The prototype superconducting 4bit processor developed at Yokohama National University is based around unshunted Josephson junction (JJ) devices manufactured using a Nb/AlOLₓ/Nb superconductor IC fabrication process. These were used to build energy-efficient superconductor digital electronic structures, adiabatic quantum-flux-parametron (AQFP), for a hybrid RISC data flow machine.
Despite cooling to 4.2K, the low switching energy of 1.4 zJ for each JJ junction means the overall system power consumption is significantly lower than current room temperature semiconductor devices.
This opens up the possibility of ultra-low-power, high-performance microprocessors and other computing hardware for the next generation of data centres and communication networks as well as neuromorphic machine learning chips.
“The digital communications infrastructure that supports the Information Age that we live in today currently uses approximately 10% of the global electricity. Studies suggest that in the worst case scenario, if there is no fundamental change in the underlying technology of our communications infrastructure such as the computing hardware in large data centres or the electronics that drive the communication networks, we may see its electricity usage rise to over 50% of the global electricity by 2030,” said Christopher Ayala, an associate professor at Yokohama National University, and lead author of the study.
The research was published in Journal: IEEE Journal of Solid-State Circuits. “In this paper, we wanted to prove that the AQFP is capable of practical energy-efficient high-speed computing, and we did this by developing and successfully demonstrating a prototype 4-bit AQFP microprocessor called MANA (Monolithic Adiabatic iNtegration Architecture), the world’s first adiabatic superconductor microprocessor,” said Ayala.
“The demonstration of our prototype microprocessor shows that the AQFP is capable of all aspects of computing, namely: data processing and data storage. We also show on a separate chip that the data processing part of the microprocessor can operate up to a clock frequency of 2.5 GHz making this on par with today’s computing technologies. We even expect this to increase to 5-10 GHz as we make improvements in our design methodology and our experimental setup,” said Ayala.
“We demonstrate register file R/W access, ALU execution, hardware stalling, and program branching performed at 100 kHz under the cryogenic temperature of 4.2 K. We also successfully demonstrated a high-speed breakout chip of the microprocessor execution units up to 2.5 GHz,” he added.
Next: Demonstrating a superconducting 4bit processor
“These demonstrations show that AQFP logic is capable of both processing and memory operations and that we have a path toward practical adiabatic computing operating at high-clock rates while dissipating very little energy.
However, superconductors require extremely low temperatures to operate successfully, which would increase the overall power consumption of the system.
“The AQFP is a superconductor electronic device, which means that we need additional power to cool our chips from room temperature down to 4.2 Kelvin to allow the AQFPs to go into the superconducting state. But even when taking this cooling overhead into account, the AQFP is still about 80 times more energy-efficient when compared to the state-of-the-art semiconductor electronic devices found in high-performance computer chips available today.”
The switching energy of 1.4 zJ per JJ comes when driven by a four-phase 5-GHz sinusoidal AC-clock at 4.2 K.
Now that the team has proven the concept of a superconducting 4bit processor, the aim is to optimize the chip and build larger and faster processors.
“We are now working towards making improvements in the technology, including the development of more compact AQFP devices, increasing the operation speed, and increasing the energy-efficiency even further through reversible computation,” Ayala said. “We are also scaling our design approach so that we can fit as many devices as possible in a single chip and operate all of them reliably at high clock frequencies.”
In addition to building standard microprocessors, the team is also interested in examining how AQFPs could assist in other computing applications such as neuromorphic computing hardware for artificial intelligence as well as quantum computing applications.
- SYNTHESIS TOOL FOR LOW POWER ADIABATIC LOGIC SHOWS 10,000X IMPROVEMENT
- LOW COST SUPERCONDUCTOR BOOSTS POWER CONNECTIONS
- WHY COLD COMPUTING MATTERS FOR THE NEXT GENERATION OF COMPUTING
Other articles on eeNews Power
- Polymer battery deal for HAPS basestation in the sky
- Looking to AI standards in wind power
- Blue World raises €6.4m for methanol fuel cell factory
- US looks to NFA cobalt-free cathode tech
- European companies drive open source virtualisation and consolidation for the grid