Intel Next-Gen OSS Drivers Out!

[Michael]

Intel also announced a new site @ http://intellinuxgraphics.org/

The Intel Open Source Technology Center graphics team is pleased to announce the immediate availability of free software drivers for the Intel? 965 Express Chipset family graphics controller. These drivers include support for 2D and 3D graphics features for the newest generation Intel graphics architecture.

As with any free software project, this release represents only the beginning of an ongoing commitment by Intel to work with the X.org and Mesa communities to continuously improve and enhance the drivers. While these drivers represent significant work at both Tungsten Graphics and Intel?as our first release of this code?they're still in need of significant testing, tuning, and bug fixing before they will be ready for production use. We're releasing them now to demonstrate our ongoing commitment to providing free software drivers for Intel hardware.

Some of my comments about Intel's new drivers today can be found in a CNET article due out any minute, and right now I am working on some benchmarks of these new drivers for Phoronix.

Feel free to discuss.

[[Knuckles]]

This is great!

If it works nicely, I think I'll start to leave out 'nvidia gpu' on the list of requirements for my new laptop.

Looking forward to those benchmarks and hearing how it works with xgl/aiglx.

[Rob Williams]

Grats on the quote on C|Net Michael... that's awesome. Good to see Intel caring about Linux... even to go as far as a dedicated domain. Good stuff.

[Michael]

Grats on the quote on C|Net Michael... that's awesome. Good to see Intel caring about Linux... even to go as far as a dedicated domain. Good stuff.
Thanks, I do some consulting for CNET from time-to-time. I am also working on possibly an interview with the head of Intel's Linux Graphics department in the next coming days and I am also working on some other information to pass along as well.

[Michael]

No benchmarks in this article (ET was also laggy), but a few thoughts were shared, and AIGLX was tested... http://www.phoronix.com/vr.php?view=7148

If there are any specific items you would like more information on or tests, just post.

[Rob Williams]

Oh ok, I had no idea. Seriously cool though :D

In the last picture there, what's that breaking apart effect? Is that the video driver screwing up when you were trying to test out a game?

I am not that knowledgable on integrated video at all, but what was the option prior to these drivers? Were you able to play games without it before, and is the performance better or worse with an official driver? I look forward to reading your findings.

[Michael]

The artifacts in the first and third pictures is from the breaking effect when closing windows in AIGLX.

Intel's Linux drivers have been open-source for a while. Today's announcement is basically that it's updated, supports 965 Express, and has a dedicated website. Previously, the results were similar using the i915 and Enemy Territory.

[User_Program]

@ Michael

I'm intrested in what type of performance does the 965 Express have? In comparison to Nvidia and ATI. Is it safe to say that it will be able to play ut2004 and other games of the like?

Being that the 915G (Grantsdale) came out in 2004 have they improved on this ? Also is this just a linux related issue of the 915G and getting bad gameplay. Sorry I just found your news about this and am just learning. I never knew there was an open source alternative for GPU's

Though none the less this is great news thanks!

[R3MF]

let's hope this weighs on AMD/ATI's mind.............

[Michael]

The Intel 915G uses the GMA 900 while the 965 Express uses the GMA 3000. The 965 Express should technically be able to support Unreal Tournament 2004, but I have yet to try it myself (though I will when time permits).

[Svartalf]

No benchmarks in this article (ET was also laggy), but a few thoughts were shared, and AIGLX was tested... http://www.phoronix.com/vr.php?view=7148

If there are any specific items you would like more information on or tests, just post.

It's worth noting that the board you're using isn't the one you want to attempt doing gaming performance testing on- the i915 chipset has the GMA 900 GPU core integrated into the chipset.

The i965G chipset has the first potentially usable integrated GPU, the GMA x3000. Here's the from the 10000 foot level scoop on both GPUs...


