attojoule Cryogenic Communication

ended

The end of Moore’s law has led to unsustainable growth in data centre and high-performance computing (HPC) power consumption. Within the post-CMOS technologies addressing this energy crisis, those based on superconductivity are among the most promising ones. Superconducting classical computing based on single flux quantum (SFQ) pulses is a technology enabling clock speeds exceeding 100 GHz, at extreme power efficiency. Rather than compete with CMOS head on, our vision is that SFQ cores should act as coprocessors in existing HPC architectures, much like GPUs do today. Superconducting circuits are also a leading candidate for implementations of quantum computing (QC), which promises to solve certain classically intractable problems. There, SFQ logic offers a natural solution for tight integration of the signal processing required for control and readout of large-scale error-corrected superconducting quantum processors. In both HPC and QC, expanding to large scale is essential for practical impact, and thus, high-bandwidth data transfer to the cryogenic coprocessor is a key bottleneck. In aCryComm we combine top-level European expertise in HPC, superconducting electronics, quantum computing, and photonics to create an optical data bus between conventional HPC and cryogenic SFQ circuits. We expect the optical data link to outperform conventional electrical connections in bandwidth, energy consumption, thermal loading, and physical footprint. To this end, we will develop opto-electric and electro-optic interfaces, culminating in demonstrators that quantitatively characterize the data bus performance. Thanks to the inter-disciplinary composition of the consortium, we are also able to produce and promote a plan for the long-term exploitation of the cryogenic data bus in HPC and QC. We also suggest paths to commercializing our technologies, taking advantage of the unique possibility the consortium offers for transferring R&D to production in the same European facilities.