Intel could Atomise handsets in two years
After 'handset-sized' devices
Intel's increased focus on Linux and real time operating systems will see it pushing its Atom architecture into handsets within three years and launching "handset-sized" devices much sooner, the chip giant's sales boss said today.
But while the vendor is dabbling heavily in the operating system market as part of its foray beyond traditional PC-based devices, it seems it would still like long-standing partner Microsoft to slide down the form factor ladder with it.
Intel paid $884m for Wind River yesterday, giving it two embedded operating systems and a development environment to play with. It also this week beefed up its support for the Moblin Linux distro for mobile devices.
Sean Maloney, executive vice president and chief sales and marketing officer, said the purchase of Wind River was part of its effort to drive a low-power full-fat PC architecture, in the form of Atom, into non-PC environments. He highlighted the health, automotive, and entertainment markets - which the vendor has been doing ever since it unleashed Atom last March.
He also said that as well as small comms-enabled devices - MIDs and smartphones - the vendor was targeting the full-one handset market, something that would be easier if it can offer potential partners a whole software, hardware, and development package stack.
"We're not into handsets yet, but we're just moving into that, that'll be the next few years," Maloney said.
We can expect "handset-sized" devices much sooner, he added. "We can just continue to nibble away at that."
But while Intel is hoovering up RTOS expertise that it hopes will help it offer a credible embedded and/or handset platform, Intel doesn't seem to want to keep that market to itself at the expense of long-time partner Microsoft.
"We don't have a version of Windows CE running down there," Maloney pointed out.
Asked if Intel would like to see Microsoft finally make the jump to handsets, he said "it's a question for Microsoft, not for us...I'm sure over time Windows will move down."
Intel, and Maloney, have of course targeted these markets in the past, with its Xscale ARM-based architecture. This effort came to naught, and the Xscale business was offloaded to Marvell. But Maloney said it would be a (forgivable) mistake to think it had ever dumped its ambitions to play in embedded and mobile, pointing out that even as it was winding down its Xscale effort, it was kicking off the Atom project. ®
Who will follow Intel?
Intel has a history of diversification then refocusing and anyone would be a fool to follow Intel down any path other than standard x86 territory like servers, PCs and laptops.
Intel have embraced and discarded:
* bubble memory
* 8-bit micros (8051 etc)
* 16-bit micros (80251 etc)
* Various embedded chips 8255, interrupt controllers,...
* USB device chips
* Commercial products such as USB microscopes
They have bought and shutdown various software outfits too, including Dialogic.
Anything outside their core x86 business has been shafted.
Now who in their right mind thinks it is wise to follow them down some new path that isn't part of their core business?
Woah, Intel bought Wind River?
That's the real story here. They have the Real-Time-OS market pretty much sewn up - almost every small, high-availability embedded system runs on VxWorks.
- We aren't talking phones here. This is serious industrial kit.
Microsoft don't have any RTOS products at all - and Linux isn't one either.
Can it be that after decades of failure to kill x86, Intel management have become blind to its problems?
Intel's engineers tried for years to shake off the ghastly x86 architecture (producing iAPX432, i960, Xscale, iTanium), while customes avoided x86 whenever they had a choice, in comms hardware, workstations, servers, games consoles and of course Macs. But the Microsoft monopoly, and Intel's superior fab processes repeatedly saved x86 from the knacker.
Multicore ARM with dedicated media processing add-on has advantages that surely trump Intel's Atom and successors. The whittling away of the Windows monopoly seems to leave as Intel's only advantage: more and better fabs than anyone else. x86 may go nowhere in handhelds, and only serve to accelerate the shriveling of the Wintel CPU cash cow, starting with netbooks stealing from the mainstream Windows CPU business.
When Microsoft switched Xbox from Xbox to PowerPC, they didn't think many moves ahead. IBM no longer could be bothered with Apple's PowerPC volumes for Mac, so one might expect up-and-coming Apple's Mac to fade from the scene. But with IBM not delivering PowerPC, and Intel out of the console market, it was a no brainer for Apple to step in and become Intel's premium customer (OS X had been running on x86 long before PowerPC). Instead of marginalising Apple, MS gave Apple parity, able to build as many Macs as they want from the same parts supply that feeds MS's OEM partners, as cheaply as them. No supply problems, no cost problems, no performance problems any more. What a blunder by MS.
Or have I got it all wrong?
were is the mass produced low power AVC Encoder SOC
iv said it before and ill say it again, were is the mass produced generic low power Hardware AVC Encoder SOC capable of H@L4.1 D1 at least for these Realtime OS's and consumer devices of all types to start using ?
the AVC/H.264 H@L4.1 Encoder tech is available off the shelf for a long time now, as has 11n/Wimax longer range chipsets, but noones bothered to put them together in a single or dual SOC package to take advantage of the global mass market demand for static and portable SD AVC reltime encoding at 24/25fps....
this grab for the second best realtime OS (QNX 6 being the far better optopn OC) seems like a perfect time to take the available AVC H@L4.1 and current Wimax chipsets and combine all 3 parts into something the worlds end users could really make use of TODAY at a generic level, Intel \and others please do so ASAP....
Atom only makes sense in a Windows environment
Current and future Atoms draw far too much power to be competitive with ARM based embedded systems. The only reason a designer would use an Atom is for backward compatibility with older software.
Once the decision to use a non-Windows OS has been made choosing Atoms is no longer a reasonable option. Atom is a pig. Slick software design tools can't fix that.