Intel has come full circle by bringing on Mike Bell (ex HP <-  ex Palm <- ex Apple) to lead smart-phone referencedesign.  Don’t just work with top players but enables others also – what a novel idea ?  Why didn’t we think of this before, wait a minute, we did but then we stopped.  Intel dropped doing reference designs for UMPC /Smart-phone space over three years back. But this is Intel,  they have money and influence to make up for the lost ground with the right people leading the effort.

On Linked in the appearance of Title ” Smartphone marketing Manager” for Intel employees is either sign of major changes or employees making themselves more marketable. Who wants to be associated with just plain PCs ?

So what happens to the tablet UMPC?  This is under seasoned Intel VP Mooly Eden, MPG ( mobility platform group) , they traditionally seems to always end up with products which makes Intel lot of money.  We certainlyhope their will be a lot of  successful iPAD like tablets with Intel hardware.  Other progeny of UMPC, Netbooks have peaked but are here to stay and UMG has done exceptional job with them.

Editors Note: Got to admit author is biased because of his past association with Intel. (Still cannot seem to get myself to buy AMD even if it is better deal and in spite of living in Austin, TX  –  stomping grounds of AMD.)

The acquisition of P.A. Semi (a California chip design start up) was first indication that Apple not happy with current chip offering and wants to do its own chips. Since it early PowerMacs day, Apple is used to influencing the chip hardware directly. Motorola and to some extent IBM were much more willing to be cooperate with Apple on this. Intel is completely different ballgame. Intel likes to be in-charge and complete control of its chips.  No wonder Intel Atom has not been able to make inroads into Apple mobile devices.

And now Apple’s acquisition of Austin chip design company Intrinsity gives us confirmation of this direction.  From  grapevine in Austin  “Intrinsity has been directed to do SOC” (System-on-Chip)  for Apple.  Given Intrinsity’s low power and high performance experience this has potential of producing some awesome custom SOCs for Apple.

So where is Intel headed with Atom ?  Answers lies in Finland … next time …

If we ask Intel answer will be most likely no.  Why ? because Intel defined UMPC as having x86 (official term from Intel is IA – Intel architecture) processor and chipset

Lets try asking Microsoft( Or Origami folks). Most likely answer here will be no as well because they defined UMPC as running a windows based Operating System.

As you can read and our  History & Definitions section. iPAD fits our definition of  UMPC  without any modifications at all.

Here is the text from that section:

Definition

The current widely-used definition for UMPC is: A mobile device with diagonal screen size from 4″-10″ running any OS and supporting more than one computing/communication function.

Further Details:
Power – A mobile device has a battery by definition (at least for now)
Hardware Platform – No restriction
Touch screen – Common but not required

Next time : so what is Intel and Microsoft up to and Apple’s hardware plans, hint: Intel’s new marketing designation: Smart-phone marketing manager, Apple buys chip design start-up in Austin ?

In 2005-2006, Intel and MS started pushing a new product category called UMPC. The Intel and MS UMPC teams who created this concept insisted on machines with 7” screens and touch panels. Basically, they were small tablets. At the time, however, most ODMs/OEMs did not want build or sell UMPCs as defined by Intel/Microsoft. They were skeptical about this definition of UMPC and its market viability. But they loved the low prices of the hardware and software offered by Intel and MS. OEMs/ODMs tell us that at the time they really wanted to build notebooks with smaller screens (7”-10”) and sell them at lower prices but still higher margins. However, both MS and Intel opposed building cheap notebooks. MS most likely was interested in expanding its OS install base, not in to replacing one install base with cheaper install base. And Intel most likely did not want to cannibalize its high margin business of laptop chips. The ODMs/OEMs told umpc.com that some of the Intel leaders on UMPC team also wanted to build cheap notebooks but where being stopped by Intel’s famous “Disagree and Commit” adage.

Side note: Why was 7” set as the screen size? Simply because it was the only cheap mass produced LCD available at that time.