GMA 900

- 256 bit architechture
- 1.3 Gpixel/1.3 Gtexel fill rate
- No T&L support on chip
- OpenGL 1.4 support
- Vertex Shader model 2.0
- Single unified graphics pipeline

GMA X3000

- 256 bit architechture
- ~5 Gpixel/5 Gtexel fill rate
- Hardware T&L
- OpenGL 2.0 support
- Pixel Shader model 3.0
- Vertex Shader model 3.0
- 8 unified graphics pipelines


Very big difference between the two. While still not a barn burner, it's largely on a parity with the low-to-middle end GPUs that are typically fielded in laptops and most of the UMA motherboards out there. You probably want to see if they carry through on the promise that the x3000 offers by getting a G965 based motherboard.

[Michael]

We will have access to 965 Express motherboards for testing, though we have not had the time yet to run through our array of tests. We hope to report some results soon.

[Svartalf]

We will have access to 965 Express motherboards for testing, though we have not had the time yet to run through our array of tests. We hope to report some results soon.

Cool. I can tell you that I can only hope there's something compelling in the drivers they just released- well, with the GMA X3000, that is. It would be nice to see AMD turn around and step up to the plate like Intel has and do the same sort of thing. The performance with the x300 GPU on my Athlon64 laptop has been nothing short of disappointing under Linux- I largely can't do any 3D game porting work with it (Thankfully for me, I'm on Disciples 2 right now, the prior project I was on is just being beaten on one last time before it goes officially Gold...) What's sad is that it works reasonably well under Windows- so it's not the GPU itself.

[Licaon]

so are they truly (http://marc.theaimsgroup.com/?l=linux-kernel&m=115536806403908&w=2) opensource?

[Svartalf]

so are they truly (http://marc.theaimsgroup.com/?l=linux-kernel&m=115536806403908&w=2) opensource?

Apparently not... <*SIGH*> One would have hoped for more from them- but I guess it's still time to kowtow to the media companies- never mind that they're much, much smaller than the other industries they keep screwing up just to prop up their own business model. What's so disappointing is that the electronics industry keeps letting them call the shots.

[James]

That's unfortunate, but is there any downside to not having Macrovision code in the drivers (as mentioned in the link)?

I'd rather not have Macrovision enabled.

[Svartalf]

That's unfortunate, but is there any downside to not having Macrovision code in the drivers (as mentioned in the link)?

I'd rather not have Macrovision enabled.

Without it, you also don't get any hardware accelerated MPEG2/MPEG4 playback either- the technical info has to be "tied together" per the deals all these jokers made with the media companies back a while back. And if you've got control of the Macrovision pieces, you can turn them off, which the media companies don't want. It's all about their control of your use of the content in question- they don't want any "uncontrolled" use, period.

To be sure, it's not a real loss, per se- but it would have been nice to say, have the headroom to display effortlessly as you're capturing a video feed (i.e. make a MythTV box just that little bit better...), but then we're still going to have to deal with another decade or so of obnoxious media company influence in this area.

[drag]

Without it, you also don't get any hardware accelerated MPEG2/MPEG4 playback either- the technical info has to be "tied together" per the deals all these jokers made with the media companies back a while back. And if you've got control of the Macrovision pieces, you can turn them off, which the media companies don't want. It's all about their control of your use of the content in question- they don't want any "uncontrolled" use, period.

Well that and then there is the legal software patent restrictions that into the use of mpeg* decoding functions. With that Intel's hands are tied as much as anybody's elses out there. They very simply cannot call the shots on weither or not to release open source drivers on that, (or at least they are not willing to absorb the costs of buying licenses for every single OSS user that may want to use the code.)

But personally I couldn't give a crap less. If their mpeg decoding support is that great it will be a simple matter of reverse engineering the binary for it. You couldn't do that in the U.S., but other countries it would be fine legally. Since it's userspace it should be pretty simple.

They did that with OpenBSD driver's for Intel's newer wifi stuff. Intel released code for the wifi kernel portion, but stuck the binary portion in user space. So it was fine according to the Linux kernel developers. The OBSD developer setup Linux and wrote a program to intercept calls to that binary. that way he was able to figure out to write OBSD drivers without having to actually violate the license by examining the intel binary directly.

But if it's like the VIA onboard 'mpeg2 accelerator' or whatever.. then the Free software stuff software-only decoders used less cpu time then their 'accelerated' drivers.