Smaller is Bigger
ESC 2007 Tracks Technology

The San Jose version of the Embedded Systems Conference (ESC) in this year began with Al Gore throwing down the gauntlet.  Socially and environmentally conscious companies need not sacrifice the environment for profit.  In fact, the bottom line and the ozone layer may have mutual allies as emphasis on more efficient systems makes sense for both.  If the types of technology under discussion and on display at ESC are applied to the problem of creating more efficient systems, Gore explains, embedded engineers can make a significant difference in the troubling trends threatening Earth’s environment.

Once the keynote ended, the ESC show floor opened and the cacophony of massively parallel trade-show high-tech high jinks commenced.  Our Embedded Technology Journal editorial algorithms are finely tuned to sift through the din of product pitches and press releases to extract clear trends – sweeping generalizations that, while possibly correct for the industry as a whole, may in fact not be completely true for any specific system design project in existence.

With that not-so-fine print disclaimer, however, here are our projections for embedded systems technology in general (your mileage may vary, offer void where prohibited, actors may portray real people, embedded system generalizations may be illegal in CA, TX, NJ, and several other red and blue states) OK, that’s enough disclaimers.  Really, here are our trends now:

Embedded systems will:

1. have significantly increased functionality demands imposed by higher customer expectations, convergence, and ubiquitous connectivity
2. involve exponentially more lines of embedded software to meet those functionality demands
3. probably use either a Microsoft Visual Studio or Eclipse-based IDE for software development and debug
4. face an exponentially more difficult debugging challenge in getting all those lines of software to work together correctly
5. require dramatically more processing power to execute all that software
6. make use of multiple processors and multiple cores to achieve the required performance
7. demand more memory and more memory management to support the software and processor requirements
8. use FPGAs for flexible connectivity and algorithm acceleration – and sometimes as system-on-chip platforms
9. be built more often from pre-configured off-the-shelf hardware boards and modules
10. utilize virtual prototypes of hardware platforms for embedded software development and debug
11. use either supported versions of open-source operating systems like Linux or proprietary, royalty-based operating systems like Windows Embedded or Nucleus
12. often employ multiple, independent operating systems for real-time processes and more feature-rich non-real-time applications
13. take advantage of virtualization layers to insulate processor and operating system changes from each other
14. use pre-built, off-the-shelf middleware to integrate various peripherals and utilities
15. require more security than ever before - for the intellectual property involved in the engineering of the system, for the OEMs’ and content providers’ protection, and to safeguard the personal information of end users
16. rely heavily on standards to provide a framework to satisfy most of the requirements above
17. be developed in far less time than in previous years

Your Embedded Technology Journal team applied these trends to the incredible number of presentations and demonstrations that we consumed at ESC.  We have, for your perusal, our selection of the most interesting stories from the most interesting companies we visited:

Microsoft, the world’s largest software company (and one of the world’s largest companies in general) is determined to have a major presence in embedded systems design.  They have invested significant resources in building products and infrastructure to assure that presence, and they continue to increase that investment.  As a result, they have a compelling story to tell to embedded development teams, and most should at least give their story a listen.  It goes like this: 

1. Microsoft’s software development tools have probably the largest following in the industry.  In terms of the number of qualified software engineers proficient in their use, we’d speculate that nothing else comes close.  If you want an ample supply of potential programmers for your project, choosing a Microsoft development platform is the safest bet.
2. Microsoft is determined to win certain high-value market segments.  In these areas (you know who you are), the company has invested such significant development resources that most of your development work is already done for you.  You’d be kinda crazy not to take advantage of it unless you’re one of those died-in-the-wool NIH types.  If you are – take a look in the next cube.  That guy is using a Microsoft platform for his product development, and he’s already finished.
3. Microsoft has such a large following in the desktop development arena that much technology is flowing over into embedded.  In addition, if your embedded application needs to connect to a desktop computing environment as an integral part of its operation – guess who’s on the other end of the phone?

Wind River, the world’s leading embedded… what are they these days?  For a long time they seemed to be an embedded operating system company, then they were an embedded tools company, then they were a device software optimization company, then they were the world’s ambassadors to corporatized open source socialism.  Now, they simply have a better understanding of the embedded systems world than perhaps any other company on the planet.  While other big companies dabble in embedded design, Wind River lives or dies by their embedded prowess.

