LAST WEEK Intel dropped what seemed to be a bombshell by saying that its upcoming Clover Trail Atom chip will not support Linux. Since then a little more information has surfaced that goes some way to justify the firm's decision to ignore what is arguably the most important operating system on mobile devices.
Since Intel's Clover Trail admission something odd happened. Intel sent over a statement saying, "There is no fundamental barrier to supporting Linux on Clover Trail since it utilizes Intel architecture cores and those same cores (Penwell) on Medfield support Android today (for tablets and phones)," something that we had already reported in the original article.
It is important to understand Intel's point in the statement, since Clover Trail is based on Intel Architecture (IA), more commonly known as x86, therefore the Linux kernel should run without much trouble. Intel cleverly avoided the underlying meaning of providing Linux support in Clover Trail, support of the chip's additions to the basic IA that has been around for decades, the features Intel hopes will make Clover Trail a winner in the tablet market.
Intel knows very well that simply being able to run the Linux kernel on IA is not enough. The chip firm has ploughed a lot of code into the Linux kernel over the years to provide fine-grained support for everything from the Xeon's VT extensions to its lauded network interface cards. The firm knows that if it wants to win business with tablet OEMs it has to do a lot of the legwork in making sure that the operating system, be it Windows or a Linux based distribution, can fully utilise every feature of the chip, otherwise the OEM is wasting money on a processor where it isn't getting any value out of part of it.
Intel's statement hinted at the real reason why it isn't supporting Linux on Clover Trail. The firm said, "We are focusing our current efforts for this Clover Trail product on Windows 8 because that's where we think the biggest market opportunity is for us in tablets right now. If something changes and we see customer interest or further market opportunity on a Linux-based OS for tablets we have the ability to bring products to market in future."
I read with interest the articles by some journalists writing they didn't believe the news that Intel wouldn't support Linux on Clover Trail, which is a fair reaction given how important Linux has become, with some believing Intel's marketing statement that the firm will produce a chip that supports Linux in the future. Sadly Intel's statement was what could kindly be called smoke and mirrors.
To understand Intel's decision not to support Linux on Clover Trail, one has to look at the market Clover Trail is being aimed at.
Intel is pitching Clover Trail to the low-end ultrabook market and in particular the tablet market. An Intel source told me "the tablet market is actually the Ipad market" and said that Android tablets "are not gaining any traction". These comments are not new and are crystallised in analyst reports. The truth is that Apple's Ipad is the device that consumers are buying, regardless of innovative designs such as Asus' Transformer range or cut-price units such as the Nexus 7 and Kindle Fire.
According to the source, Intel could support Linux on its tablet chip, where Linux effectively means Android. However I was told, "Intel is fed up ploughing money into optimising Android and not getting anything back." When asked whether Intel had received bags of cash not to support Linux, the source said no, and that Windows support was only there because Windows 8 tablets are the only devices that have some chance of claiming some - though the source admitted not a majority of - market share against Apple's Ipad.
Now you would expect Intel to say it didn't receive cash from Microsoft to drop Linux support but the source also confidently said that the firm could build a prototype Windows Phone device, but just like Android tablets "it won't make us any money". Intel's former PC chief, Mooly Eden told us that the firm made a business decision not to go ahead with Windows Phone 8 at this time, and given the unpopularity of Windows Phone devices, that certainly jibes with what our source is saying.
So if Intel won't do the leg work to push code into the Linux kernel, will it open up the specifications to Clover Trail so Linux kernel developers can do it for themselves? I've been asking Intel that question for the best part of a week now, a question that requires essentially a "yes" or "no" answer, but have yet to receive an official reply. However the same source told me that it's "unlikely" that Intel will open up Clover Trail specifications, because it sees no need to do so.
Intel will continue to support Linux on Medfield and the future iterations of that chip for the simple reason that Android has a lot of market share in the smartphone market, according to our source. Intel is already working with Motorola Mobility, or effectively Google, and has put considerable effort into the Android x86 development environment, even with rival chip vendor Qualcomm saying that it is trivial to build Android executables on the x86 architecture.
Those journalists won over by the response of Intel's marketing machine reported there is an upcoming chip that will support Linux and they will more than likely find that it ends up in a smartphone rather than a tablet. And being fair to Intel, it doesn't have a duty to support all operating systems, after all no one is complaining that Medfield doesn't support Windows Phone 8.
Intel's decision not to support Linux on Clover Trail is not a master plan concocted with Microsoft to undermine Linux. After all, AMD has made exactly the same decision with Hondo. Rather, it's a recognition of Android's inability to get a serious foothold in the tablet market despite having innovative designs. At the end of the day, chip vendors' decisions to support Linux - whether in the server, desktop or mobile market - are based on the hope of making money.
While Windows 8 tablets might not prove to be popular, the truth is, as our source put it, "it has the best chance of making [Intel] money". µ