While ARM system-on-a-chips (SoCs) are dominating the fast growing cloud client segment (smart phones, tablets, e-readers etc.) Intel has finally begun marketing his 1st generation Moorestown to generate design wins needed for next year’s Medfield 2nd generation SoC “product delivery en-mass” successes. Marketing in a true, positive sense by giving professional quality SoC information.
- Update: Finally the information detailed below is surfacing even from the most Windows slate concious manufacturers. See: MSI waiting on Intel Oak Trail for Win 7 tablet, Android version will hit before end of the year [Aug 23].
This week the CTO and Research Chief, Justin Rattner first time has given a “behind the scenes” look into their effort. See Gizmodo’s: Intel’s Chief Wizard Conjures the Cloud, Apple and a Phone That Keep Secrets. His answer to the question “Why is Intel not really in mobile phones yet?” is worth to be quoted here:
There was a lot of concern that what became Atom would cannibalize the laptop business, which didn’t materialize. Instead we created a whole new category around netbooks! But that was the big fear, and that cost us probably a good two, maybe even three years, before everybody was convinced we could introduce a cheaper, slower, more energy efficient product and not damage the main revenue.
Not by coincidence we have had this week another “behind the scenes” look given by Shreekant (Ticky) Thakkar, “father of Centrino” and lately (for years) the leading authority on Moorestown. See the interview with him in tom’s hardware: Tom’s Talks Moorestown With The Father Of Centrino. Here we have another quote worth to be included here:
[Re: Why did it take Intel until now to come out with its own SoC?]
That’s kind of a complex question. Let’s talk about the notebook, which was my last platform before I worked on this one. The notebook platform has very little motivation to shrink in size, especially in desktop replacements. Several years ago, we were trying to get the desktop side to adopt a lot of the notebook’s capabilities. [Ed.: Presumably, this refers to the mobile-on-desktop effort back from the Core Duo days.]
But the desktop industry and users had very little motivation because of the developed component ecosystem for power delivery, heatsinks—the whole nine yards needed to build a desktop. And a similar thing has evolved around the notebook. The need for space and capabilities are very different between these platforms. When not driven by the constraints of size and capability, these platforms can use existing components and programable logic to do, for example, video decode and encode. There’s little motivation for them to move to an SoC-like environment.
But when you come to such things as handheld devices, set-top boxes, and embedded systems, all of these have size and power constraints. Constraint is the mother of necessity that drove us to designing SoCs. We needed a such-and-such size chip with certain capabilities and power—high performance CPU, memory controller, graphics controller, video controller, decoder/encoder. You have to wire all of those things up into that limited real estate. That’s what drove us into building an SoC for this class of devices—need more than anything else.
When reading those articles I’m suggesting to go through the comments as well. They will give you an idea of typical external reactions to the views of these most competent Intel insiders.
And a final remark. This is first time Intel clearly communicates why Microsoft Windows (as we know it today) is not running on these x86 processors:
Ticky Thakkar: We had a great, ultramobile, PC-class device platform before. It was low power…but not low enough. So the big challenge for us was to figure out how to deliver low idle power in this kind of form factor. We have work going on in our labs with OSPM—OS Power Management. We want to use Intel’s process strength to get as much out of the process as possible in terms of performance integration and the low active power you can get from the smaller transistors. We introduced the notion of power gating at the island level. We want to use only the power needed to do the activity that you’re doing, not switch on anything else. Today, most operating systems, such as the larger Windows-style OS, do quite the opposite. It’s like opening your front door and the whole house lights up. That’s not what we wanted to do here.
Tom’s hardware: Which explains probably part of why Moorestown for phones doesn’t support PCI and thus Windows, but larger form factor Atom platform versions, including the forthcoming Oaktrail platform, do.
Ticky Thakkar: There are many factors at work here. … In transitioning from a PC to a handheld device, we don’t need to use some of the PC I/Os. We kind of got rid of PCI Express and put in handheld I/Os that are more pertinent to what we need. Things such as MIPI I/Os. MIPI is the handheld I/O organization. To give you an idea, the difference between LVDS and the MIPI interface, just the interface link power is about a 125 mW difference. All these kinds of things allowed us to get to lower power.
To understand the essence of Windows capable “Oak Trail” SoC derived from Moorestown and to be made available in “early 2011” read this Gizmodo news nugget:
Intel “Oak Trail” Is Official: Tablet Processors With Windows, Android, and MeeGo Support.