Isn't this EXACTLY the way Intel behaved a few years ago, causing the Windows Vista debacle? Have you read some of the email transcripts that came out in the court battle over that? It is pretty obvious from those that Intel believes they are big enough that they don't have to care in the slightest about their "good name", or whether something actually works properly or not. So, number one, this should come as no surprise, and number two, comparatively speaking this is absolutely "small potatoes" as far as Intel is concerned.

jw

There are vendors on the market who make netbooks with not only intel chipsets: Samsung, HP, MSI...

I love the look of my Mini 12. But my next laptop will be without intel inside.

This situation should get better soon.

I've seen a couple of graphic demos of Moblin on Menlow platforms and in both cases they were test platforms for developing software for Moorestown. I've also had it confirmed that Moorestown uses a PowerVR core/GMA 500 so when Moblin 2.1 for handhelds preview/alpha comes out (Intel tell me before Christmas) then we should at least see a new driver/kernel combination and as it moves forward, Moblin has to continue to support GMA 500.

Chippy.

The root of the problem is the Linux Kernel itself. It doesn't have a stable API (binary compatible for kernel modules). It should have a way to say 2.4.x is stable API, for which a module compiled on any 2.4.x will always work with any other 2.4.x.

They can claim 2.5.x is a development/unstable API that it is subject to change and make up their mind before they get to the "stable 2.6.x API". Once they have a stable 2.6.x API, they can do all their experiment, mucking around in 2.7.x.

The way they do it right now is such a waste of time and resources that it is moronic. Every little 2.6.x.y release breaks all the existing kernel modules and require new compilation from source all over again, and that's every 4-6 weeks or so.

Worse yet, they are too busy playing the freakin' tweakin' game that they don't have any decent documentation for every new or changed "features" in the kernel. Many device driver projects who have less than full-time devotion to writing device drivers will find that all the structures and API's in the kernel have been replaced completely, or broken up into many different new structures with different context that need different initialization and nuiances of timing or mutex locks required, or disappeared in the new kernel revision. It is worse than a shot in the dark even if you find the time at all to read the Linux Kernel Mailing list.

Look at Windows market, their API changes every new major release, but remain stable with each major relase. Even with that many vendors can't even write decent or stable device drivers until months after official releases. They don't even have resources to support their devices on a newer major Windows release, or it's the "if you want support for such and such, you have to pay more for it". They would rather tweak the hardware or firmware slightly so they can sell a new device to enhance their revenue stream, milking the consumers instead of providing decent support and services for the existing customers... Uh... the suckers who already paid.

The whole thing $uck$ right now.

Yes Intel, it's all too late now. I bought my netbook on the assumption that you provide working Linux drivers. If they don't arrive soon, I shall return the netbook!

well I'm a linux user from Nicaragua , I have a Dell mini
inspiron 1010 with GMA 500 , I've read all about this GPU but , all that I saw was not good news for mi ubuntu and, I really hope that intel as a company takes responsability for this situation and find the way to give us a chance , and develop a decent driver or support because I felt like betrayal from intel

thanks
PD: sorry for my english

Asus EEE is another brand that promises Linux compatibility. A warning, however, for linux-fans that consider buying the beautiful 1101HA: it also contains the GMA500.

A new Intel inside logo that that mentions the GMA500. We might get a better idea what we are buying before we buy. If they feel justified in what they did, they ought to be eager to brag about it. Weren't they telling us that they would have support for Linux. If they did, I don't see that Intel is even trying address their credibility. If not they certainly seem to been rather quiet about the expectations they worked so hard to achieve. I fell for the A075h. I can't even tell how crippled things are. Even with Adam Williamson's help what is the best that can be acheived? I don't even care about 3D. Video is not that important to me. But scrolling and moving windows around shouldn't regress by 15 years. How can Intel even find vendors that can be so Linux blind. There can't be that many that have been around for 15 years. Come on Intel don't let this become the new standard by which you are measured, I know that Intel can do better.

If you look around that is how most vendors are acting nowadays.

They say that they support Linux, but in the end, it only a way to get a foot in the community or to get publicity.

None of the so called 'Linux friendly' companies is really supporting it.

- Intel supports some chipsets and some of their development tools;
- Asus sold their soul after the EEE PC sucess;
- NVidia releases binary only drivers;
- ATI has improved, but still releases partial documentation;
- Some of Google software is WINE based or still to come status;
- Nokia is using Maemo, but does not support PC Suite or similar in Linux;
- Lots of companies are using Linux internally, because it is cheap, but only provide Windows/MacOS X clients

I could still make a more exhaustive list, but I guess most of us here will get my point.

I see your problem. You are an Intel fan. That is like saying you are a Microsoft fan merely because most computers come with their products.

Please get over the whole Intel thing. Their chipsets are average at best. But unfortunately are all over the place (sound familiar?).

Other options exist. ARM on smaller devices. AMD/ATI and so forth on larger ones. Don't whine about one bad chipset of the many available. Make a better choice, to not use Intel whenever possible.

A few years ago, Intel made a "promise" (reminds you of Microsoft doesn't it?) that they would not release any new hardware until they developed a working reference driver for Linux.

Intel is too big of a company to play the victim here. No one forced them to use this hardware, or forced them to accept proprietary drivers.

Intel has the problem of cranial-rectal inversion. Even their free drivers have been underfunded. I heard it was 3 years before TV-out worked on the GMA 9xx series on Linux. The drivers that shipped with Ubuntu 9.04 leaked memory like a sieve.

Thanks for writing this article. Hopefully it wakes them up. I have heard they love Linux and aren't not fans of Microsoft. Well, you'd never know it. Personnel is policy.

