Building Brain-Inspired AI Supercomputing System

IBM (NYSE: IBM) and the U.S. Air Force Research Laboratory (AFRL) today announced they are collaborating on a first-of-a-kind brain-inspired supercomputing system powered by a 64-chip array of the IBM TrueNorth Neurosynaptic System. The scalable platform IBM is building for AFRL will feature an end-to-end software ecosystem designed to enable deep neural-network learning and information discovery. The system’s advanced pattern recognition and sensory processing power will be the equivalent of 64 million neurons and 16 billion synapses, while the processor component will consume the energy equivalent of a dim light bulb – a mere 10 watts to power.
IBM researchers believe the brain-inspired, neural network design of TrueNorth will be far more efficient for pattern recognition and integrated sensory processing than systems powered by conventional chips. AFRL is investigating applications of the system in embedded, mobile, autonomous settings where, today, size, weight and power (SWaP) are key limiting factors. The IBM TrueNorth Neurosynaptic System can efficiently convert data (such as images, video, audio and text) from multiple, distributed sensors into symbols in real time. AFRL will combine this “right-brain perception capability of the system with the “left-brain” symbol processing capabilities of conventional computer systems. The large scale of the system will enable both “data parallelism” where multiple data sources can be run in parallel against the same neural network and “model parallelism” where independent neural networks form an ensemble that can be run in parallel on the same data.


AFRL was the earliest adopter of TrueNorth for converting data into decisions,” said Daniel S. Goddard, director, information directorate, U.S. Air Force Research Lab. “The new neurosynaptic system will be used to enable new computing capabilities important to AFRL’s mission to explore, prototype and demonstrate high-impact, game-changing technologies that enable the Air Force and the nation to maintain its superior technical advantage.”

“The evolution of the IBM TrueNorth Neurosynaptic System is a solid proof point in our quest to lead the industry in AI hardware innovation,” said Dharmendra S. Modha, IBM Fellow, chief scientist, brain-inspired computing, IBM Research – Almaden. “Over the last six years, IBM has expanded the number of neurons per system from 256 to more than 64 million – an 800 percent annual increase over six years.’’


30 Billion Switches Onto The New IBM Nano-based Chip

IBM is clearly not buying into the idea that Moore’s Law is dead after it unveiled a tiny new transistor that could revolutionise the design, and size, of future devices. Along with Samsung and Globalfoundries, the tech firm has created a ‘breakthrough’ semiconducting unit made using stacks of nanosheets. The companies say they intend to use the transistors on new five nanometer (nm) chips that feature 30 billion switches on an area the size of a fingernail. When fully developed, the new chip will help with artificial intelligence, the Internet of Things, and cloud computing.

For business and society to meet the demands of cognitive and cloud computing in the coming years, advancement in semiconductor technology is essential,” said Arvind Krishna, senior vice president, Hybrid Cloud, and director, IBM Research.

IBM has been developing nanometer sheets for the past 10 years and combined stacks of these tiny sheets using a process called Extreme Ultraviolet (EUV) lithography to build the structure of the transistor.

Using EUV lithography, the width of the nanosheets can be adjusted continuously, all within a single manufacturing process or chip design,” IBM and the other firms said. This allows the transistors to be adjusted for the specific circuits they are to be used in.


Nano-LED 1000 Times More Efficient

The electronic data connections within and between microchips are increasingly becoming a bottleneck in the exponential growth of data traffic worldwide. Optical connections are the obvious successors but optical data transmission requires an adequate nanoscale light source, and this has been lacking. Scientists at Eindhoven University of Technology (TU/e) now have created a light source that has the right characteristics: a nano-LED that is 1000 times more efficient than its predecessors, and is capable of handling gigabits per second data speeds.

NANO LEDWith electrical cables reaching their limits, optical connections like fiberglass are increasingly becoming the standard for data traffic. Over longer distances almost all data transmission is optical. Within computer systems and microchips, too, the growth of data traffic is exponential, but that traffic is still electronic, and this is increasingly becoming a bottleneck. Since these connections (‘interconnects’) account for the majority of the energy consumed by chips, many scientists around the world are working on enabling optical (photonic) interconnects. Crucial to this is the light source that converts the data into light signals which must be small enough to fit into the microscopic structures of microchips. At the same time, the output capacity and efficiency have to be good. Especially the efficiency is a challenge, as small light sources, powered by nano– or microwatts, have always performed very inefficiently to date.
The researchers in Eindhoven believe that their nano-LED is a viable solution that will take the brake off the growth of data traffic on chips. However, they are cautious about the prospects. The development is not yet at the stage where it can be exploited by the industry and the production technology that is needed still has to get off the ground.
The findings are reported in the online journal Nature Communications.


