The "Bloody Mess" That Is Intel's Poulsbo Driver

AdamWill writes "Phoronix writes about the mess that is the Linux support situation for Intel's new graphics chipset, the GMA 500 — aka Poulsbo. Near the end they refer to my own post on the topic ('Okay, so after a whole day spent bashing around at this crap, I can very confidently and conclusively say, it's utterly broken'). Intel has a reputation as one of the most clued-up open source-friendly hardware companies, but if they can't sort out the mess surrounding the driver for this chipset — which is already used on the Dell Mini 12 and Sony Vaio P, and will be used on many future Intel-based systems — that reputation will take a serious hit."

Reputation?

[clang_jangle]

I thought the intel video chipset reputation was already something like "it sucks, ATI or nvidia are much better choices".

Wonder if this is one of the reasons?

[HangingChad]

Microsoft threatening Intel unless they knock off the Linux integration [boycottnovell.com]. Now, all of a sudden, Intel is having all kinds of problems with their Linux drivers.

Coincidence or anti-competitive behavior in action?

GMA 500 is based on PowerVR

[Anonymous Coward]

I thought there was an upto-date PowerVR driver in X.org already: the not-so-obvious solution may be to merge the Intel GMA500 driver & PowerVR driver to create a better GMA 500.

Of course this assumes that there is an upto date PowerVR driver and that the GMA500 is close enough to existing PowerVR chips to make it worth the effort...

Re:!gonvidia

[Anonymous Coward]

If being open source means the open source community will fix it, why aren't they contributing and fixing these drivers already? Why should it matter which drivers you get?

If it's open source, fix it.

[tjstork]

If the Linux community wants open driver development, then, it should write them. Intel made an open source driver, and now the author is condemning the code? Geez, how about fixing it! If you want something to be community owned, well that community has to step up. It's not Intel's responsibility.

Re:Wonder if this is one of the reasons?

[dattaway]

Coincidence or anti-competitive behavior in action?

I've noticed lots of Microsoft news articles recently too.

Re:Reputation?

[Pinky's Brain]

Really though, you'd think Intel would negotiate an IP license which would allow them to release good drivers. It seems IMG should be getting ready to release some sort of Linux drivers around this time though ... perhaps this will address the GMA500 situation too?

http://groups.google.com/group/beagleboard/browse_thread/thread/ec1427fdb8f9ef8d/14af5abb79383525?lnk=gst&q=POWERVR#14af5abb79383525 [google.com]

Re:Wonder if this is one of the reasons?

[Pecisk]

I think it is more about the lines of "omg, there is new sublaptop market here, quick, we need solution. Damn, our video chip uses too much power. Ok, there is some niche chip which could suit us. But there is lot of NDA and proprietary stuff. Heck, let's ride with it and see if it sticks. If not, we will abandon a driver."

It is clearly a totally different video card with different chip (which have closed parts not developed by Intel). So it ends there where usually such drivers goes - to trash can.

Re:Bit of a tangent

[Creepy Crawler]

I'd agree. I had a rather nasty return on a DV6990 HP laptop. It was trash, but that's aside the point.

I went and bought a T61, all intel down to the graphics card. Better wattage drain and complete open source drivers. Ubuntu detects everything on here, with exception to the HD APS system, which I can do without (it drains batt 2w extra).

And then, I find out that Intel releases everything about their 3d system.. And because of that, Linux devs are working on a Graphical Memory Manager, called GEM. Come to find out, it only works for Intel because they're soo open. They know they sell hardware, not their drivers.

Hopefully, AMD/ATI will follow and do the same. Wonder where that leaves nVidia...

Re:Reputation?

[Sockatume]

My understanding is that it's not for 3D gaming. The GMA500 performs poorly on that front but it supports a lot of video decode features that are AWOL in the GMA945 etc.

Re:Bit of a tangent

[makomk]

Funny story really... originally, everyone was working on something called TTM, which did much the same thing as GEM does now, and was designed to be suitable for all graphics cards. Progress was slow (mostly volunteer developers) but steady. Then the Intel driver's developers decided they wanted to go their own way, and ripped it out in favour of their own simpler solution, GEM - which only worked for Intel graphics hardware. Since Intel was the only open hardware at the time, they got their way. So the real reason that GEM only works for Intel hardware is because it was designed to support Intel graphics and Intel graphics only.

