heroes of Newerth or any Unigine based game when it comes out, so it excludes the possibility of getting open-source right now.
fglrx is working fine on my gentoo, but I'm still being angry the video tearing, which is just enormous. And that video tearing is just left as garbage by ATI, I don't wonder why, as I don't think this should be so complicated to correct if the only wanted to.
Nvidia is able to deal with tearing, why not ATI ?

Anyway, for all that reasons, I think i'll go back to nvidia on my next video card, as their blobs that I have to use anyway on ATI are better as they don't tear video. I have been using nvidia blobs for years before my 4870 and I didn't noticed any particular problem, so I don't mind reverting to nvidia.

My 8400gs has always had tearing with the blob, even when drawing with ogl. Don't expect the tearing to go away with nvidia.
Btw, the player used doesn't matter for me.

[castlefox]

the problem is that there is no sustainable (financially) environment for foss drivers developers outside the HW companies. (its been stated numerous times here by developers that bounties/donations don't work.)

and i don't see this changing in the future

So... what is the proper website to make a donation?? Im feeling charitable today.

[allquixotic]

Well, what practical use do the open source drivers have if they cannot advance in an unencumbered, legally-compliant way, due to software patents? The limited GL subset that can be implemented without violating any patents is still going to leave a big gaping hole in the feature set that (Murphy's Law) all the really cool games are going to absolutely require, every single time. And there are really compelling reasons to use floating point textures, for example.

Even if we had unlimited manpower, the only thing the open drivers could do is follow the Fluendo patent licensing model, like they did with MPEG-LA: you develop a plugin infrastructure for mesa, you write 95% of the code in the open, and for those who want support for those patent-encumbered features, you go to the company that licenses the patents and buy a proprietary add-on module that comes as a sort of GL Feature Set Pack, instead of, as Fluendo does, a GStreamer codec pack. You still get encumbered proprietary code in your otherwise-open system, though, so from a free software purist perspective, you may as well just use the binary drivers anyway.

So this really calls into question the future viability of open source graphics drivers, without regard for any particular project or implementation, until such time as software patents are abolished. The subset of OpenGL that can be supported will seem smaller and smaller and smaller, as the standards continue to advance, and as popular applications start to up the ante on the GL version they require.

And then if you look back at the current reality of the situation, we don't even have unlimited manpower, or even sufficient manpower. We can't even run titles that have existed in the public for 4+ years and do not depend upon any patent-encumbered features of OpenGL. And of course, there's always an excuse for the bad performance / inability to run apps. If the app is open source, then the app developer probably hasn't really optimized their code because they don't have the expertise / manpower. If the app is closed source, then the open source driver developers can't see into the black box to figure out what it needs from the open drivers, or how to go about improving support for that app. So in either case there are excuses for the lack of performance or total inability to run / render correctly, even if we totally disregard the shadow of software patents.

Open source graphics drivers are dead to me. I give up. It's a really depressing reality to resign to, but until the larger issues can be changed at a societal level, open source graphics are a toy project for people whose interest is to explore and learn about graphics programming. Since that isn't the field that pays the bills for me -- I'm more into information security and desktop applications -- my only interest in the open source graphics drivers is (1) because they provide, or are purported to provide, a useful set of functionality to run applications I care about, and (2) they can do (1) while also being free software, which is important to me ethically and also for development / practical reasons (e.g. kernel API moving target, Catalyst KCL failing to build on recent development kernels, lack of vendor support for tainted kernels, etc.)

It looks like this is a situation where you can just never win, and as disheartening as it is, I will have to completely give up trying to support and encourage and help this effort, until the top-level issues of manpower and software patents (at least, software patents in the OpenGL spec, if not overall) can be resolved. Maybe I will refocus my efforts on those issues instead, although I've already been supporting those causes with activism and donations...

Best of luck to the folks who try and stick it out and get some kind of meaningful practical use out of the open drivers, but I don't intend to participate in Phoronix discussions about them anymore. I guess it's just "Happy Hacking" (literally for the sake of hacking itself) that's left.

[BlackStar]

If the app is closed source, then the open source driver developers can't see into the black box to figure out what it needs from the open drivers, or how to go about improving support for that app.

For the record, this is totally wrong. Consider what the driver does: it receives commands from the application and acts upon them. Driver developers can trace this command stream to learn effectively *everything* about an application without ever looking into its code.

If an application says, "BufferData-DrawElements-BufferData-DrawElements", the devs can see that the frequent BufferData calls are stalling the GPU and work to optimize VBO uploads. Same goes for frequent state changes, viewport resetting, 1-triangle draw calls and everything else schizophrenic, unoptimized applications do.

Provided enough time and manpower, of course.

[BlackStar]

My 8400gs has always had tearing with the blob, even when drawing with ogl. Don't expect the tearing to go away with nvidia.
Btw, the player used doesn't matter for me.

Same here with the Quadro, mobile version of this chip (135NVS). I did manage to get tear-free video at some point, by stopping compiz and using VDPAU but the desktop remained tear-y. Performance sucked, too.

[ssam]

the problem is that there is no sustainable (financially) environment for foss drivers developers outside the HW companies. (its been stated numerous times here by developers that bounties/donations don't work.)

and i don't see this changing in the future

How about phoronix collects donations, and use that to hire a developer.

There are plenty of people willing to chip in some money for free graphics on linux.

~13k was pledged for nouveau (but i dont think the nouveau devs accepted it)

http://www.pledgebank.com/nouveaudriver

~8k was raised for the open graphics card project.

http://www.linuxfund.org/projects/ogd1/

i don't see why someone like phoronix couldn't raise enough for a full time dev, or several student projects.

[energyman]

well, if 'desktop' oriented linux companies like Canonical would stop to create unneeded 'eye candy' like 20+ brown gtk themes or fonts nobody needs and would hire driver devs instead - that would help a lot.

Redhat, Novell, IBM etc employ devs who do amazing work to give us up to date chipset, networking, controller, etc drivers. Canonical?

[crispy]

How about phoronix collects donations, and use that to hire a developer.

There are plenty of people willing to chip in some money for free graphics on linux.

~13k was pledged for nouveau (but i dont think the nouveau devs accepted it)

http://www.pledgebank.com/nouveaudriver

~8k was raised for the open graphics card project.

http://www.linuxfund.org/projects/ogd1/

i don't see why someone like phoronix couldn't raise enough for a full time dev, or several student projects.

How much would be needed to cover a fulltime developer for say a 3 year contract? Don't forget this news will probably hit slashdot

[energyman]

How much would be needed to cover a fulltime developer for say a 3 year contract? Don't forget this news will probably hit slashdot

200 000€? If you get one for cheap?