What really happened? Well, the market determined its own course. UMPCs, as defined by MS/Intel, met with only limited success. After this, Intel’s new UMG (Ultra Mobile Group) executives started pushing a new term MID (mobile-internet-device), compelling rehash marketing of an old vision of a do-all-mobile-device. Meanwhile the ODMs used the learning they had from building 7” tablets to build 7” laptops and then moved on to 8” and 10”. ASUS’s EEMPC and Inventec’s Kojinsha were results of this effort. By this time the MPG (Mobile Platform Group, responsible for Intel’s laptop business) executives had seen the light and went on to officially support ODM/OEM efforts to make small inexpensive notebooks (and left UMG to focus on MID). But the issue with cannibalizing higher margin notebooks remained. Enter the Netbook hardware platform from Intel, dubbed Atom, with a few limitations over notebook hardware and officially differentiated as for light work. As a result, the Atom processor offerings from Intel were split: an Atom line for Netbooks and another Atom line for MIDs.

So did OEMs/ODMs get their wish by creating an inexpensive but high margin product for notebook (called Netbook now) market? Not really. Again the market ran its own course. Every OEM/ODM got into the fray of building the new netbooks, the prices were driven down and

netbooks have turned out to be even lower margin than notebooks for OEM/ODMS.

The winner in all of this is of course the consumer. Consumers now have greater choices at lower prices.

So who would Intel try to woo next? Motorola is experiencing its own difficulties in this market (they did launch an ARM-based MID-like device, the V6096, but we are still trying to figure out exactly what this device is supposed to be). What about Nokia, who has met with limited success with their N800 internet tablet? That would be good match, but is Nokia going to bite? Their problem with N800 was not its hardware platform, so do they really need a different platform (for example, Nokia has Flash Lite working on several devices by leveraging their influence with Adobe, an important achievement which most ARM devices do not have). Instead, the problem is the user interface. The blind belief that higher resolution is always better has tainted the vision of many mobile devices.

Other large companies that Intel will probably target are Samsung and LG, especially Samsung for its early enthusiasm for UMPC. That team must be looking for next thing to design and so far all innovation from Samsung in this area has been incremental.

We think Intel should stop thinking like an incumbent, and instead think more like a newcomer in the mobile device space. Newcomers do best by partnering with many other newcomers instead of trying to partner up with gorillas. But, the Intel execs running the MID show came from running gorilla businesses and with gorilla ambitions. Can they successfully change their expectations of the business model to fit the mobile market?

So, what can Intel really offer which ARM doesn’t?

Here are some general guidelines learned from our experience architecting some of the early UMPC products in the market.

• How the device will be used needs to be defined before choosing platform.
• Starting the process by negotiating with hardware and software vendors to use their platform is a backwards approach to design. What matters most at this stage is how the device will be used!
• Designs which do not look beyond gadget enthusiast are only good as a design learning experience.
• Connectivity and GPS should be baseline requirements for a mobile device.
• Unless the device UI is Windows, which serves a limited but important audience, the UI and man-machine interface (MMI) design should be in forefront when designing device hardware and software.

So what would be ideal device which is possible with current hardware/software? Stay tuned for our answer to this question.

Platform and Chipset Selection

As a UMPC designer, selecting a CPU/Chipset platform is one of the important decisions. Although this is not the first decision to make for a product, it is a rather important decision and one frequently asked us. It is also an area of deep expertise for umpc.com’s technical folks, so we will publish on this before addressing first concerns like device functionality and user interface. So, to start let us examine Intel’s platform offerings (processor/chipset) for UMPC/MID mobile devices, starting from the earliest offerings.

Gen 0 – The platform used by the very first UMPC devices by ASUS, Founder, and Samsung.

Dothan CPU (Pentium M-based) 900 Mhz/1GHz + 915GMS + ICH6 Chipset.

This is same chipset used in laptops until mid 2005. All the first generation UMPCs from Intel used this chipset. Power was too high, resulting devices with fans and short battery life.

Gen 0.5 – McCaslin Platform. (Intel A100-A110 (Stealey)) Launched Q3 of 2007

Stealey CPU 600Mhz /800 Mhz + 945GU + ICH7-U. This CPU is repackaging of the Dothan CPU into smaller package and binning parts to get lower voltage and lower power parts. These improved parts were sold as Intel A100-A110 CPUs. The smaller package was supposed to reduce the device size. But power (and as a result, heat) are still too high for this package for several reasons, so the device is still large and includes a fan. The cost of platform is even higher because of the 8-10 layer PCBs which are required by smaller package with closer pins. At the same time, other components are still standard laptop components, which keeps the PCB a large size despite the increase in layers.

