‘Unleashing the Internet Experience.’ IDF Keynote analysis.

Yesterday the traditional ultra mobile PC was pretty much ignored by Intel as they went from Notebook to MIDs without a nod at the UMPCs. And when I say UMPCs, I mean Microsoft Origami-style, Ghz-class mobile productivity devices running Windows operating system. Intel are clearly focusing on a new market for ultra mobility. Its a new customer target (can we hear them saying ‘2nd PC’?) and has total separation from the notebook business.

It makes sense of course, but for anyone hoping for leaps and bounds in the ultra mobile productivity space it seems that Intel is happier to let people like Sony do their own thing with the higher-powered and much more expensive ULV Santa Rosa and Montevina platforms. I feel sure that we’ll see new Ultra Mobile productivity devices but sub $1000 is not going to be realistic and I don’t think we’ll see the numbers of new models increasing dramatically. Its going to be a niche market.

So what’s happening in the ultra mobile space according to Intel? Here’s a detailed analysis of the keynote speech given by Anand Chandrasekher – ‘ (Click ‘Read more’ if you don’t see the full version of this report.)

“The handheld mobile experience today is not great. It’s a subset of the full Internet. You don’t get adequate performance. It’s not compatible”

I think that gives you all the clues you need as to where Intel are looking. Its not about productivity, its about mobile Internet. The Full Internet Experience, or ‘FIE’, as we have talked about a lot here on UMPCPortal. Intel defines it as three things.

• Software compatibility for developer and user.
• Mobile responsiveness and visualization.
• Always on. Always connected.

Two of those points I detailed in my report ‘What’s is the FIE’ but the third there, Always on, Always connected is absolutely correct. Although it actually relates to the FIE device itself, its critical. Hotspots are useless for ultra mobility. 3G is where its starts (not 2.5G Apple!) and as there’s a huge opportunity in the U.S. with WiMax, Intel are obviously putting efforts into highlighting this.

Anand went on to talk a bit more in detail about software compatibility and took direct aim at the ARM-based devices with a slide that I’m sure we’ll discuss below. It trys to highlight the difference between the x86 (IA) and ARM-based Internet experiences. Although ARM is not really the problems here (its actually the software coding and porting process) it does demonstrate the key point about FIE and I don’t disagree with it. In fact, it doesn’t actually show the extent of the problem. Speed isn’t even considered in the diagram. Here’s the slide.

If MIDs are all about FIE then, unless the ARM software development process suddenly starts getting more efficient and catches-up with the x86 development process, its critically important that they are based on x86 architecture. [Question: Why doesn’t Intel use the ‘i86’ term anymore?] One thing that’s not mentioned at all at this stage is component cost.

Anand then starts to talk about Menlow. Much of it is what we’ve heard before. 10x power reduction, size reduction and retaining performance levels. Its never been clear to me exactly what the 10x reduction meant but it all became clear in the demonstration that was given on stage. Side-by-side, a 5.5 W TDP CPU that you’ll find in Intel based UMPCs from 2006 was compared with a Menlow CPU. If we trust Intel that this was a like-for-like processing power demo (there’s no mention of clock-rates or of how much data was actually crunched) then the results are quite amazing. Where your 900Mhz Celeron tops out at 5.5W, Silverthorn, the CPU part of Menlow, tops out at under 0.5W. This is not a demonstration of power saving states it a demonstration of power saving under load and that, for Intel is a big step.

However, it might not be good enough. At the point where the silicon goes sub 1W, Intel need help. Its all very well putting a sub 1W processor in a ultra mobile PC but if the screen uses old backlight technology or the WiFi unit is taken from the 2006 parts shelf, the design is seriously compromised in terms of battery life. Intel needs help. Radios, screens, disks and motherboard design are the key areas. Of course, battery tech can help too but my felling is that the battery technology will follow when the sales numbers increase.

As for radios, Intel are looking into integrated Wifi, BT, GPS and WiMax. These are areas in which Intel already have expertise and Menlow will use these. Its interesting to note that they won’t bother with 3G silicon. Option are the partner for 3G and 3.5G cellular radios and according to the presentation, Option have a new mini card ready for Menlow.

Anand then went on to talk about other partners and showed a range of devices. A few McCaslin-based devices where shown (Q1 Ultra, Inventec/Kohjinsha) along with the demonstrator from EB, the R3 from Asus, Benq’s MID and prototypes from Quanta, Compal and Inventec. All of the new MID devices (shown in the slide below) look like they use the 4.8″ 1024×600 Samsung screen but that’s only a guess at this stage.

There was no mention of disk technology or screen technology. Screen tech was mentioned back in April where Samsung detailed their new screens. 4.8″ 1024×600 is the most likely candidate here and its a 0.9 watt device (screen and backlight) but what about disks.

