Intel's BIG science

Tim Dean | Nov 6, 2008 12:42 PM

Tim Dean reports on the crazy big science Intel's R&D arm is working on.

Intel doesn’t muck around when it comes to research. In fact, Intel as a company spends more on pure research than does Australia as a nation. We’re talking $US6 billion devoted to exploring the far reaches of technological capability, and not all of it focussing on processor technology. And on day two of the Fall IDF in Taiwan, recently, the gathered technophiles were treated to a sneak peek of some of the wild and wondrous creations that lie just over the horizon.

The scene was set by some precipitous forecasting from none other than technology’s Chief Hyperbolist, Ray Kurzweil. When it comes to the future, few have laid down more concrete – or astounding – predictions than Kurzweil. Not only does he see technological advancement as possessing some kind of inevitability, but he suggests it’s only a matter of time before the exponential increase in processing power leads us to developing machines that are more intelligent than us. (Then again, I’ve watch enough trash TV to think many of us may have overrated human intelligence in the first place...) When this day comes, we’ll have reached a "profoundly transformative" moment that Kurzweil calls – with his usual knack for understatement – the 'singularity', because what happens beyond this point is virtually unpredictable.

The question that Intel is focussing on is not when this day will come (or why we might want it to come at all), but how to get there. Intel Senior Fellow and research guru, Kevin Kahn, laid out a number of 'gaps' to be overcome between computers and people. They were:

• Computing - referring to raw processing power.
• Sensing - which is the ability for machines to observe what’s going on outside their little beige boxes.
• Signalling - or how they communicate to each other and to us.
• Power - which is a big problem when you think that Google now builds its data centres near power stations.
• Actuation - which is making them actually move and interact with the world.

On the first problem, Kahn revealed some intriguing notions about manufacturing future processors in an entirely different way to today. Instead of building them 'top down', as they do now, by etching away unwanted material to make your processor, Kahn talked of building processors from the 'bottom up'. This means actually assembling them atom by atom – or true 'nanotechnology'. According to Kahn, Intel already lays down its high-K dielectrics in this way, so it’s not just pie in the sky stuff, and future Intel processors could be manufactured in this way.

Kahn also mentioned emerging processor technologies that move away from simple on/off voltages of today. One approach is to go 'non-Boolean' by using multiple states, such as spin state or spin direction, to encode more than one bit of information at a time. We could even use carbon nanotubes in place of silicon as the basis for future processors. Intel is exploring all these ideas, although Kahn suggests it will be at least a decade before they see the light of day.

On the signalling front, Intel is very interested in silicon photonics. The big benefit of photons for communications is they don’t generate heat like electrons do. But the trick is to create them with the right properties, which is where Intel’s silicon lasers come into the picture. Intel has already developed a hybrid laser built into silicon, and the next generation in development is a laser that is more easily integrated into silicon chips, making them easier to manufacture. The aim is for a terabit photonic transceiver that can be built alongside a conventional silicon microprocessor. Intel has already built a prototype 200Gb/s silicon photonics test chip, so progress is rolling along nicely.

Wireless communication is also being looked at, specifically at how to squeeze the most from the radio spectrum. However, one of the most impressive demonstrations of the whole IDF was not to do with wireless communication, but wireless power. Harkening back to the days of inventor Nikola Tesla, Kahn demonstrated a device that can beam energy over the air to power a remote device. The demo unit, called a wireless resonant energy link, looked like something from a 1920s sci-fi flick, with rings of copper about half a metre in diameter sitting about a metre from a smaller collection of rings connected to a light bulb. And you guessed it, switch on the main ring, and the bulb lights up. It works by resonant magnetic fields which are picked up by the receiver coils and turned into a current. The device is entirely safe, and you can stick your arm between the coils without any undue affects. According to Kahn, such a technology could be used to line the walls of buildings or cubicles in an office and then used to power small devices remotely. Wiggy stuff.

Another cute demo dealt with helping robots to interact with humans, presumably before they take over the world. The demo consisted of a robot arm developed in Intel’s Seattle research lab. This arm had a special sensor integrated into its hand, called electric field pretouch. It functions in a similar way to the electro sense possessed by some fish, which allows them to detect the presence of other objects nearby. For the robot, the sensor sits in between vision and touch and enables the arm to detect an object, follow it and grasp it cleanly.

Possibly the strangest demo of the conference was of a device called the ultrasonic phased array. This small tablet shaped device is kind of like a prototyping machine, which creates physical objects from a 3D computer rendering. Except instead of assembling a real object, this device uses ultrasonics to create the feeling of an object. Just input the shape you want and you can wave your hand over the tablet and the ultrasonics will give you the impression you can actually feel the object. Such a device may one day be useful for remote sensing where a human operator can get a real hands on feel of what’s going on somewhere else.

The final demo comes straight from science-fiction: programmable matter made from self assembling 3D nano robots. I’ve seen some early Intel research on this front from previous Intel Research Days, and I can guarantee you Intel is definitely keen on this technology. The idea is to have millions of tiny robots that interact with each other to create full objects that can be held, manipulated and even modified on the fly. This could be for prototyping, as mentioned above, or to produce a representation of a remote object. Intel has already played around with getting the concept to work by building larger devices, a few centimetres across, called catoms, that interact with each other using magnets in 2D. Smaller catoms have also been made that use electrostatics, including the latest work, which uses catoms 500 microns in diameter. Definitely a technology to watch.

Kevin Kahn, and Intel's crazy science research.

It’s easy to think of Intel as just a semiconductor manufacturer, but it’s forays into some wild and wondrous areas of speculative research are truly impressive. Whether they’ll lead us to the maw of the singularity is anyone’s guess. But there’s no question that if even a few of these technologies demoed at IDF Taiwan reach production, they have the potential to have a huge impact on our lives.