Gen 1 – Menlow Platform. Branded as Intel “Atom”, Launched end of Q2-2008.

This is the first true UMPC platform designed from ground up.

Silverthorne CPU 600 Mhz/1.6 Ghz. Poulsbo SCH (combined system controller hub).

Intel had to start this design for this as early as 2004-5. The goal was to further shrink power, but unfortunatelly smaller Silverthorn devices still require a fan. However, Intel finally listened to its customers and created a standard package size version of Menlow, making Netbooks take off with low cost PCBs (recall that the smaller package requires a more expensive PCB). If the popularity of clamshell/Netbook/UMPcs is any indication, then this platform will realize the UMPC vision of a portable companion device, without the artificial limitations (touch screen only, no keyboard, etc.) imposed by Intel/Microsoft.

Now more on the current Intel Atom based chipsets:

Atom Z is 800Mhz-1.86 Ghz CPU with deep sleep capabilities. The package is much smaller and as result the ball-pitch between pins is lower. This makes expensive 8-10 layer boards a requirement. Atom Z is paired with System Controller Hub (SCH aka Poulsbo). This is the CPU most often used in MIDs.

Atom N is 1.6Ghz CPU with restricted deep sleep capabilities. It is larger package with ball-pitch of pins allowing for standard 4-6 layer mobile boards (which are cheaper than 8-10 layer boards). This CPU is commonly paired with Intel chipset 945GSE in Netbooks and other UMPCs.

For unknown reasons (as far we can understand none technical) Intel does not sell Atom N with full capable deep sleep enabled. We think this will make for a much more interesting platform for UMPC and Netbooks. The standby power will go down, resulting in longer battery life while cost will remain low because of 4-6 layer PCBs.

Gen 2 – Moorestown Platform. “Atom” Brand will probably continue, expected launch Q4-2009.

Next-gen Atom CPU + separate integrated chipset (still no SOC). This generation will still have two packages, but if Intels’s push towards MIDs (Linux-based) continues, then platform almost definitely will not support Windows (any version but CE/Mobile). If platform power does go down by 10X as been touted by Intel mobility execs then finally no fan will be required. Products based on this platform will probably not launch until late 2009.

Design Choices

During the Gen 0 – Gen1 product launches some companies decided to build devices based on standard Intel mobile chipsets—this resulted in the breakaway popular ASUS EEEPC clamshell/netbook device. The Intel evolution described above through Gen 1 is an ideal platform for this type of netbook device.

That is, the design choice for sub-10” clamshell/ tablet devices to run Windows XP or any powerful Linux variant is unquestionably the Intel Atom platform. This is because the only other options are Via and AMD, which have much too high power for the equivalent performance.

As for the design choice for 4-6” MID/UMPC running non-Windows OS, this is entirely different story. Ironically, the “MID” device platform that Intel’s marketing engine is aggressively pushing, is not well suited for such a small device. But Intel is giving up the Windows XP, or equivalent, market by downgrading the next-generation performance capabilities, without increasing the peripheral integration nearly enough. Next week, we will discuss what is the best platform for a 4-6” MID. As usual input is welcome, keep sending your emails to info@umpc.com.

The sub-notebook of yesterday is the UMPC clamshell of today and will exist as a cheap portable entry level device of future.

Regarding the device platform, the Intel-Atom will become the platform of choice of all devices running Windows, while ARM-based devices become standard in all non-Windows machines. Watch for the Qulacomm Snapdragon to do well in this space.

Size and Resolution

There is race to miniaturize devices and increase resolution of the screen at the same time. We think this is wrong trend. 1024×600 resolution on a 4.8″ diagonal screen is just not very useful. We will discuss optimal resolution for a given size device, held at a given distance from eyes in the design guidelines section.

Clamshell Devices

UMPC on clamshell form factor evolving as net-book. In our opinion this is expected to be the large growth market.

MID UMPCs will keep getting thinner and smaller to fit in pocket, for now any reasonable device in this space is based on non-Intel Atom (most likely ARM) solution.