|November 2028 is the 40th anniversary of ESD. Click here read other 2028 lookbacks.|
Back in 2008, I had many discussions with some of our leading technologists at TI about what technology would be like in 20 years. As we approach the year 2028, I'm amazed at how clear our vision was. We correctly anticipated both the market drivers that spawned a new era of amazing products and the technological developments that made it possible. In a 2008 paper, we predicted that four main market drivers–robotics, green engineering, full immersion, and healthcare–would present insurmountable challenges to developers until advances in IC technology (price, performance, power, integration, development environments) made it possible to break down development barriers that had plagued embedded systems designers for decades. Well, we've accomplished all that and have seen these advances lead to new ways for designers to differentiate their products. The explosion of creativity and prosperity thanks to embedded systems has greatly benefited humanity.
In 2008, TI predicted that there would be four market drivers which would drive the embedded systems industry:
- Robotics systems exist to do things we don't want to do or can't do and to serve as interfaces between the human and machine. The key technologies necessary for pervasive robotics included developing natural interfaces based on voice and video I/O. Not only now in 2028 do vending machines understand what I say, they visually recognize me, remember what I purchase, and look out for my health. Best of all, they deliver! Of course, they dispense only the treats my doctor prescribes. I have to find an off-network machine to give me those circa 2008 Twinkies I love so much. (They just get better with age.) I hope none of the newer vending machines see me do this–they tend to tell on me.
- We predicted that green electronics would be big and that it would focus mainly on harvesting and conserving energy, a huge concern back in 2008 (remember Global Warming). Today our devices harvest energy on macro and micro levels: from renewable sources (macro)–in other words, the sources formerly known as “alternative”–as well as scavenging power from their environment (micro). These devices also conserve energy: compared with 2008, even though each of us have more electronic devices per square meter, we now use less energy in daily living (macro) and have ultra-low-power devices that are so power efficient (micro), they literally “put more green in our pockets.”
- In 2008, we were just beginning to see the promise of full immersion systems to transform entertainment and communications from a passive to truly interactive experience. By then we had achieved higher resolution and greater color depth but were ready for the next technological leap that would take us from simply looking through a plate glass window to engaging all seven of our senses (the other two senses weren't discovered until 2014 when magnetic sensing began to manifest itself) to make these experiences real. Since these systems give us the illusion that we are really someplace else, we've seen, as we suspected, the rise in home-based “staycations” and changes in how we travel in general.
- Health innovations have focused on personal health and now use technology to do three things: manage chronic diseases, predict catastrophic illnesses, and make the last months of life comfortable at home. Healthcare has become more personal, redefining the concept of a doctor's visit through videoconferencing. House-based monitoring systems now track our blood pressure throughout the day, noticing trends and scheduling doctor's appointments on our behalf. One major triumph is that cancer is now much more preventable, detectable, and treatable thanks to implantable sensor systems. When I took part in this study 20 years ago I was nearing retirement. Now am I not only not retired, but I'm looking at options for my next career. This, I owe to the new life expectancy of 120 years. Our longer “senior years” are more productive and enjoyable thanks our robotic healthcare helpers and prostheses.
Improving the state-of-the-art IC technology made it all possible.
- Price: Moore's Law stated that we could double the number of transistors we could integrate every two to three years. Its only purpose, to give us lower cost and any correlation to performance or power, no longer applies. In 2008 there was even talk of slowing integration rates because we're having trouble taking advantage of the capability we had. But even though we can now put hundreds of billions of transistors on a device and the cost of these devices are in the range of $1 per billion transistors, we now are comfortable with how to take full advantage of the technology.
- Performance: Increasing performance through multiprocessing is a concept that has been around for 40 years. In 2008, we predicted we'd see SoC devices that integrate sophisticated “heterogeneous multiprocessor” architectures incorporating many diverse processing blocks together, to relatively simple “homogeneous multicore” devices grouping cores similar in nature. Today, processors look like microsocieties with each core performing its own assigned task, perhaps unaware of the overall application. And, just as we predicted, devices have thousands of these blocks integrated into those microsocieties.
- Power: We achieved ultra-low power by moving to transistors that run at lower voltages, operate at lower frequencies, and power down when not in use. We now have predictive circuitry, which turns on circuits only when they're needed. Power scavenging was also a major breakthrough. Many embedded applications arose as we became comfortable with the concept of the perpetual device–a device that scavenges energy from its environment. With the perpetual device, we were able to implant medical devices with a lifetime source of energy and stress sensors buried in bridge support columns.
- Integration: We'd still be stuck with clunky 2008 products if we weren't able to simplify system-integration roadmaps with three aspects. The first step was to create the system at a board level with multiple devices on a printed circuit board–chips-on-board (COB) architecture. Secondly, we put the system in one package using stacked die or multichip packaging–a system-in-package (SIP). The ultimate integration was to put the system on one substrate–system-on-chip (SOC). But this SOC quickly became a “sub-system on a chip” and was integrated in as part of a larger system-in-package (SIP). Integration, it turned out, was solved not by SOC, but rather by the SIP.
- Development environments: Year 2028 processors have hundreds to thousands of cores, something almost unimaginable back in 2008 when the complexity of multiprocessing just four cores frustrated designers. Over the years, developers demanded tools to design multiprocessor systems without having to explicitly program each core. As we reluctantly predicted, we did achieve design nirvana whereby we have significantly reduced the design complexity of a sophisticated system. It is comparable to reducing the complexity of a multidimensional crossword puzzle to that of two-dimensional TicTacToe. Developers now rely on their development environments to manage all their lower-level expectations, including component configuration, integration, interfacing, and management. Development tools also determine what is efficient to execute in parallel and how to partition it.
For the first few decades of the embedded systems industry, one could predict innovation in the electronics industry by trying to figure out how to fit larger systems into our pockets through integration. For example, the computer moved from a large room, to a desk top, to our pocket. The fun of this approach was the next step: how to embed these same systems in our clothing or bodies so we don't even have to think about them. As we look back at the evolution of electronics, we can argue that it wasn't higher performance that brought computers to our pockets, it was reduced power dissipation. When we reduce power consumption by half, we reduced battery size by half as well. What happened when we reduced power consumption by orders of magnitude? Coupled with a billion transistors per $1, we created perpetual devices that we embed anywhere we can imagine. This exciting innovation is still the key to our future.
The one thing we didn't anticipate back in 2008 or ever, was how collectible TI's original Speak'N'Spell became. I'm still surprised that a vintage 1978 Speak'N'Spell recently went for $14 billion dollars at last year's worldwide Soth-E-bay's auction. Wow. It's a good thing we put an S'N'S in that TI time capsule. Now if only we could remember where we buried it (yes, eliminating memory loss is the next breakthrough in technology–I hope).
Gene Frantz is one of the world's foremost experts in digital signal processing. In 2002, he was named a TI Principal Fellow, joining an elite group of technology innovators at the top of TI's technical ladder. As DSP business development manager, Frantz is responsible for creating new businesses within TI utilizing DSP technology.