Digital

Surround Computing Is About to Change Our Lives

(Image via Shutterstock)

(Image via Shutterstock)

Computing is in upheaval. After 40 years of microprocessor improvement, an even more fundamental advance is underway, with enormous implications for how we will live.

We are uniting traditional central processing units (CPUs), the “brains” of computers, with graphics processing units (GPUs) to enable much easier information processing with faster performance and better energy efficiency. In fact, the changes in computing chip architecture underway today may be the most significant since the 1970s.

We need this advance. More data will be created in the next three years than in all of human history. The Internet of Things and cloud computing—the combination of which is sometimes called the Internet of Everything—are the next phase of the dramatic information revolution launched with the advent of the IBM PC in 1981 and accelerated further by the World Wide Web starting in 1991.

As recently as the early 1990s it was still something of a novelty for an individual to own a computer. Fast forward to today, when we have many billions of connected computing devices. Many people not only have a PC or laptop—or more than one—but also a smart phone, GPS unit, gaming and entertainment console, and a tablet computer. Smart watches and glasses are emerging. Technology used to be experienced while sitting at a desk. Now, it goes with us everywhere and is increasingly embedded in appliances around us.

Gartner Group predicts that by 2020 there will be 30 billion connected devices. Imagine processing all the information collected from these and figuring out how to make good use of it. That global mission, already underway, will require fundamental advances on many fronts of tech and will result in the distribution of computing intelligence everywhere.

Technology will surround us and blend seamlessly into our environment. It will be discoverable, readable, recognizable, locatable, addressable, and controllable. I call this evolution “the Surround Computing era.” We will be able to interact naturally with technology and it will assist us in innumerable amazing ways that are only beginning to unfold. It will recede into the background, streamlining everyday life and fundamentally changing the way we live, work, learn, play, and relate to one other and to the world.

In the Surround Computing era connected devices will sense and anticipate our needs, often relying on information stored and synthesized in the cloud. Surround Computing will pervade and transform industries including retail, health care, automotive, and many others. Some early examples of this kind of environment already exist in products like Siri, Kinect, Google Glass, and virtual reality devices such as Oculus Rift.

The Surround Computing era is within reach, but still requires a number of fundamental technical advances, beginning with semiconductor architecture. This imminent flood of information will require us to be able to analyze all of the data that is produced and visualize the results to make it useful.

For most of computing history, CPUs have done the heavy work of computing, executing the instructions of computer code serially, or one after the other. That’s fine for many activities like financial accounting. But the demands of modern devices and applications spawned the rise of GPUs that process instructions in parallel, which enabled brilliant graphic displays and much faster computation ability. The kind of activities that benefit from this sort of parallel processing range from high energy physics on supercomputers to medical diagnostic equipment such as MRIs, and more recently to gesture and face recognition.

Simply adding more CPU cores and making their operation more efficient will not keep up with the demands of the Surround Computing era. We must harness the GPU’s parallel computing capabilities to satisfy these new requirements.

For roughly the last decade, CPUs and GPUs have coexisted within PCs, workstations, and, more recently, supercomputers. Nearly 10 percent of the world’s top supercomputers now use a combination of CPUs and GPUs. The GPU, however, has operated largely independent from the CPU inside computers in order to complete various tasks. This has effectively slowed both types of processors and limited the total output of these systems, inhibiting their full capabilities.

Realizing there had to be a better way to deliver more output, we at AMD created the APU, an accelerated processing unit that combines the CPU and GPU onto a single computer chip. It allows computing and visualization to come together at much lower power and cost.

Now we are taking the advantages of the APU a big step further with true heterogeneous system architecture that seamlessly combines the two computing approaches into one holistic design.

With heterogeneous system architecture (HSA), the serial and parallel processors equally share computing resources, including memory. That combination enables faster, more energy-efficient devices and more intuitive applications.

AMD just released the first processor with HSA capability. It delivers more sophisticated technology that better interprets speech, gestures, and facial expressions, and also enables new applications for virtual realism, big data mining, and security.

This HSA innovation, now overseen by the Open HSA Foundation, which comprises many of the largest technology companies in the world, is fundamental to realizing the vision of Surround Computing. In addition to AMD, members of the foundation include ARM, Imagination Technologies, MediaTek, Oracle, Qualcomm, Samsung, Texas Instruments, and more than 40 others. Together, these companies are behind two out of every three processors used in intelligent devices.

