computer chip

Engineers develop new revolutionary chip design for better AI computing

  • Twitter
  • Facebook
  • Google+
  • Pinterest

Circuitry, crafting ways to deliver electricity from Point A to Point B, is the foundation of the modern world. With the age of electronics helping mankind to automate factories, standards of living greatly improved. However, even since the days of the Edison-Tesla rivalry, we have always sought to make circuits more efficient. Hence, in the case of computers, this means building a better computer chip.

Computer chips of all types stem from a common origin. The general consensus is that the modern computer chip came into being by the ’60s. This was due to the implementation of multiple circuits embedded in a semiconductor material.

A recent breakthrough in the chip hardware came from EPFL

Engineers at EPFL (École Polytechnique Fédérale de Lausanne), one of the world’s leading universities may have stumbled onto gold. Their  Laboratory of Nanoscale Electronics and Structures (LANES) just developed a new, two-in-one chip. This new invention essentially fuses two essential processes into one unit.

Normally, traditional chips follow the von Neumann style of design. This means they have separate compartments for logic functions and memory. The former basically deals with the chips ability to process and interpret information sets. The latter refers to the storable data on the circuit for future task execution. Together, these two compartments, with totally different functions, make up so-called logic-in-memory architecture. One of the biggest bottlenecks in this process is the energy and time wasted to relay data between the two sets.

EPFL’s team has found a way to use 2D (two dimensional) materials in the chip interface. This allows them to house both logic functions and memory functions together. This way, they aim to conserve energy and resources having to transfer data across a larger gap, making the chip more efficient.

This material is called MoS2, and allows for circuits to function much more efficiently

At just a mere 3 atoms in width, MoS2 is so thin that it is practically 2D. Replacing the reliance on silicon, a good semiconductor, meant finding a better one. Only recently discovered by scientists to have good charge bearing capacities, this material is a gamechanger for microchips. And all because it perfects the way current-generation hardware designs.

Most chips require the implementation of a floating-gate field-effect transistor (FGFET), which fuses several functions across one chip. Usually, such FGFET’s make up components of smartphone and camera memory storage (RAM). But, with this stunning new advancement, lots of precious processing power is freed up. That means with this layout, coupled with MoS2’s electrochemistry, entire systems could become faster and leaner.

One of the most exciting potential applications of this technology is in AI (Artificial Intelligence). This phenomenon, now more widely popular than ever, is when a computer, with its memory and logic, is allowed to execute decisions and “learn” from the outcome. And AI is, quickly rising as a vital part of every computer industry, like graphics cards for video games to factory automation.

While unveiling their new invention, the team also explained its aims and its manufacturing process

Andras Kis, the leading member of EPFL’s LANES project, recently announced the creation. During his talk, he mentioned how the new chip streamlines almost any machine using it. EPFL’s goal, he echoed, was to help make machines more efficient, smarter, and sleeker. Furthermore, to keep costs as low as possible, the manufacturing process is quite efficient too. Over 80 chips can come from a single batch of MoS2 at once.

For more on the future and present of computer chips, stay tuned!


error: Content is protected !!