How To Erase Chips Remotely

A military drone flying on a reconnaissance mission is captured behind enemy lines, setting into motion a team of engineers who need to remotely delete sensitive information carried on the drone’s chips. Because the chips are optical and not electronic, the engineers can now simply flash a beam of UV light onto the chip to instantly erase all content. Disaster averted.

This James Bond-esque chip is closer to reality because of a new development in a nanomaterial developed by Yuebing Zheng, a professor of mechanical engineering and materials science and engineering in the Cockrell School of Engineering. His team described its findings in the journal Nano Letters.


The molecules in this material are very sensitive to light, so we can use a UV light or specific light wavelengths to erase or create optical components,” Zheng said. “Potentially, we could incorporate this LED into the chip and erase its contents wirelessly. We could even time it to disappear after a certain period of time.”

To test their innovation, the researchers used a green laser to develop a waveguide — a structure or tunnel that guides light waves from one point to another — on their nanomaterial. They then erased the waveguide with a UV light, and re-wrote it on the same material using the green laser. The researchers believe they are the first to rewrite a waveguide, which is a crucial photonic component and a building block for integrated circuits, using an all-optical technique.


Electronics: How To Dissipate Heat in A Nanocomputer

Controlling the flow of heat through semiconductor materials is an important challenge in developing smaller and faster computer chips, high-performance solar panels, and better lasers and biomedical devices. For the first time, an international team of scientists led by a researcher at the University of California, Riverside has modified the energy spectrum of acoustic phononselemental excitations, also referred to as quasi-particles, that spread heat through crystalline materials like a wave—by confining them to nanometer-scale semiconductor structures. The results have important implications in the thermal management of electronic devices. Led by Alexander Balandin, Professor of Electrical and Computing Engineering and UC Presidential Chair Professor in UCR’s Bourns College of Engineering, the research is described in a paper published in the journal Nature Communications.


The team used semiconductor nanowires from Gallium Arsenide (GaAs), synthesized by researchers in Finland, and an imaging technique called Brillouin-Mandelstam light scattering spectroscopy (BMS) to study the movement of phonons through the crystalline nanostructures. By changing the size and the shape of the GaAs nanostructures, the researchers were able to alter the energy spectrum, or dispersion, of acoustic phonons. The BMS instrument used for this study was built at UCR’s Phonon Optimized Engineered Materials (POEM) Center, which is directed by Balandin.

Controlling phonon dispersion is crucial for improving heat removal from nanoscale electronic devices, which has become the major roadblock in allowing engineers to continue to reduce their size. It can also be used to improve the efficiency of thermoelectric energy generation, Balandin said. In that case, decreasing thermal conductivity by phonons is beneficial for thermoelectric devices that generate energy by applying a temperature gradient to semiconductors.

For years, the only envisioned method of changing the thermal conductivity of nanostructures was via acoustic phonon scattering with nanostructure boundaries and interfaces. We demonstrated experimentally that by spatially confining acoustic phonons in nanowires one can change their velocity, and the way they interact with electrons, magnons, and how they carry heat. Our work creates new opportunities for tuning thermal and electronic properties of semiconductor materials,” Balandin said.


Nanocomputer: Carbon Nanotube Transistors Outperform Silicon

For decades, scientists have tried to harness the unique properties of carbon nanotubes to create high-performance electronics that are faster or consume less power — resulting in longer battery life, faster wireless communication and faster processing speeds for devices like smartphones and laptops. But a number of challenges have impeded the development of high-performance transistors made of carbon nanotubes, tiny cylinders made of carbon just one atom thick. Consequently, their performance has lagged far behind semiconductors such as silicon and gallium arsenide used in computer chips and personal electronics.

Now, for the first time, University of Wisconsin–Madison materials engineers have created carbon nanotube transistors that outperform state-of-the-art silicon transistors. Led by Michael Arnold and Padma Gopalan, UW–Madison professors of materials science and engineering, the team’s carbon nanotube transistors achieved current that’s 1.9 times higher than silicon transistors. The researchers reported their advance in a paper published in the journal Science Advances.

carbon nanotube integrated circuits

This achievement has been a dream of nanotechnology for the last 20 years,” says Arnold. “Making carbon nanotube transistors that are better than silicon transistors is a big milestone. This breakthrough in carbon nanotube transistor performance is a critical advance toward exploiting carbon nanotubes in logic, high-speed communications, and other semiconductor electronics technologies.”

This advance could pave the way for carbon nanotube transistors to replace silicon transistors and continue delivering the performance gains the computer industry relies on and that consumers demand. The new transistors are particularly promising for wireless communications technologies that require a lot of current flowing across a relatively small area.


