Power/Performance Bits: Feb. 16

Superconducting microprocessor
Researchers at Yokohama National University created a superconducting processor with zero electrical resistance.

Huge amounts of power are being used by computers today, and compared to the human brain, they are many orders of magnitude less efficient. Superconductors have been a popular approach to making computers more efficient, but this requires extreme cooling down to 10 kelvin.

What makes this different is that they sought to create a processor that’s adiabatic, meaning that, in principle, energy is not gained or lost from the system during the computing process.

The new microprocessor prototype, called MANA (Monolithic Adiabatic iNtegration Architecture), is composed of superconducting niobium and relies on hardware components called adiabatic quantum-flux-parametrons (AQFPs). Each AQFP is composed of a few fast-acting Josephson junction switches, which have a switching energy of 1.4 zJ per switch. The MANA microprocessor consists of more than 20,000 Josephson junctions (or more than 10,000 AQFPs) in total. It is a hybrid of RISC and dataflow architectures operating on 4-b data words.

Today, the processor is running at 2.5GHz, but they expect this to increase to the 5GHz-10GHz range. Even taking the costs of cooling into consideration, they say that it is about 80 times more energy-efficient when compared to a comparable 7-nm FinFET device.

Since the MANA microprocessor requires liquid helium-level temperatures, it’s better suited for large-scale computing infrastructures like data centers and supercomputers, where cryogenic cooling systems could be used.

Integrating LEDs into chips
When functionality is split across multiple chips or die, production costs increase rapidly. This has fueled the long-running trend of increased integration. But not everything is as easy to integrate as digital electronics. Optics has been an area that has presented extreme difficulties because silicon has an indirect bandgap and does not normally emit light. But researchers at MIT have fabricated a silicon chip with fully integrated LEDs, bright enough to enable state-of-the-art sensor and communication technologies. The advance could lead to not only streamlined manufacturing, but also better performance for nanoscale electronics.

Optical components are usually fabricated using III-V semiconductors. Jin Xue, a PhD student in MIT’s Research Laboratory of Electronics (RLE), led the research that came up with a silicon-based LED with specially engineered junctions that enhance brightness. This created enough light to transmit a signal through 5 meters of fiber optic cable. GlobalFoundries manufactured the LEDs right alongside other silicon microelectronic components, including transistors and photon detectors.

While the LED is not yet as efficient as those manufactured using a III-V material, Professor Rajeev Ram, who leads the Physical Optics and Electronics Group in RLE, is hopeful for the future. According to Ram, III-V semiconductors have nonideal surfaces, riddled with “dangling bonds” that allow energy to be lost as heat rather than as light. In contrast, silicon forms a cleaner crystal surface, and it is this what they hope to make use of in the future.

Wireless for Industry 4.0
Researchers at Universitat Oberta de Catalunya and Technische Universität Darmstadt are working on ways to make wireless communications more feasible for industrial manufacturing plants. The group aims to make wireless technologies that have power and reliability comparable to fiber optics and that could replace cabled connections.

The team created a parameterization of a millimeter-band signal propagation model, a wireless technology capable of transmitting a huge amount of data per second, in an industrial environment.

“This study is aimed at making communication less expensive and more flexible by incorporating mobile devices into the manufacturing process, something that could be very useful in moving towards Industry 4.0, since it allows, for example, connecting freely movable robotic arms to the production process or establishing communications for data reporting, and controlling or stopping the different components of the process in an emergency. But it could also allow the worker to be a part of the process,” said Cristina Cano, a researcher from the Wireless Networks (WiNe) group at the Internet Interdisciplinary Institute (IN3) at Universitat Oberta de Catalunya.

But industrial settings have unique environments that change the way wireless signals propagate, Cano added. “There are several models for millimeter bands in office and urban settings, but there are hardly any in industrial settings. These sorts of facilities differ in many ways which could interfere with the behavior of the wireless signal, such as the height of the ceiling, the material of the walls and floors, or the type of machinery they contain. Our research has allowed us, for the first time, to establish the parameters for an industrial environment.”

The researchers were able to measure the behavior of this type of signal at the ALBA synchrotron, an electron accelerator located in Barcelona that has characteristics that resemble different industrial environments in large production plants.

They found that typical surfaces in industrial plants, such as reflective pipes, are very beneficial for this type of communication: “Specifically, we were able to establish a 110-meter link, the largest communication link achieved with the IEEE 802.11ad standard to date,” said Cano.

The team has made their model accessible to the research community to help draft protocols that can ensure reliability in industrial wireless communications.