DRM Changes For Linux 2.6.32 Kernel

[phoronix]

Phoronix: DRM Changes For Linux 2.6.32 Kernel

David Airlie has just called for the first DRM pull request for the Linux 2.6.32 kernel. To no surprise, R600/700 3D support can be found within the Direct Rendering Manager code as with kernel mode-setting support for these newer ATI Radeon graphics cards. Also found within the Linux 2.6.32 DRM is the much-worked VGA Arbitration code, the KMS page-flipping ioctl, Radeon TV-Out, and the Intel DRM code now supports dynamic clocking for improved power management. The list of highlights for the Linux 2.6.32 DRM so far can be found on the dri-devel mailing list...

http://www.phoronix.com/vr.php?view=NzU0Nw

[avilella]

What is the dynamic clocking improvement in the Intel DRM code? What sort of benefit should the average laptop user get from it?

Cheers,

Albert.

Phoronix: DRM Changes For Linux 2.6.32 Kernel

David Airlie has just called for the first DRM pull request for the Linux 2.6.32 kernel. To no surprise, R600/700 3D support can be found within the Direct Rendering Manager code as with kernel mode-setting support for these newer ATI Radeon graphics cards. Also found within the Linux 2.6.32 DRM is the much-worked VGA Arbitration code, the KMS page-flipping ioctl, Radeon TV-Out, and the Intel DRM code now supports dynamic clocking for improved power management. The list of highlights for the Linux 2.6.32 DRM so far can be found on the dri-devel mailing list...

http://www.phoronix.com/vr.php?view=NzU0Nw

[rvdboom]

I'm not sure I read David's mail correctly :
I see 3D acceleration for R600 in non KMS-mode
And I see R600 KMS + acceleration.
I assume the above "acceleration" includes also 3D with KMS.
Am I right?

[Tommy666]

Ok , but how about VIA DRM code ? Will be added into 2.6.32 ?

Hope to see support for chrome9 asap, because sometimes I wanna put my laptop through the window...

[7oby]

What is the dynamic clocking improvement in the Intel DRM code?
Downclock GPU render clock when GPU is idle:
http://lists.freedesktop.org/archives/intel-gfx/2009-August/003879.html

Similar to intel Speedstep and AMD Cool'n Quiet except that GPU Voltage isn't lowered. Just Freq is reduced and Power = C * Freq * VID^2.

What sort of benefit should the average laptop user get from it?
Longer power on time whilst on battery.

[bugmenot]

Isn't DRM = bad ??

[whizse]

DRM as in Direct Rendering Manager, not DRM as in Digital Restrictions Management, ;)

[Fixxer_Linux]

Does this 3D support for R600/R700 means that I'll be able to play UT2K4 or Heroes of Newerth with my ATI HD4870 with this open-source driver, getting ride of fglrx ?

[m4rgin4l]

Phoronix: DRM Changes For Linux 2.6.32 Kernel

David Airlie has just called for the first DRM pull request for the Linux 2.6.32 kernel. To no surprise, R600/700 3D support can be found within the Direct Rendering Manager code as with kernel mode-setting support for these newer ATI Radeon graphics cards. Also found within the Linux 2.6.32 DRM is the much-worked VGA Arbitration code, the KMS page-flipping ioctl, Radeon TV-Out, and the Intel DRM code now supports dynamic clocking for improved power management. The list of highlights for the Linux 2.6.32 DRM so far can be found on the dri-devel mailing list...

http://www.phoronix.com/vr.php?view=NzU0Nw

Hmmm...I hope it doesn't take a few months for the Catalyst Drivers to be compatible with it.

I wonder if Ubuntu is going to enable Radeon KMS by default in the upcoming releases.

[lbcoder]

Hmmm...I hope it doesn't take a few months for the Catalyst Drivers to be compatible with it.Catalyst will NEVER be compatible with it. You are going to have to entirely disable it in order for catalyst to work. That means ensuring that the applicable modules aren't loaded at bootup. No matter anyhow since catalyst is almost obsolete now, just a few more bug fixes for the open source driver and you'll never need catalyst again.

[lbcoder]

Does this 3D support for R600/R700 means that I'll be able to play UT2K4 or Heroes of Newerth with my ATI HD4870 with this open-source driver, getting ride of fglrx ?Thats the idea ;)
Though it is something of a work-in-progress, some bugs remain yet, so soon...

[nanonyme]

DRM as in Direct Rendering Manager, not DRM as in Digital Restrictions Management, ;)Even though I aren't pro-DRM, I think it still makes sense to use Rights instead of just twisting it into what you think it means. There is no such word as Restriction in the term.

[agd5f]

r6xx/r7xx 3d support works with both kms and non-kms.

[remm]

Yes, 3D is a bit glitchy, and definitely not performance oriented at the moment, but the basics work on my HD4850. With KMS.

[whizse]

Does this 3D support for R600/R700 means that I'll be able to play UT2K4 or Heroes of Newerth with my ATI HD4870 with this open-source driver, getting ride of fglrx ?

At least not HoN, there's still problems with Mesa for this particular game.

[m4rgin4l]

Catalyst will NEVER be compatible with it. You are going to have to entirely disable it in order for catalyst to work. That means ensuring that the applicable modules aren't loaded at bootup. No matter anyhow since catalyst is almost obsolete now, just a few more bug fixes for the open source driver and you'll never need catalyst again.