Tiny Diamonds Revolutionize Nanotechnology

Nanomaterials have the potential to improve many next-generation technologies. They promise to speed up computer chips, increase the resolution of medical imaging devices and make electronics more energy efficient. But imbuing nanomaterials with the right properties can be time consuming and costly. A new, quick and inexpensive method for constructing diamond-based hybrid nanomaterials in bulk could launch the field from research to applications. University of Maryland (UMD) researchers developed a method to build diamond-based hybrid nanoparticles in large quantities from the ground up, thereby circumventing many of the problems with current methods.

The process begins with tiny, nanoscale diamonds that contain a specific type of impurity: a single nitrogen atom where a carbon atom should be, with an empty space right next to it, resulting from a second missing carbon atom. This “nitrogen vacancyimpurity gives each diamond special optical and electromagnetic properties. By attaching other materials to the diamond grains, such as metal particles or semiconducting materials known as “quantum dots,” the researchers can create a variety of customizable hybrid nanoparticles, including nanoscale semiconductors and magnets with precisely tailored properties.


If you pair one of these diamonds with silver or gold nanoparticles, the metal can enhance the nanodiamond’s optical properties. If you couple the nanodiamond to a semiconducting quantum dot, the hybrid particle can transfer energy more efficiently,” said Min Ouyang, an associate professor of physics at UMD and senior author on the study.

The technique is described in the June 8 issue of the journal Nature Communications.


Electronic Circuits Mimic The Human Brain

Researchers of the MESA+ Institute for Nanotechnology and the CTIT Institute for ICT Research at the University of Twente in The Netherlands have demonstrated working electronic circuits that have been produced in a radically new way, using methods that resemble Darwinian evolution. The size of these circuits is comparable to the size of their conventional counterparts, but they are much closer to natural networks like the human brain. The findings promise a new generation of powerful, energy-efficient electronics, and have been published in the journal Nature Nanotechnology. The approach of the researchers at the University of Twente is based on methods that resemble those found in Nature. They have used networks of gold nanoparticles for the execution of essential computational tasks. Contrary to conventional electronics, they have moved away from designed circuits. By using ‘designless‘ systems, costly design mistakes are avoided. The computational power of their networks is enabled by applying artificial evolution. This evolution takes less than an hour, rather than millions of years. By applying electrical signals, one and the same network can be configured into 16 different logical gates. The evolutionary approach works around – or can even take advantage of – possible material defects that can be fatal in conventional electronics.

One of the greatest successes of the 20th century has been the development of digital computers. During the last decades these computers have become more and more powerful by integrating ever smaller components on silicon chips. However, it is becoming increasingly hard and extremely expensive to continue this miniaturisation. Current transistors consist of only a handful of atoms. It is a major challenge to produce chips in which the millions of transistors have the same characteristics, and thus to make the chips operate properly. Another drawback is that their energy consumption is reaching unacceptable levels. It is obvious that one has to look for alternative directions, and it is interesting to see what we can learn from nature. Natural evolution has led to powerful ‘computers’ like the human brain, which can solve complex problems in an energy-efficient way. Nature exploits complex networks that can execute many tasks in parallel.


Super-Efficient Light-Based Nanocomputers

Stanford electrical engineer Jelena Vuckovic wants to make computers faster and more efficient by reinventing how they send data back and forth between chips, where the work is done.

In computers today, data is pushed through wires as a stream of electrons. That takes a lot of power, which helps explain why laptops get so warm.

Several years ago, my colleague David Miller carefully analyzed power consumption in computers, and the results were striking,” said Vuckovic, referring to David Miller, the W.M. Keck Foundation Professor of Electrical Engineering. “Up to 80 percent of the microprocessor power is consumed by sending data over the wires – so-called interconnects.”

In a Nature Photonics article whose lead author is Stanford graduate student Alexander Piggott, Vuckovic, a professor of electrical engineering, and her team explain a process that could revolutionize computing by making it practical to use light instead of electricity to carry data inside computers.

infrared lightIn essence, the Stanford engineers want to miniaturize the proven technology of the Internet, which moves data by beaming photons of light through fiber optic threads

Optical transport uses far less energy than sending electrons through wires,” Piggott said. “For chip-scale links, light can carry more than 20 times as much data.”

Theoretically, this is doable because silicon is transparent to infrared light – the way glass is transparent to visible light. So wires could be replaced by optical interconnects: silicon structures designed to carry infrared light.

One-Atom-Thin Silicon Transistors For NanoComputer

Researchers at The University of Texas at Austin‘s Cockrell School of Engineering have created the first transistors made of silicene, the world’s thinnest silicon material. Their research holds the promise of building dramatically faster, smaller and more efficient computer chips.

