MIDs to Become "Most Popular Computer Type"

Shreekant Thakkar, UMG chief platform architect, Intel Fellow, Intel Corp of the US, believes that mobile Internet devices (MID) will eventually become the most popular type of computer in the world, evolving increasing "intelligence" by utilizing information from MID onboard sensors, and providing users with the same Web experience as is provided by PCs. Here, he discusses the ways in which he expects the MID to evolve.

##CONTINUE##

Would you describe your concept of the mobile Internet device (MID) platform architecture of the future?

I believe that the most popular computer type in the world, in terms of user quantity, will be the MID. As 3G mobile phones, WiMAX and other technologies spread globally, mobile broadband will become more common than it is now.

Mobile phones as they are today will remain in use, but I suspect they will gradually be replaced. MID-like computers will no doubt evolve to become more and more "intelligent" and the first step in that evolution will be utilizing information from MID onboard sensors. For example, position information from a global positioning satellite (GPS) sensor could be used to implement an application that searches out the closest and cheapest gas station, as a Web service. Already games utilizing acceleration sensors have appeared for the iPhone from Apple Inc of the US and other platforms. I think we'll see an increasing number of applications making use of equipment sensors.

The next step will be predictive applications. For example, the device might automatically show the user a map of how to get to a meeting place from the arrivals gate at the airport. Depending on local traffic congestion and other factors, it might suggest a taxi, bus or train. Increased MID computational power makes it possible for applications to recognize the specific context involved, using information from sensors, the Web or elsewhere.

Why did Intel feel it had to release products for MIDs? Isn't there competition with products for notebook PCs?

Yes, they do compete with compact notebooks with 8-inch displays, the so-called NetBooks, but that's really only a question of where you draw the line. Intel as a company must expand the market scale. The front-runners in MIDs are products like the iPhone and the N810 from Nokia Corp of Finland. The MID market was established by the iPhone, and in 2008 grew in scale to about 10 million units. It is still an immature market, and it's clear that it will continue to grow. The decade starting in 2010 should reach a shipment volume of hundreds of millions of units annually. In addition, the Atom microprocessor used in the MID platform is not limited to MID applications. It should be possible to apply it in other product sectors as well, and we believe that this will expand the sale of markets addressed by Intel. 

We feel that MIDs must be able to provide users with the same Web experience as is provided by PCs. Existing mobile phones, even if they can connect to the Internet, don't really enable people to utilize the Web in the same way that a PC does. Social networking services (SNS) like YouTube, FaceBook and MySpace are immensely popular, but mobile phones provide only very limited access to these services. I think there is an urgent need for a highly portable device that can deliver the same experience that a PC can deliver.

Since Intel announced the Atom processor in March 2008, there have been comments about it being superior to the ARM architecture in terms of software compatibility. But is it possible that the problem is with the software, not the microprocessor hardware?

When software is ported to the ARM architecture, then yes, you're right. What we're talking about is the difference in how long it takes. Suppose a new Flash player appears for PCs, for example, and Websites begin using it. Equipment using Atom will be able to make use of that Flash player immediately, providing users with the same experience they get from a PC.

It takes time to port software over to the ARM architecture, though, and the software has to be ported again every time a new version is released. As long as that delay and effort continues to exist, the ARM architecture can't approach the PC experience.

Which design guidelines were given top priority when it came to developing the Atom processor?

We assigned top priority to assuring 100% compatibility with PC microprocessors like the Core 2 Duo. Intel has a policy of shipping products assuring instruction set architecture compatibility. This is a fundamental difference from the ARM architecture, which has had a different instruction set with each central processing unit (CPU) core generation, and I believe it is one of Intel's strengths.

Intel used to have an ARM architecture microprocessor, too, but a microprocessor that software won't run on isn't worth much. A few years ago we recognized that using Intel Architecture (IA) was critical. I suppose we could have adopted a different instruction set architecture for low-power, but we elected to stay with IA. Development proceeded from the IA basis, keeping dissipation as low as possible for portable equipment. We enhanced the design to boost computational power as much as possible while always keeping an eye on dissipation.

Work continues on the "Moorestown" (codename) platform for MIDs, scheduled for release in 2009 or beyond. This is a 2-chip design, consisting of the Lincroft microprocessor and the Langwell I/O controller hub, with the microprocessor single-chipped with the graphics processing unit (GPU), image processing circuit, memory controller, etc. Why did you move these other components, integrated in the I/O controller in your existing products, over to the microprocessor?

The GPU is going to be about the same size as the CPU core, and the image processing circuit will have to be able to code and decode high-definition TV (HDTV) imagery. The volume of data being handled has increased significantly, and we had to expand the memory bandwidth. We want to be able to use the 45nm manufacturing technology we're using for the microprocessor on circuits like these, which we expect to grow even larger in scale in the future. I'm afraid I can't provide details of just what sort of circuits are being integrated into the Lincroft, but I can say that it will deliver improved graphics draw and image processing performance.

The Langwell is more than just an I/O hub; it also integrates a solid-state disc (SSD) controller, power supply management circuit, etc. In the Menlow (codename used for the Centrino Atom, the current ID platform) we needed some external components like the auxiliary storage controller and some power supply management parts, but these external components are no longer needed. We expect to be able to achieve reductions in cost, dissipation and mounting footprint.

-----------------------------

Interviewed by Tomohisa Takei

Source：TechOn!