(This also set development of the equivalent functionality for other hardware back a bit, since the developers were set on TTM, and GEM was useless to them as it stood.)

Re:Wonder if this is one of the reasons?

[pestilence669]

I argue for coincidence. First of all, even Microsoft relies on Linux for their infrastructure. As much as they try, it's just not possible to do many "things" on Windows. I wish I was allowed to go into more detail. Intel does realize the importance of solid Linux support. They are a leader in this area in ways they simply don't have to be. To snub Linux could compromise their foothold on the embedded market, which is significant. It's just not possible that this is intentional. Mistakes happen. I blame the outsourcing, which was probably some middle-manager's perceived ticket to promotion and praise. Companies this big can't always put their best resources on EVERY project. Since Slashdot has made this problem known, I can't imagine it'll exist for more than 30 days. The Slashdot effect is a great motivator in shifting corporate priorities. Seriously. I hate Intel. I do. In this case, however, I don't see that they've done much wrong. They will most likely issue a massive fix and add this component to their maintenance queue. Priorities.

Intel caught me on this one too :(

[HRbnjR]

I fell in love with the Poulsbo based Panasonic CF-U1 ruggedized MID. Once I saw Intel did the graphics hardware and that they had a Linux driver, I bought the thing. Knowing Intel has been doing such a great job maintaining their desktop Linux stuff (i810 driver, etc) I just trusted them, and as you can see by this article, what a mistake that turned out to be.

Re:!gonvidia

[ericrost]

No one had noticed before because no one had tried to do that stuff.

Shock, you mean an untested, unused feature in their driver didn't get the bugs worked out of it till someone tried to use it? That's SO different to a an open source driver. nVidia has been the best thing to happen to the linux desktop since sliced bread. I can play 3d games, accelerated video, and have a slick composited desktop that doesn't freeze every time I try to switch terminals. Any time I tried to do those things with any other product I came up short.

Re:Wonder if this is one of the reasons?

[Tycho]

It seems to me, that when Intel hypes a new forthcoming CPU that it often fails to mention the poor performance of the chipset to be paired with that CPU. The CPU ends up using the newest and greatest process, but saddles the CPU with a process at least two process nodes back. Not surprisingly, when the CPU and processor are matched together, the performance is often atrocious, due to the low performance chipset. Intel also has bad habit of attempting to save silicon die area by just dropping portions from the chip. Another company making the same chip could never get away with dropping the portions that Intel does in their designs. Intel gets away with it though because it essentially has a monopoly on x86 chips and Intel sets their poor performance as the baseline for the industry. Basically, Intel no longer cares about anything else other than raking in the cash.

I look forward to seeing Microsoft and Intel ending up being broken up and no longer being a force in the computer industry. Intel and Microsoft have each acted so egregiously, that Antitrust charges are probably being worked on by government lawyers for both companies.

Re:Compositing = Easy

[ogdenk]

On your 2001 iBook it was a mobile Rage 128-based GPU and it wasn't capable of using Quartz Extreme for 3D-accelerated compositing. They went from the Rage 128-based chip to a Radeon 7000 I believe in the G3 iBooks.

It was done in software with Quartz and some 2D acceleration. Still worked f**king great though. Impressively snappy even on an old 350mhz G3 tower. Much more usable than the XRender-based compositing offered as an alternative to XComposite in KDE4.

Re:Reputation?

[Chops]

It seemed to me that i810 was fine up until Intel got involved with it. I have an unusual chipset (855GM on a desktop with no LVDS output), and new versions of Intel drivers keep totally failing to work on it in various exciting ways. Before Intel engineers started showing up on xorg bugzilla (i.e. when the module was called 'i810' instead of 'intel'), this happened once in a blue moon and I got responsive, polite fixes reasonably quickly. Now, it happens constantly, and I have to beat the engineers over the heads just to stop them closing a bug with comments which more or less translate to "we can't be bothered, sod off". When bugs do get fixed, it tends to take them a respectable fraction of a year to do it.

Interacting with Intel engineers on xorg bugzilla has sort of made me yearn for the days when GNU/Linux hardware drivers were crappy, desperate efforts slapped together with enormous difficulty without any specifications to work from.