Smartphones, cloud computing, and the Internet of Things will dramatically increase our quality of life, according to veteran EDA investor Jim Hogan (right). But speaking at a Cadence-sponsored lunch at DVCon on March 4, 2014, Hogan also noted that these new technologies are causing "abundant chaos" in functional verification.
Abundant chaos is not totally a bad thing. It also opens new opportunities for better tools and methodologies. And it may also result in "exponential" growth for DVCon itself, "because the problems we have to solve are so huge," Hogan said.
Hogan titled his talk "Fracturing of the Value Chain and Re-Aggregation." He identified five technology trends that will impact our lives in the near future, as described below. He then pointed to some of the resulting verification challenges.
Trend #1—Smartphone penetration and types of use have only begun.
Smartphone penetration is already pretty amazing, according to reports cited by Hogan. According to one estimate, there were 1.5 billion smartphones in 2013 and over 5 billion mobile phones. In a few years, he said, almost everyone on the planet will have access to all the information that people in the developed world have today, and then some.
I was quite surprised by a chart that showed smartphone penetration rates for various countries around the world. The U.S. is pretty far down the list, with 56.4% penetration. At the top are United Arab Emirates (73.8%), South Korea (73%), and Saudi Arabia (72.8%). Another interesting data point is that the bill of materials cost for an iPhone is about $200, and the cost for a Samsung Galaxy 4 is about $241. The people who are making most of the money, Hogan observed, are the carriers.
Trend #2—The Internet of Things is going to be massively diverse and driven by 32-bit processors.
According to Hogan, Internet of Things devices will share these characteristics:
- Integrated sensors
- Very energy efficient
- Wired and wireless
- 32-bit functionality and performance
- Embedded memory
- Cheap "like dirt"
- Huge security issues
Health monitoring will be a major application. Hogan cited one example—the biostamp (left) from MC10. This is a wearable, remote health monitor that can be used for pre-natal or post-natal care. It also comes in handy for Hogan's elderly uncle in Texas, who sometimes forgets to turn on his thermostat at night.
The Internet of Things will not be driven by 64-bit processors, Hogan said. Instead, cheap 32-bit processors will be the "ubiquitous workhorse." They will cost about 7 cents each. Compare that to the cost of the processor in an Apple or Samsung smartphone, which is around $20 each. And to keep costs down, the Internet of Things devices will use a lot of open-source software as well.
Trend #3—Software application behavior drives hardware architectures.
A system on chip (SoC) is not something you synthesize at a high level, Hogan said, it is an aggregation of IP. How does the architecture know what to do? That depends on the behavior that the software dictates through the drivers.
"EDA touches software on the emulator today," Hogan said. "Over time, that interface between the emulation architecture and the behavior of the software is going to be where the action is, and it will make up half of the poster sessions at this [DVCon] conference."
Software defines the user experience, and automobiles are a great example, Hogan said. While he doesn't yet personally own a Tesla, he expressed admiration for My Tesla, a software-defined instrument panel that lets customers create their own environments.
Trend #4—The software applications will run on open-source components. Every system will have its own somewhat unique software stack and corresponding hardware.
Hogan's explanation: "Get used to it. Open source is the way it's going to go."
Trend #5—Cloud applications will be the source of most of the value for Internet of Things and mobile devices. As a result, they are likely to become the economic winners, with growth in sales and profits over the next ten years.
Hogan showed a chart that predicted that cloud computing revenues will grow from $40.7 billion in 2011 to $118.7 billion in 2014 and $241 billion in 2020. Another chart showed high compound annual growth rates for cloud infrastructure services, topping off at 41.7% for infrastructure as a service (IaaS). Yet another showed compound annual growth rates for a number of public cloud services, reaching up to 49.1% for office suites.
"This is going to be an enormous business and it will create a lot of value. Hopefully it will need a lot of semiconductors to do that," Hogan noted.
How will we verify all this?
While the value is in the system, Hogan said, the challenge is verification. "It's a tremendous problem. There are discontinuities everywhere. So, this is an opportunity."
Here are some of the challenges Hogan noted:
- Multi-core software development and hardware/software verification
- Complex interconnect protocols, with competing standards
- Complex low-power design features
- SoC software and IP integration with 10s to 100s of IP blocks
- Analog/mixed signal verification
And extending up to the system level, these questions emerge:
- How do we bring up software stacks and use them for verification?
- How do we verify the chain from sensor to hub to cloud?
- How do we represent the SoC environment?
A concluding remark:
"EDA might not have "E" in front of it in a few years. It may be System Design Automation or something. Software is going to be the thing that drives all this."
Related Blog Posts
Lip-Bu Tan at DVCon 2014: EDA/Silicon Ecosystem Crucial to Innovation
25th Anniversary: Hogan on EDA History and Three Little Words