|November 2028 is the 40th anniversary of ESD. Click here read other 2028 lookbacks.|
An interesting thing about predictions is that just about every prediction about mechanical physics has been wrong. If what was predicted by the science fiction writers and futurists of the mid-20th century came to pass, we would be colonizing the moon, commuting in a flying automobile, using nuclear reactors in the home, and zipping to the supermarket with the aid of a jet pack. Almost without exception, the predictions about energy sources, mobility, and speed have not only been wrong, they've been extremely wrong.
If in doubt, think about the world and technology described in an early episode of Star Trek or in the movie 2001: A Space Odyssey –remember when one of the main characters made a phone call from the space station's phone booth? In that version of the future, we've figured out how to travel effortlessly across space, but the mobile phone didn't make it. Clearly, I haven't yet flown beyond the moon (or even made it into outer space), and except for a brief span of a decade or so, we still aren't flying supersonic for commercial flight.
But while these predications of our ability to conquer our physical shortcomings have proved untenable, a few of the predictions that were made about information and our understanding of the world around us have indeed been met and even exceeded.
As we sit here in the year 2028, waiting to catch a glimpse of asteroid 35396 as it hurtles past the earth, we're looking back on some of the changes of the last 20 years that have enabled huge advances in information access, connectivity, and utilization. Don't worry–I won't miss the asteroid–my telescope will alert me when it's time to look. Hopefully though, it will miss us.
Twenty years ago in 2008, we were in the middle of an information explosion. In front of our computer terminals, we had access to huge amounts of data on every conceivable topic–just a mouse click away.
Today, virtually ALL of the world's accumulated knowledge is available on-demand. Historical documents including copies of manuscripts transcribed by Franciscan monks in the 13th century; personal medical data including all of the x-rays you've ever had taken; the contents of every single book in the New York Public Library; even life records of the world experienced around ourselves for the fortunate few outfitted with 360-degree video capture. What's changed is that I no longer need to access it with my fingers. With just a few words, I can find out just about anything and interact with it on the flexible display embedded in the sleeve of my rain jacket.
Some of the most compelling advances in the past 20 years have been those in the interfaces between humans and devices. It's not just that we don't have to use a keyboard anymore. (No one misses QUERTY!) The interfaces we have today are more intuitive and efficient, with minimal movement (voice, pointing) to control a huge amount of complexity. It's the interfaces between the digital world and the physical world that make embedded technology meaningful. And these interfaces are one of the key reasons MIPS Technologies acquired analog/mixed-signal IP leader Chipidea.
Oh, excuse me a moment, my personal medical monitoring device is telling me that my blood sugar is low. I'm just going to grab a quick bite to eat. Today's mainstream refrigerator not only closely monitors its own inventory and communicates with the grocery warehouse to keep itself fully stocked, but it monitors the nutritional content of the food I eat to make sure I am keeping to my 2,369 calorie a day diet. It's quite a feat to predict exact dietary requirements factoring in the temperatures I've experienced throughout the day.
What has enabled this complete view of the systems in our lives–from medical monitoring, diagnosis, and advice, to our financial health, to the smart devices that control everything in our connected homes–are, of course, dramatic advances in embedded systems. We've moved from 1988 when a desktop computer was single- to ten-digit MHz, with single-digit megabytes of memory; to 2008 when frequency, capacity and performance each increased by about 1,000 times; to today where those numbers have reached heights that are virtually inconceivable.
Today's embedded systems are prevalent and seamless. They are networked together, constantly accessing and feeding information back to each other. The programmable macro objects of yesterday have given way to softer embedded systems with softer boundaries between the system and the network.
Did I mention that I am driving while I am writing this? My sustainable vehicle is rocketing down the freeway while making sure to keep a comfortable distance from other cars. What? We're there already? That's OK, my car will drop me off and then park itself. Twenty years ago, we were already seeing the results of the first efforts of these types of smart vehicles with SoCs like the EyeQ2 from Mobileye for real-time visual recognition and scene interpretation in intelligent vehicle systems.
Driving the performance of all of these systems has been a great progression in processor technology. Huge advances in raw processing performance have been accompanied by dramatic advances in power and programming efficiency. The embedded designer of 2028 is working with highly abstracted systems and software. Gone are the imperative procedural languages of the past. Today's designer works with a rules-based, natural language description process, describing what he wants the system to do and allowing the system to ask for clarification when appropriate or inconsistent with the description of what is desired. The process of programming algorithms in a procedural language seems quaint, but quite painful as described by those unfortunate enough to have had to do it.
And while the materials that our processing engines are built on have continued to evolve, what made a great difference is that prior to rules-based programming, we offered configurable programmable functionality in RTL that could be implemented as hardware or rules on the processing engines.
Today's rules-based systems are vital for complex interoperable environments where people can't begin to contemplate and define all of the behaviors needed. With such a high abstraction layer, people don't know or care what is going on underneath. Today's systems go a long way toward controlling themselves.
The systems' IP vendor provides the flexible high-performance, low-power engine that enables the vast amount of computation in today's systems. The challenge is making all of the technology affordable and accessible enough so that the world's eight billion residents can take advantage of it in the next generation of embedded devices-devices to which they can teach/suggest rules, and that are incorporated into our daily lives. Who knows what is still to come?
John Derrick is chief operating officer of MIPS Technologies, with responsibility for both the Analog and Processor Business Groups. He joined the company in January 2008 as president and general manager of the company's Processor Business Group.