I meant that if it is going to be compatible with the kernel, I already knew that I would have to disable KMS. I guess it all depends on what Ubuntu does for 10.04.

Anyway, I see your point, but I think it'll take a few more months (maybe a year?) for the open source driver to match Catalyst feature set.

[Qaridarium]

Does this 3D support for R600/R700 means that I'll be able to play UT2K4 or Heroes of Newerth with my ATI HD4870 with this open-source driver, getting ride of fglrx ?

Yes! that will be the future! but!:::BUT!!! you need to wait a half year becouse the opensource driver do still not have mainfeatures like OpenGL2.1 or OpenGL3.2

in the future the opensource driver will be full-featured!

soooo only wait and see how fast the opensource driver grow!

[Qaridarium]

Catalyst will NEVER be compatible with it. You are going to have to entirely disable it in order for catalyst to work. That means ensuring that the applicable modules aren't loaded at bootup. No matter anyhow since catalyst is almost obsolete now, just a few more bug fixes for the open source driver and you'll never need catalyst again.

bugfixes??? LOL... the opensource driver do not have OpenGL2.1 or OpenGL3.2...

YES FIX THAT BUG!

[Remco]

Even though I aren't pro-DRM, I think it still makes sense to use Rights instead of just twisting it into what you think it means. There is no such word as Restriction in the term.
That's just what it is. DRM is used to manage restrictions to content. It is never used to give you more than the 100% control you would have had without DRM.

[nanonyme]

That's just what it is. DRM is used to manage restrictions to content. It is never used to give you more than the 100% control you would have had without DRM.Actually it is used to enforce that people who own full rights to digital content can get their demands on how you should be allowed to use their digital content after purchasing a copy respected. Yes, it's about restrictions, yes, it's not fully legal in all countries but it is about protecting copyrights for digital content, therefore Digital Rights Management is justified. (actually DRP - Digital Rights Protection - would be more precise, DRM has little to do with management)

[Remco]

Actually it is used to enforce that people who own full rights to digital content can get their demands on how you should be allowed to use their digital content after purchasing a copy respected. Yes, it's about restrictions, yes, it's not fully legal in all countries but it is about protecting copyrights for digital content, therefore Digital Rights Management is justified. (actually DRP - Digital Rights Protection - would be more precise, DRM has little to do with management)
I'd argue that Digital Restrictions Management (or Enforcement) is also justified, since it disrespects the customer by limiting what he can do with his property (which goes far beyond what Copyright Law would limit).

[nanonyme]

I'd argue that Digital Restrictions Management (or Enforcement) is also justified, since it disrespects the customer by limiting what he can do with his property (which goes far beyond what Copyright Law would limit).Actually technically I think their claim is that it doesn't become your property when you buy it. You just obtain right to use it, with limitations. Whether this is actually legal is another issue.

[Remco]

Actually technically I think their claim is that it doesn't become your property when you buy it. You just obtain right to use it, with limitations. Whether this is actually legal is another issue.
Without signing a contract, by buying it, you own the thing. "Owning" something in the normal sense means that you have the right to do anything with it except distribute copies of it (this is also true of TVs and clothes). There is no way that a distributor can only give away the right to 'use' a product in a particular way without you signing a contract.

But DRM always enforces extra restrictions, which are not mandated by Copyright Law. So in reality, DRM has nothing to do with the distributor's rights. It has to do with the distributor's power.

[nanonyme]

Without signing a contract, by buying it, you own the thing. "Owning" something in the normal sense means that you have the right to do anything with it except distribute copies of it (this is also true of TVs and clothes). There is no way that a distributor can only give away the right to 'use' a product in a particular way without you signing a contract.Well, that's how they see it. I didn't say it's in any means according to any laws. ;) (nor did I actually say it's against either; in either case contact your lawyer and sue them if you're unhappy)

[rbmorse]

Without signing a contract, by buying it, you own the thing. "Owning" something in the normal sense means that you have the right to do anything with it except distribute copies of it (this is also true of TVs and clothes). There is no way that a distributor can only give away the right to 'use' a product in a particular way without you signing a contract.

But DRM always enforces extra restrictions, which are not mandated by Copyright Law. So in reality, DRM has nothing to do with the distributor's rights. It has to do with the distributor's power.

Depends largely on where you live.

[Lory]

I've tried out this new code, on my HD4650. The X starts, openGL working faster than with fglrx. However I've a big problem:
I have a second card. A Hauppauge PVR350. I have a dual-head configuration (radeon-monitor/hauppauge-TV). Whenever I start mythfrontend the X restart. Has anyone any idea why?

Plus I have random graphical glitches if the compositing is on.

[Aostrich]

Pulled (http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=44040f107e64d689ccd3211ac62c6bc 44f3f0775) into the linux main tree!

[whizse]

The Intel GPU reset support also went in. That's pretty cool, all the recent stability work Intel's done is really paying off.

[moomoo]

unfortunately even with the latest changes I'm still getting some random lockups with KMS on my 3470 mobility few seconds after starting up X :( Has anyone else been experiencing something similar? I found that using a non-preemptive kernel and building drm as a module instead of having it builtin helps a bit, but sometimes (say one time every five) it would still hang.. And when the laptop crashes, nothing ever gets written in the system logs, so I have no idea on how to get some debugging info to file a bug report with ! Any ideas on how to catch this kind of wrong behaviour?

[whizse]

At least not HoN, there's still problems with Mesa for this particular game.

Updates to HoN have fixed the problems Mesa had with the shaders.