EDA has to expand just to keep pace with its core customers’ challenges.
By Frank Schirrmeister
The electronic design automation (EDA) industry has now been trying for at least a decade and a half to catch up with software developers, for two main reasons. First, there are so many of them that it would be great to expand EDA into that domain. Second, semiconductor companies, i.e. the core customers to which the EDA industry sells, have had to add more and more software responsibility to their portfolios. In fact, they now often have more software than hardware developers, as I’ve tried to demonstrate in the past in previous posts like, “Putting Kurzweil’s Singularity to The Mobile Test.” As a result, EDA naturally has to expand to address the challenges its core customers face.
As a reality check that these changes are really happening, I went back 16 years to the announcement of a previous foray into system-level design and re-read the release and article about the Felix initiative, “Cadence Partners with Electronics Innovators to Accelerate Delivery Of New Environment for Software and Hardware System Design”. The partners were ARM, debis Systemhaus, Magneti Marelli, Motorola, National Semiconductor, SGS Thomson, Symbionics, and Ericsson. Well, ARM is still providing IP and Magneti remains a constant in the automotive sector. The rest of the companies has been acquired, split up into semi and system, re-structured, morphed back from a system-on-chip company to an analog one, and then acquired and re-focused from a mobile phone leader to a network infrastructure provider.
So where are all those software developers? And what can EDA and system-level design do for them? That’s where it becomes a bit more complicated. At DAC 2012 during the ESL Symposium, I presented a slide with data provided by Research2Guidance showing how by the end of 2011 more than 1.9 million apps were already available though apps stores. While there certainly is a huge number of programmers behind those, the vast majority of these applications can be written with vendor-provided software development kits (SDKs). The SDK itself is essentially part of the cost of developing the devices which would not be successful without the apps fueling them.
But interestingly enough, in the world of iOS and Android, as indicated in the graphic associated with this post, there were dependencies even from the apps down into the hardware, naturally lending itself to EDA to provide early development platforms to enable software development and verification as early as possible. From the left to right, firmware and drivers can be developed for IP blocks, sub-systems can be executed together with their firmware, operating systems, drivers and middleware, and the system-on-chip (SoC) itself can be executed in simulation and hardware acceleration, all the way up with test applications and apps the user sees – especially in the context of optimizing for low power.
The reason for dependencies here is that semiconductor companies and even IP providers like ARM are optimizing their hardware offerings for specific operating systems – Linaro, for example, allows ARM and its licensees to make sure the architecture specificity of their hardware can be properly exhibited to Linux in order for that to be optimized for the hardware. The OEMs, on the other end, are taking OSs like Android and are adding their specific differentiation to stand out from the “Android-crowd” – HTC’s “Sense” and Google’s add-ons to the software-enabled user experience of the Nexus 10, for example. The resulting landscape is one in which apps developers must make a choice as to which platform to develop for first, and market analysts are closely following what platform is leading.
Well, how is that landscape developing? The next level of truly hardware-independent applications is already here. They are based on HTML5. Recently I was able to play with a Chromebook at a friend’s house. It essentially brings up a browser and that’s it. One adds apps to the browser and customizes the user experience. The Chromebook becomes a portal to access all your applications on the web, validating the Sun-Oracle vision from 1996 when they introduced the original Network Computer, and this time around the infrastructure seems to be ready. And this is not only on devices that look like a thin laptop. At the recent Mobile World Congress in Barcelona, Mozilla and 11 network operators announced how Mozilla expands Firefox OS globally into cell phones, creating a similar experience for mobile devices.
Now, that is an interesting trend because the apps now are truly becoming even more independent from the hardware, making their developers more difficult to catch for us in EDA. A respectable number of software developers are porting Chrome and Mozilla onto the various hardware platforms, but the developers of end applications are becoming even more independent from the hardware. Another related trend becomes clear when looking at some of the first applications like the Financial Times App. The landing page sums it up crisply: “The FT web app […] is available via your Safari browser at app.ft.com rather than from an app store. The web app is our most complete app to date and we regularly add new features and sections to it. These are available instantly, without the need to download a new version. […] The web app replaces our apps that were available in the App Store.”
HTML5, combined with devices essentially just running a browser that becomes the next-generation operating system, so to speak, allows one to bypass the app store environment. It will be interesting how this plays out given that, according to Research2Guidance, in 2012 smartphone users spent US$8 billion for paid apps in the top 5 app platforms.
The race for the dominating platform in mobile devices is on and fascinating to watch. For us in EDA it seems like a constant, multi-segment race to catch those software developers, and HTML5 just added a new segment to the race.
—Frank Schirrmeister is group director for product marketing of the System Development Suite at Cadence.
Leave a Reply