Made of a one-atom-thick layer of silicon atoms, silicene has outstanding electrical properties but has until now proved difficult to produce and work with. Deji Akinwande, an assistant professor in the Cockrell School’s Department of Electrical and Computer Engineering, and his team, including lead researcher Li Tao, solved one of the major challenges surrounding silicene by demonstrating that it can be made into transistors —semiconductor devices used to amplify and switch electronic signals and electrical power.

The first-of-their-kind devices developed by Akinwande and his team rely on the thinnest of any semiconductor material, a long-standing dream of the chip industry, and could pave the way for future generations of faster, energy-efficient computer chips.

Buckled honeycomb lattice structure of silicene

Apart from introducing a new player in the playground of 2-D materials, silicene, with its close chemical affinity to silicon, suggests an opportunity in the road map of the semiconductor industry,” Akinwande said. “The major breakthrough here is the efficient low-temperature manufacturing and fabrication of silicene devices for the first time.”

Despite its promise for commercial adaptation, silicene has proved extremely difficult to create and work with because of its complexity and instability when exposed to air. To work around these issues, Akinwande teamed with Alessandro Molle at the Institute for Microelectronics and Microsystems in Agrate Brianza, Italy, to develop a new method for fabricating the silicene that reduces its exposure to air. In the near-term, Akinwande will continue to investigate new structures and methods for creating silicene, which may lead to low-energy, high-speed digital computer chips.

The research work has been published in the journal Nature Nanotechnology.


How to “Grow” Billions Of Light Dots Directly On Chips

Researchers from the University of California, Santa Barbara (UCSB), in collaboration with the DARPA, succeeded to grow lasers directly on microchips, a breaktrhrough that will enable the mass-production of inexpensive and robust microsystems that exceed the performance capabilities of current technologies.

Defense systems for instance, such as radar, communications, imaging and sensing payloads rely on a wide variety of microsystems devices. These diverse devices typically require particular substrates or base materials and different processing technologies specific to each application, preventing the integration of such devices into a single fabrication process. Integration of these technologies, historically, has required combining one microchip with another, which introduces significant bandwidth and latency limitations as compared to microsystems integrated on a single chip. Although many photonic components can now be fabricated directly on silicon, realizing an efficient laser source on silicon has proven to be very difficult.
Now, the engineers at UCSB showed it was possible to “grow” or deposit successive layers of indium arsenide material directly on silicon wafers to form billions of light-emitting dots known as “quantum dots.” This method of integrating electronic and photonic circuits on a common silicon substrate promises to eliminate wafer bonding, and has application in numerous military and civilian electronics where size, weight, power and packaging/assembly costs are critical.
laser on chipsDARPA’s Electronic-Photonic Heterogeneous Integration (E-PHI) program has successfully integrated billions of light-emitting dots on silicon to create an efficient silicon-based laser. The Defense Advanced Research Projects Agency (DARPA) is an agency of the United States Department of Defense responsible for the development of new technologies for use by the military.
This method of integrating electronic and photonic circuits on a common silicon substrate promises to eliminate wafer bonding, and has application in numerous military and civilian electronics where size, weight, power and packaging/assembly costs are critical“.“It is anticipated that these E-PHI demonstrator microsystems will provide considerable performance improvement and size reduction versus state-of-the-art technologies,” said Josh Conway, DARPA program manager for E-PHI. “Not only can lasers be easily integrated onto silicon, but other components can as well, paving the way for advanced photonic integrated circuits with far more functionality than can be achieved today.


Flexible, Paper-Thin Television

Next to the transistors, wiring is one of the most important parts of an integrated circuit. Although today’s integrated circuits (chips) are the size of a thumbnail, they contain more than 20 miles of copper wiring. Junhao Lin, a Vanderbilt University Ph.D. student and visiting scientist at Oak Ridge National Laboratory (ORNL), has found a way to use a finely focused beam of electrons to create some of the smallest wires ever made. The flexible metallic wires are only three atoms wide: One thousandth the width of the microscopic wires used to connect the transistors in today’s integrated circuits. The discovery gives a boost to efforts aimed at creating electrical circuits on mono-layered materials, raising the possibility of flexible, paper-thin tablets and television displays.

This will likely stimulate a huge research interest in monolayer circuit design,” Lin said. “Because this technique uses electron irradiation, it can in principle be applicable to any kind of electron-based instrument, such as electron-beam lithography.”

One of the intriguing properties of monolayer circuitry is its toughness and flexibility. It is too early to predict what kinds of applications it will produce, but “If you let your imagination go, you can envision tablets and television displays that are as thin as a sheet of paper that you can roll up and stuff in your pocket or purse,” commented Sokrates Pandelides, Professor at Vanderbilt University and Lin’s Advisor.
Lin’s achievement is described in an article published online by the journal Nature Nanotechnology.