An Open-Source Intel GMA 500 Driver Appears

[phoronix]

Phoronix: An Open-Source Intel GMA 500 Driver Appears

If you're an owner of a netbook or other hardware containing an Intel Poulsbo / GMA 500, the Linux 2.6.39 kernel should be rather exciting. Entering the Linux kernel's staging tree is an initial open-source driver for this notorious Intel graphics processor derived from Imagination Technologies' PowerVR SGX graphics core...

http://www.phoronix.com/vr.php?view=OTE0NQ

[damipereira]

Excuse me if I disagree but we don't need "open-source acceleration" first we need good acceleration, with supported and easy to install drivers, if then you can have it open-source, better.
I support open-source and all, but you can't say users want open-source, users want games to work, and drivers to install, if then they know it's open-source, they will be happy about it.

[MaestroMaus]

Excuse me if I disagree but we don't need "open-source acceleration" first we need good acceleration, with supported and easy to install drivers, if then you can have it open-source, better.
I support open-source and all, but you can't say users want open-source, users want games to work, and drivers to install, if then they know it's open-source, they will be happy about it.

If that's your argument then you are agreeing...

The closed driver is horrible in multiple ways and the open ones for other Intel IGP's are good (enough). So in this case good drivers == opensource drivers.

[Dard]

First and foremost, we need stable drivers.
The proprietary ones are NOT. I could only avoid crashes by deactivating several features and it still sometimes crashes upon wakeup.

Since the proprietary drivers are no longer maintained, it seemed like I would never get a stable driver.
An open source driver can reach stability. As long as 2D is not too slow, I couldn't care less about any further acceleration or 3D.

What I don't understand yet if with this new code will there still be a dependency to the binary drivers?

[allquixotic]

The dependency to the binary drivers is essentially being stripped out. They're opening the floor for someone (probably not Alan Cox, mind you) to come in with a driver for the PowerVR SGX core to do the 3d accel. They're doing part of the kernel work for you, and providing a kernel-based precedent for others to contribute acceleration code.

It's a good thing: an actual DRM driver for Poulsbo in the mainline kernel. It's probably the first step towards a proper 3d + video stack for Poulsbo.

But I agree with Michael that the work currently being done by FSF supporters on REing the PowerVR SGX is going to influence this. I'm not sure how different the (considerably older) SGX core in the GMA 500 is from the latest chips packaged in smartphones and so on, but if they are at all similar, documentation for one SGX chip should at least streamline the information gathering processes for REing the remaining SGX chipsets. This is roughly how Nouveau works -- they work on pattern matching, and hopeful guessing that two chipsets might happen to work in similar ways with similar hardware design patterns.

Of course, it doesn't always pan out ideally as you expect it to. But once you have a pretty good knowledge about a given chip family by a given vendor, chances are that there will be a reduced amount of effort to get to the same level of knowledge about other chips in the same family, by the same vendor.

And modesetting is always a real pain in the butt, so they have a nice easy jumpstart with the Intel GMA logic driving that.

[damipereira]

If that's your argument then you are agreeing...

The closed driver is horrible in multiple ways and the open ones for other Intel IGP's are good (enough). So in this case good drivers == opensource drivers.

I agree that open-source drivers in this case are better, but I disagree with the sentence
"but still it's open-source hardware acceleration that is really needed to please end-users" because it doesn't really matter if it's free or not, open is not really needed but a nice to have. So I would remove the open from there and add stability and easiness.

[allquixotic]

I agree that open-source drivers in this case are better, but I disagree with the sentence
"but still it's open-source hardware acceleration that is really needed to please end-users" because it doesn't really matter if it's free or not, open is not really needed but a nice to have. So I would remove the open from there and add stability and easiness.

Depends on whether you value your freedom, or not. Don't assume that everyone's values are the same.

[»John«]

So I would remove the open from there and add stability and easiness.

Not only in my experience does "free and open" map almost directly to "stable and easy" for obvious reasons - the blob's history clearly shows that it's creators have no interest whatsoever in properly maintaining it, which leaves the users with broken and unsupported crap that's bound to eventually become completely unusable.
On the other hand, free and open code can be seamlessly integrated into any distro out there and properly maintained as long as someone cares enough to do the work - if that's not worth calling "stable and easy", I have absolutely no idea what is.
That's why I believe "free and open" is a "must have" rather than "nice to have" and why I really hope that all those who care get a full blown free and open code ASAP.

[deanjo]

Don't assume that everyone's values are the same.

I believe that is the point that he is actually making. People also have various definitions of what is freedom and what is not. Having freedom of a source code that someone can tinker with vs the freedom of being able to enjoy full featured solution. Most people that use computers are users, not programmers.

[damipereira]

I believe that is the point that he is actually making. People also have various definitions of what is freedom and what is not. Having freedom of a source code that someone can tinker with vs the freedom of being able to enjoy full featured solution. Most people that use computers are users, not programmers.

Exactly I disagree with the common assumption that everyone need open drivers, most of us need good drivers. Nvidia drivers, even if they fail at 2d some times, are very good.