I am now back from an absolutely exhausting trip to Embedded World 2010 and remain incredibly excited about the ARM Partnership’s prospects in this space. Cortex-M4 is less than two weeks old, fits perfectly alongside the wildly popular Cortex-M3 and Cortex-M0 (the M1 is all about FPGA), and the people I spoke with have grand plans for its future. 5 licensees are already in place, with 2 in the shadows, and you can imagine many more in the future pipeline. On top of that, I spotted a few Cortex-A8 boards in several booths, well outside of mobile phone applications. The ARM Partners are ramping up big-time and I think the world will be completely different this time next year.
Last month I observed that Intel’s booth at Mobile World Congress was both uninspiring and comparable in size to Tensilica’s, which felt rather odd. I thought latter’s booth quite nice, by the way (e.g. no offense!). Last year I understand that the Intel presence at Embedded World was somewhat overwhelming with banners touting the next 15 billion intelligent, connected devices by 2015. This year I found Intel’s Embedded World booth smallish and uninspiring. Something is happening here…could they be beginning to pull back?
Anyway, the key questions I pondered throughout Embedded World included:
- Why is the embedded world so fragmented?
- Why/where will proprietary architectures (Intel and many others) be successful?
- Why/where will the ARM architecture (through the Partnership, of course) be successful?
- Is one model inherently better than the other in this space?
[Note: if you are cut/pasting this content to another web site, please stop here and include a link to the full text -- http://wp.me/ptyFX-hB -- as I analyze traffic to understand interests and drive future post topics. Thanks]
First, let me point out that fragmentation is just the other side of diversity. Diversity in applications drives diversity in requirements which drives diversity in suppliers and solutions. Yes, the size, difficulty, and cost of our challenges only increase but I see no real reason diversity will fade. The old joke goes along the lines of “the only things that have ever truly converged is the alarm and the clock”; seriously new opportunities always appear so any process of convergence can never complete. The Embedded space has been diverse for decades upon decades upon decades and no single company has the silver bullet.
Different architectures will have differing degrees of success in proportion to how well they serve the application, customer and market.
Atom will do well for cases where you need a board with the software foundation in hand, you want to add magic at the board/system/application-software level, and a lower-volume somewhat cost-insenstiive product can result. It won’t be smallest, optimized to nth degree, or cheapest BOM cost, but the development costs will be reduced (which is key). I am not sure how much Atom really unlocks in this space relative to other x86 offerings that have been in play for over a decade though. However, it does work very well and I respect that greatly.
I think folks will acknowledge that the collective ARM Partnership offerings are virtually the definitive example of diversity, and based upon collaborative differentiation. Success will first build in the reverse of the above case – where size, optimization, BOM cost rise in importance. Hello high volume! BTW, NXP have been very prolific with their Cortex-M0 based silicon down to $0.65 promotions. As for optimizations, have a look at Energy Micro who are winning awards. Look mom, no heat sinks.
The key gap in the latter case is development costs. If that were reduced, it would be unbeatable. Hmm, do we see a strategy, shall we race towards the next 15 billion?
For some more Embedded World coverage, have a look at the ARM blogs starting here and especially here. I took 194 pictures at Embedded World, which can be found here.
Here are some related press releases from the ARM web site
