Intel Pulls In ShadowFB Support For KMS

[phoronix]

Phoronix: Intel Pulls In ShadowFB Support For KMS

While Ubuntu 10.10 will have no i8xx driver fix for those with this vintage Intel hardware that's been plagued with stability problems and other issues since Intel introduced their Linux kernel mode-setting and GEM driver, there is now a workaround upstream for this issue. Originally the plan was to add back user-space mode-setting support to the Intel X.Org driver that would not use the Graphics Execution Manager (GEM) and this code-path could be enabled by i8xx customers to workaround the cache coherency issues while losing KMS support, but a new workaround was devised...

http://www.phoronix.com/vr.php?view=ODU3Ng

[droidhacker]

The elephant remains.... WHY DON'T THEY JUST FIX THE DRIVER?

[deanjo]

The elephant remains.... WHY DON'T THEY JUST FIX THE DRIVER?

Kinda puts in perspective how despite being a open platform doesn't necessarily mean that something will be or is more easily fixed despite having it open for all to see, analyze and commit doesn't it? And this is on a relatively "simple" graphics hardware.

[BlackStar]

Kinda puts in perspective how despite being a open platform doesn't necessarily mean that something will be or is more easily fixed despite having it open for all to see, analyze and commit doesn't it? And this is on a relatively "simple" graphics hardware.

AFAIK the issue is that most motherboard manufacturers used to customize the 8xx and 9xx BIOSes in some way, making driver development very difficult. In fact, drivers for 8xx and 915 chips suck even on windows - I used to work with such IGPs and they were trivial to crash using OpenGL/D3D. Things started getting better once the 965 was released.

In other words, it's not as if they can wave a magic wand and make the problems go away. There are probably hundreds of different combinations that would have to be tested and verified individually (think how many laptops and desktop motherboards there are with such chips). Simpler to just disable hardware acceleration and provide a more stable out-of-the box experience. Enterprising users can always re-enable it (and report bugs when it doesn't work).

[droidhacker]

Kinda puts in perspective how despite being a open platform doesn't necessarily mean that something will be or is more easily fixed despite having it open for all to see, analyze and commit doesn't it? And this is on a relatively "simple" graphics hardware.

True, but being an open platform means at least that it may POSSIBLY be more easily fixed by being seen... Don't know if you've read this thread, but it should stand as a MONUMENT to this possibility; http://www.phoronix.com/forums/showthread.php?t=24480

*** on much more "complex" graphics hardware

[deanjo]

On the same hand, the solution could have been fixed by a closed source developer as well with bug reports. It's done all the time.

[drag]

The elephant remains.... WHY DON'T THEY JUST FIX THE DRIVER?

Hrmm...

Probably because their is no point to fixing it and the advantages to users is minimal.

Modesetting is what matters, which is what is required and is what they are working on.

X's software acceleration for 2D often will have better results then what you get through any 'hardware acceleration' from 8xx. XV would be nice, but there are work arounds.

3D performance was shit when it's new, was shit when we had decent drivers, was shit regardless of what OS your using, and will always be shit from now and until the end of eternity. There is nothing Intel can do about it now: that boat had sailed over 10 years ago.

[smitty3268]

ShadowFB will probably be faster for most people on this chip anyway, the only real problem now is the lack of Xv and 3D support.

Does Intel accept patches from community members? I know they do practically all their work in-house, just like Solaris was open source but completely controlled by Sun. I'm guessing no one interested in patching Mesa still has a piece of hardware as crappy as the 8xx series...

[hechacker1]

3D performance was shit when it's new, was shit when we had decent drivers, was shit regardless of what OS your using, and will always be shit from now and until the end of eternity. There is nothing Intel can do about it now: that boat had sailed over 10 years ago.

No. Performance was decent before they moved over to DRI2 and GEM. I actually had compiz (beryl) running at a stable 60FPS with 855gm hardware. And opengl such as unreal tournament and google earth also worked just fine.

It's sad that now I'd have to roll back to an ancient Xorg release and driver just to get that same capability. Unfortunately, those versions are no longer maintained or recommended.

I guess it's time to move on and forget about hardware acceleration.

Frankly, with the latest Xorg + Intel driver I actually can run most opengl and still have a stable system. But compiz doesn't work anymore and segfaults.

When they moved over to support DRI2 they basically abandoned all the old hardware and focused on their new chipsets.

It's weird because the argument is often made that having open source means it's easier to keep old hardware around and supported.

[kraftman]

On the same hand, the solution could have been fixed by a closed source developer as well with bug reports. It's done all the time.

It could, but there are many advantages using Open Source model.