One company, called Semiqon, is looking at the technology of cryogenic CMOS to potentially develop future silicon-based transistors capable of operating at the ultra-low temperatures required by quantum processors. The transistors are specifically engineered to perform optimally at temperatures of 1 Kelvin and lower, which is the temperatures where most quantum computers operate.

“We get rid of the electrical noise associated with transistors operating at room temperature,” said Himadri Majumdar, CEO of Semiqon, in a recent interview with Design News. Majumdar added that the ability to operate at cryogenic temperatures eliminates the impurities that can adversely impact performance.

According to Majumdar, Semiqon is initially targeting producing cryogenic transistors in its pilot line foundry. While quantum computing is an obvious market, Majumdar also sees potential uses in space applications where cryogenic-level performance is needed. You can a view a video of Semiqon’s technology here.

Reversible computing

Also being investigated is applying the concept of reversible computing to produce chips that dissipate zero or next to zero heat, which recovers energy through reversible logic gates via a pendulum-like resonator device. A startup called Vaire Computing is trying to apply reversible computing concepts to develop chips that dissipate little or no heat, which would solve the problem of heat buildup in AI applications and lessen the need for massive cooling devices.

According to the company, reversible computing has been under investigation for a long time. However, the issue has been finding a suitable business-use case, which AI now provides, said Rodolfo Rosini,  CEO of Vaire Computing. He added that the hurdles with developing chips capable of reversible computing involve scalability and developing a resonant device.

Energy-efficient processors

While minimizing chip power consumption is a key goal, there’s also the issue of designing chips to handle ever-larger amounts of data to effectively manage power use. Thus, some research efforts are taking a fresh look at chip architectures in an effort to manage power efficiency.

One such effort, spawned at well-known Carnegie-Mellon University, produced a startup company called Efficient Computer, which is tackling head-on the issue of heat generated by programmable general-purpose processors through what is termed as a Fabric processor architecture. Brandon Lucia, CEO of Efficient Computer, said in a recent interview with Design News the company has developed its initial prototype processor, called EI. This processor is designed to enhance computing by intelligently distributing workloads and connecting instructions to reflect the application dataflow. According to the company, this design ensures high energy efficiency and strong performance, all while maintaining full programmability.

In conjunction with the EI processor, Efficient Computer has also developed a compiler, named effcc, that enables the design engineer to maximize the benefits of the EI processor, without requiring having to rewrite code. The compiler works with most major embedded languages.

The processor is aimed at wearables, edge networks, spacecraft, and other applications. According to Lucia, the processor technology can be scaled up, and future developments could potentially include chiplets.

According to designnews.com