The company has re-invented itself many times over the years, and that is exactly in keeping with what the market has done.  In fact, Wind River should be a case study in high-tech companies rapidly adapting to market and technology shifts.  Today, with a significant investment in supporting Linux and a robust set of proprietary tools, systems, and utilities built around that offering, the newest incarnation of Wind River is not taking their market position for granted.

At ESC this year, Wind River was announcing new on-chip debugging support with Wind River Workbench 2.6.1 On-Chip Debugging Edition.  The new release has broader multi-core support, enhanced Linux debugging, and an Eclipse-based development environment.

LynuxWorks was following the reliability and security thread with announcements on upgraded versions of their popular LynxOS and Luminosity.  In both cases, the enhancements were targeted at multi-core systems that require some hard real-time, high-reliability, possibly secure processing components.  Then new version of LynxOS (5) supports symmetric multiprocessing (SMP), allowing multiple processors in shared memory to share processing duties to balance workloads.  It also provides a new Application Binary Interface (ABI) that allows Linux applications to run unmodified along with POSIX applications.  At the show, LynuxWorks also announced Luminosity 3 – an Eclipse-based IDE that supports the features of LynuxWorks operating systems.

Mentor Graphics has apparently finally completed the integration of their embedded division (formerly ATI).  Few traces of the non-Mentor-branded acquisition exist outside the enormous market share of the Nucleus operating system in the baseband section of cell phones and in similar applications.  Mentor is usually the outlier of the big three (or four) EDA companies, and in the embedded world, that position works to their advantage.  While the other EDA companies seem to focus almost exclusively on high-end ASIC system-on-chip development, Mentor invests significant resources in the board, FPGA, and embedded systems spaces.

At the show, Mentor was announcing a new API for the Inflexion UI for Nucleus.  The new API helps in the rapid creation and deployment of GUIs for embedded devices running Nucleus.  In addition to the Nucleus announcement, the company was flexing its “breadth” muscles, reminding us that they have a rich portfolio of hardware design, software design, and hardware/software co-design and verification products.  In addition to the former ATI embedded software products, their embedded offerings include comprehensive board design tools, FPGA design and verification tools, and hardware/software co-design co-verification as well as a relatively new line of electronic system level design (ESL) products. 

ARM and MIPS continue the divide-and-conquer approach to the processor IP market.  Both companies have provided outstanding processor IP for years, targeting mostly different segments of the high-volume ASIC world with high-performance processor cores.  Now, both companies are showing a full understanding of the reality that processor IP alone does not make a platform, and that customer design-ins are heavily influenced by the entire infrastructure surrounding those cores from development and debug environments to bus and peripheral IP, third-party OS, middleware and software support, and reference designs in similar applications. 

Now, ARM has announced a new processor optimized for FPGA use - the Cortex M1.  The new processor could easily become a disruptive force in the FPGA-based embedded system-on-chip market.  Now, the company is rounding up support from FPGA vendors and tool suppliers.  The first to announce support were Actel – with their non-volatile FPGA lines, and Synplicity – whose tools are widely used for FPGA design across vendors.  Expect ARM to twist the arms of companies like Xilinx, Altera, and Lattice soon, and then we’ll see what shakes out in the soft-core FPGA processor market.

MIPS continued their “biggest fish in our part of the pond” marketing strategy, pointing out that, while they are not the biggest processor IP company, they certainly have commanding market share in many segments of the market.  MIPS’s show presence highlighted design wins in digital home, communications, and networking applications where the company has seen widespread adoption of its processor IP.

Virtutech and CoWare are now both fighting the battle for attention in the virtual embedded platform arena.  Today, virtual platforms have the status of “great idea that we plan to try someday” in most embedded development teams, leaving a wide field for innovative companies to grow the market.  Virtutech has been in the business for years with their Simics product, and CoWare has gradually morphed their ESL pitch toward a virtual-platform-centric story as well.  At this point, both companies are working more to grow the awareness and adoption of virtual platform technology rather than wasting too much energy competing with each other. 

This is but a small sampling of the interesting announcements at this year’s ESC.  While rumors abounded that the show was smaller than usual, it was still far bigger than one could conveniently cover in a single visit – or a single article.   (We’ll have dedicated articles on several of the product announcements in future issues.)

Kevin Morris, Embedded Technology Journal

April 10, 2007

[back to top]