Increasingly, intelligent technology will be embedded to enable cars, buildings, roads, and our homes to communicate seamlessly. Our interactions with computing will be tailored in real time, utilizing information we share about our preferences, what we need to buy, entertainment we enjoy, and our location. We will see an inexorable growth of external sensing devices around us such as traffic systems, GPS, remote weather monitoring stations, and intelligent HVAC (heating, ventilation, and air conditioning) that improve both comfort and energy efficiency. Our technology will customize our information depending upon the context of our environment.

Imagine having picked the recipes that you want to prepare for the week and then letting the grocery store tell you what you need to buy—either aisle by aisle as you walk through the store or, more likely, simply delivered directly to your home. For businesses, sales information will be updated in real time and displayed in easy-to-understand 3D images. Such convenience was once the stuff of science fiction, but is rapidly becoming reality.

Look at the new game consoles from Sony and Microsoft based on AMD APUs. Not only do they have beautiful graphics for gaming, but they can become household servers. They aim to be a true portal for the computing, infotainment, and communications needs of a home.

Soon Surround Computing will be everywhere—and nowhere—as it disappears and blends into the environment, becoming a ubiquitous and fundamental capability for every facet of our lives.

Mark Papermaster is senior vice president and chief technology officer for AMD, responsible for AMD’s engineering, research and development (R&D), and product development organization.

Tags: , , , ,

  • fred

    This is a super article that’s well written. Halfway through, I’m wondering “who the hell wrote this!?”. An exciting vision of the near future, with rock solid step-by-step implementation (CPU-GPU-APU-HSA).

    Another key application for HSAs may be the ongoing transition of CAD-BIM-VD&C. Existing software architectures for building vis, “smart” objects and BIM, and serial simulations must give way to massively parallel virtualities (i.e. Mirror Worlds) that can deliver real-time design feedback and forecasting – before decisions are buried.

    Mark, I would love to talk with you about setting up a research project @ CalPoly SLO to pioneer new Mirror World applications for your HSAs.

    • Miriam Cox

      Hi Fred – I am interested in discussing your work. Please contact me at miriam[dot]cox at amd [dot] com.

  • amd sucks

    wow, yeah. i just want intel, intel, intel, intel, inteeeel.

    • Jason

      AMD is a very great brand, There FX 8320 is an awesome processor for 129$ and there 7870s where great for 180$ I got a nice crossfire system for under 900$. Also the Ram they make is freaking awesome and over clocks like a champ. I’m not a AMD fan boy or anything cause I love my laptop that runs an i7 4700mq and 750msli to play those games that don’t work well which is just call of duty ghost and The Secret World and PhysX is sweet. Also its great for mobile gaming. Its sad to see intels top of the line mobile processor match AMDs desktop processor.

    • HyperionZ

      poor child must be 3 years and makes his tantrum

  • http://www.waynecaswell.com Wayne Caswell

    The title and graphic caught my attention, and the article itself caused me to visit the Foundation’s website, but I was not much impressed by AMD’s attempt to offer royalty free IP and open source software to build a camp of developers supporting its APU concept. I saw through the politics and found little substance. Maybe, it’s because the article failed to describe this heterogeneous computing environment with meaningful examples.

    Will the various and disparate processor technologies (CPU, GPU…) come together on a single, tiny, low-power chip for wearables and embedables? I’m not convinced that the same processor architecture can be used for dispersed sensors and for powerful computers and consumer electronics at the same time.

    That challenge seems similar to that of the Smart Home, which has languished in a niche of high-end new homes with professional installation for over 30 years. There’s good reason for that, starting with the fact that manufacturers of different subsystems (PCs, TVs, phones, HVAC, security systems, lights, window treatments, medical sensors, etc.) all have different communication and processing needs. Some have AC power and can use Wi-Fi for high bandwidth and distance, while others must run for months or years on a hearing aid cell battery and rely instead on Bluetooth LE one Zigbee.

    And still others have no need to communicate at all. The Zilog microprocessor in my Philips SonicCare toothbrush, for example, has more than enough power for the very simple code that runs on it. At 10 MIPS, it can execute its simple instructions 10 times faster than the $3.5 million IBM mainframe that I worked on in the mid-1970′s, but Philips couldn’t even find a cheaper processor to embed.

    The bigger challenges I see are (1) scaling processor capacity upward and resolving the crosstalk of tiny circuits that otherwise would constrain Moore’s Law, possibly with quantum computing; and (2) scaling processor size downward to the size of a blood cell but in a way that is not rejected by the human body. I don’t see these designs converging in a single architecture and view efforts to do so as a distraction.