NVIDIA's Tegra takes on Intel in the MID/PMP market
http://arstechnica.com/hardware/news/2009/06/nvidias-tegra-takes-on-intel-in-the-midpmp-market.ars

When the Tegra line was launched a year ago at COMPUTEX, NVIDIA pitched it as being aimed at the mobile internet device (MID) market, and the idea seemed to be that NVIDIA could grow it up into netbooks and down into smartphones. Then at the Mobile World Congress earlier this year, the graphics giant renewed its push into MIDs with the announcement of a Tegra-based $99 hardware platform, which, at the time of the announcement, had scored some design wins with Asian ODMs.

At this year's COMPUTEX, NVIDIA is banging the MID drum yet again, and is showing off the fruits of those earlier design wins in the form of over 20 Tegra-based mobile devices that range in form factor from smaller MIDs to netbooks. I'll tackle the issue of Tegra in netbooks in a separate post, but as far as the Tegra-based MID goes, this is probably NVIDIA's fight to lose.
One form factor, two rationales, two acronyms, two markets

Intel's case for the x86-based MID/Ultramobile PC (UMPC) form factor is built around two potential usage scenarios, one for enterprise and one for consumers. The enterprise scenario is what I'll call the "legacy Windows app in your pocket" scenario, and the idea behind it is that there are all of these special-purpose enterprise Windows applications that companies would like to deploy away from the desktop, but without having to buy anyone a laptop just to run a single app. An example here would be some custom in-house Windows app that a FedEx worker might use to track inventory as he wanders around a warehouse, or a data collection app written in Access that a franchise manager would use in the course of an on-site audit of a retail location.

Tegra is bound to be a better fit in every conceivable way—performance, battery life, features—for a wireless, Linux-based MID/portable media player (PMP) than anything x86-based. This will be even more true once Adobe is done porting Flash and AIR to ARM.

So in this first use case, you don't need to run Windows on the go—you just need to run one specific Windows app on the go, hence the idea of purchasing a cheap UMPC and turning it into a portable, single-purpose widget that lets you deploy this one Windows app across an entire mobile workforce at a very low cost. (For such applications, Intel prefers the term UMPC to MID, though the hardware is basically the same.)

Intel's second, consumer-focused usage scenario is centered around Linux and is aimed at wireless carriers. The idea here is that a wireless company in Europe or Asia (they're not aiming at the US market with this yet) will offer an (x86-based) entertainment and connectivity device that runs a customized, carrier-branded Linux distro and connects to the wireless network. Customers can buy these to surf the Web, download music, and play videos, and they essentially function like giant cell phones. Both Intel and NVIDIA agree on calling this second, consumer-focused device a "MID."

(Incidentally, what both of these scenarios have in common is that they envision the MID/UMPC as an application-specific device, and not as a general-purpose computing platform that can provide a scaled down version of a desktop experience.)

Tegra isn't going to hurt the enterprise usage scenarios—the tanking economy has probably already done that one in, since nobody has room in their IT budget for an experiment like that. But I can't imagine that Tegra will not blow a big, fat hole in the second, consumer-focused scenario. Tegra is bound to be a better fit in every conceivable way—performance, battery life, features—for a wireless, Linux-based MID/portable media player (PMP) than anything x86-based. This will be even more true once Adobe is done porting Flash and AIR to ARM (more on this in the next article).

Vendors like Compal, who now are making MIDs and netbooks based on both Atom and Tegra, will figure out shortly that you get more media playback performance per watt from an ARM product like Tegra than you do from anything Intel is selling, and because it's running a customized Linux distro anyway the carriers won't care that it can't run Windows.

It's also the case that there is a huge ecosystem of ARM-based media SoCs out there to choose from, so if they don't like Tegra they can pick something else. Given that reality, why would they hitch their wagon to x86, where Intel is the only vendor with a SoC part?

It looks likely that in the long-run, the Windows-based netbook is the only place where mobile x86 is truly compelling, at least until Intel gets close enough to performance/watt parity with ARM offerings that it can win on price.

Source: http://arstechnica.com/hardware/news...pmp-market.ars