To counter the pressure on global energy resources due to increasing computational demands (yes, AI, we're looking at you), research institute imec suggests a radical shift away from traditional computing methods.
Their solution, detailed in the engineering magazine IEEE Spectrum, involves exploiting the fundamental properties of superconductors to greatly reduce power consumption, thus creating an innovative superconducting processor.
This promising technology has been in development for a couple of years so far and uses standard CMOS manufacturing techniques that can potentially deliver computing power one hundred times more energy efficient than the best current chips, and could “lead to a computer that fits the computing resources of a data center in a system the size of a shoebox.”
20 exaflops
Imec's research involved designing a new type of processor from scratch, with close collaboration between CMOS engineers and entire development teams.
Imec switched from using niobium as a superconducting material to the related compound niobium titanium nitride, as it can withstand the temperatures used in CMOS manufacturing without losing its superconducting capabilities.
The resulting superconducting chip, optimized for AI processors, resembles a typical 3D CMOS system-on-chip. However, a significant difference is that most of the chip must be immersed in liquid helium to ensure an optimal operating temperature, close to 4 Kelvin.
Compared to conventional CMOS chips, superconductors dissipate only a fraction of the energy as heat. This feature allows the ability to stack computing chips directly on top of each other, reducing the physical footprint while preserving the density gains caused by Moore's Law.
Imec estimates that a stack of 100 superconducting boards, each containing superconducting processing units (SPUs), superconducting SRAM and DRAM memory stacks and housed in a cooling environment the size of a 20 x 20 x 12 centimeter shoebox, It can perform a staggering 20 exaflops on BF16. numeric format. This is 20 times the capacity of the most powerful supercomputer currently in existence (Frontier at Oak Ridge National Laboratory, Tennessee), which consumes a total of 500 kilowatts of energy. It also offers 100 times the level of energy efficiency.
Imec's innovation is not intended to replace existing CMOS computing technologies, but rather to complement them. Researchers at the institute believe the technology will boost artificial intelligence and machine learning, and will integrate seamlessly with quantum computers.
The smaller data center footprint, made possible by superconducting digital technology, will allow these systems to be placed closer to their target applications, ultimately driving colossal advances in industries such as agriculture, healthcare and energy.