...somewhat against my better judgement at this point.

But what the heck! We can debate civilly, right?

> They put their name on a GPU that was developed by Imagination Technologies, using drivers written by Tungsten Graphics.

Well, it might be a bit more complex than that, Shawn. As licensed from ImgTec, the "GPU" may well have amounted to little more than a bunch of IP (intellectual property). True, it might have been an off-the-shelf core block, but given the need to fit it to an x86-type interconnect, rather than the usual ARM bus, I'm thinking Intel may have done an IP-level license rather than a core block one. If so, the done thing is usually for the implementer to rebrand it. For example, ARM rebrands its own cores based on the ARMv5 instruction set as "ARM9." TI has OMAP. Intel did XScale. You get the picture.

I'll grant you one thing, though -- the decision to go with something (GMA500) so similar to existing and open GMA5xx GPUs was an unfortunate one for consumers. For embedded customers, I guess it was down to branding consistency. But heck, I don't even think they should have called Menlow "Atom," but I guess it was too irresistible to try to leverage that strong a brand in the device market. And, it *is* the same processor core.

By the way, you do realize that Menlow actually shipped before Diamondville, right? So Intel actually didn't know when it branded the GMA500 that netbooks were even going to be that big a deal.

> 2) Rather than provide ABSURDLY limited Linux drivers, Intel should have just not supported Linux at all.

Not an option. In the device market, folks have to be able to evaluate with Linux. EVERYTHING supports Linux nowadays.

As far as the quality of the Poulsbo drivers, I'll pitch in my "sample size of one" impressions. It was a hoot hooking up a little FitPC2 to my Mom's HDTV and seeing Ubuntu on the big screen! Can't do that with your Diamondville netbook.

And I've had no stability issues in normal all day, every day use at the Internet cafe, running Ubuntu LTS. For the intended use of a netbook, browsing and email, it's fine and good in my experience. Running Moblin? Not so much, because there's no open source implementation of OpenGL ES yet. Once that happens, us Poulsbo device owners may get to enjoy more of the hardware's capabilities. :-)

> I feel most sorry for those folks that bought the Dell Mini 12 with Ubuntu pre-installed. Certainly if Ubuntu was pre-installed the hardware would be Linux compatible. Um, nope. Yes, perhaps Dell should have read their contract with Intel better, but come on, Intel sold their stuff with no upgrade path? I'm sorry, but that's just dumb.

I can see your point. Yet, I'm pretty sure there's a Poulsbo driver on the way for Ubuntu 9.10. Or, maybe the next LTS release in six months. At least, CompuLabs told me they'd be skipping an Ubuntu release, but that there would be a new driver coming.

So, is it really *no* upgrade path -- or just an upgrade path likely to frustrate the Linux ubergeek, who actually *looks forward to* an OS upgrade more often than every couple of years? You realize your somewhat in the minority on that one, right? At least those poor Dell Mini 12 owners out there are probably getting great battery life!

I don't know. Please remember, I'm not at all *for* closed drivers. I'm just a pragmatist about it. Had my ear bent by too many device and chip execs about the value of closing things to expect instant change anytime soon, I guess. But, it'll happen. The device world will ultimately embrace the openness of the PC and Linux, too. Today, they laugh at us. But... you know the rest.

Keep up the great work at Linux Journal. I've subscribed forever, and it's always a treat.

-Henry

The Dell mini 12 is a beautiful machine. It had so much potential to be great; it should have been the $400-$500 MacBook Air for the Average Joe. But, the component choices were terrible: the 1GB memory cap, the limited hard-drive choices (80GB 4200rpm max), and the extremely bad choice of video chipset.

Were Dell to have pulled its head out of its ass in time, the mini 12 would have been a mini 9 with a 12-inch display. Bigger screen, better keyboard (correct right-shift key), AND usable internals. Not only would it have been a perfect Ubuntu box, it also would have been a fantastic hackintosh.

Every Dell mini to date has been flawed by compromise. The 9 was too small and the keyboard was terrible, the 10 and 11z have unusable trackpad buttons, and the 12 was perfect on the outside and anything but perfect on the inside.

One of these vendors has to be capable of making a machine similar to the Dell mini 12 (thin, light, head-turning good looks), but with upgradeable components and good Linux compatibility...

I've got a Fujitsu U820 laptop. It's also got the GMA500 chip of fail. And I don't care if the drivers are proprietary or not. Even with the proprietary drivers on an older version of Ubuntu, the 2d performance is still horrible, AND there is NO 3d at all. It's kind of sad that I can't watch a low resolution video on it at full speed. What annoys me is that Intel didn't even bother trying to support it on Linux. They paid someone to write a basic crippled driver that's only barely better than using the framebuffer directly, and then just dropped it. Come on Intel, how about a little support eh?

Both sides make points that are valid. Both make some points that approach lame-ness. If I were Shawn I'd be just as upset as he is, and I would sell the offending hardware and replace it.

Over the course of my lifetime I've been burned by just about every company I was loyal to. International Harvester, Atari, Citibank, and General Motors all come to mind. The lesson I've learned is that things change, and one always does well to complete ones homework before making decisions.

Shawn's indignation is justified. Intel's position is reasonable. Life will go on. We all get a little smarter.

All my best - Chris Reich; Rochester, New York
twittername: chrisreich

Go Shawn! Tell it like it is!

http://www.bashscripts.info/
Learn to Take Control of Your Computer, NOW!!!

Vendors can do better. Good communication might cost a little more on paper, but it saves in the long run.