Personally I think that this is where Linux plays its role. MIDs simply can not use rotating disks. Too hot, too big, too fragile and too power hungry. They have to go in favor of silicon and this is where Microsoft is shown the door. They do not have an capable finger-driven operating system for small screens. They do not have a cheap licensing structure for desktop OS’s and they do not have an operating system that fits the low capacity, cheap flash drives that will go in Menlow devices. You can not put 10GB of flash drive in just to hold the operating system. By using Linux they can do these 6 things.

1. Reduce image footprint to sub 1GB levels.
2. Reduce ability for consumers to install additional software
3. Reduce license cost
4. Do the core OS development in house with existing, free, software
5. Hand out the GUI and app design to small, quick to react, opportunity hungry ISV’s / Web2.0 developers.
6. Provide the all-important market separation as seen from the consumers side.

Linux is clearly the only choice and already we’ve seen the Intel Moblin teams generating core builds that are already in use by development teams in Ubuntu and Red Flag. Although we’ll have to wait for next April to see consumer editions of Ubuntu, its still going to be an amazingly quick turnaround time and i’m sure the costs savings are going to be massive. This could be the single most important project for Linux-based software and its no wonder that Mark Shuttleworth turned up to sing the praises of it. This could be show-time for Linux.

After Mark Shuttleworth came on stage we saw Adobe. Adobe is there to help the gui and app developers. They see a big opportunity for Adobe Integrated Runtime (AIR) which can be thought of as enabling slick Web2.0 apps running outside of the browser. Flash for desktops perhaps. You can see examples of it running on the Asus R3 ‘UMPC’ (interesting choice of term by ASUS) image below.

That’s the Searchcoders Dashboard (note and RSS feeds) and AMP (Adobe Media Player – think ad-ready, DRM-happy FLV and H.264 client.) demo applications showing there. You can actually try out Searchcoders for yourself here. This will be ready from Adobe when MIDS launch next year.

Al Ramadan of Adobe said this:

the great news about AIR is that it allows the rich Internet applications that many of the developers have been imagining to step outside of the browser. It now becomes a first-class citizen of the desktop, Anand. So if you — in your dock or in your start menu or however else you get to native applications, these rich Internet applications can now exist as first-class citizens. If you take the network down as you’re using the application, it doesn’t just die on you. It continues to behave in a way that an application would, and when the network come back up it syncs. And of course you use the same development tools that you’ve been using for the Web to deliver these kinds of experiences.

After the stage shows, Anand summarized by saying that everything was on track and it looks like the silicon will ship in Q1 08. Personally I think this hints at CES demonstrators and Spring IDF launches. This is similar to what we saw with the Q1 Ultra. March announcement and summer availability. Its on track but there’s still a way to go and I think it will be almost another year until we have choices based on Menlow. And even longer until we get to Moorestown, the next-gen after Menlow. I’m not going to go into that here but you can probably look forward to higher integration (small and cheaper) with slight increases in processing capability and continued improvements in battery life. The demonstration video showing the ‘iPhone stretch,’ as i’m going to call it, was shown and you can see that over at jkkmobile. Click below.

Anand ends up with the following statement:

So if we thought cell phones were a big deal, this is a huge deal

Thats a bold statement Mr Chandrasekher!

In summary then I think we can safely say that Intel are trying hard to differentiate MIDs from mobile productivity devices. They obviously see that productivity is linked to keyboards and that keyboards means larger, more powerful and more expensive devices. My opinion is that we’ll see some high end UMPCs and tablets in this space but if Intel get their way, they will remain niche products that utilize notebook platforms.

Does this mean the end of the term ‘UMPC’ then. Intel don’t use the expression anymore. VIA don’t use the it either. Microsoft are silent and all that remains is the marketing teams that are pretty much riding on momentum that the term has at the moment. I’ve always maintained that ‘Ultra Mobile PC’ meant a wide range up devices from the N800 to the Flybook and I will continue to use the expression in that way and I have a feeling that the expression will continue to be used. If only for the fact that, of all the terms that are being used, ultra mobile PC is the only one that brings up clean search results on Google! It also happens to be the name of this site so yup, I’m a little biased there ;-)

So there’s my analysis. I enjoyed the presentation immensely (I think I’ve watched it three times now!) and it really makes me happy that my old Carrypad product definition is turning out to be something that could be enabled by Menlow. I would have liked to have seen a live stream going out (Intel, why didn’t you do that?) but for next April’s IDF, I plan to, somehow, get there myself. Beijing here we come. Does anyone want to join me / host me / sponsor me?

You can watch the complete video presentation with slides here. The transcript is here (PDF) and I’ve copied some hi-res versions of the slides into a gallery. I hope it stirred some thoughts for you. Please feel free to register and comment below.