ED4-1-INV

Towards Ultra-High-Speed Superconductor Computing ~ Computer Architecture Perspective ~

Dec.2 18:00-18:25 (Tokyo Time)

*Koji Inoue1

Kyushu University1

Moore’s Law, doubling the number of transistors in a chip every two years, has so far been contributed to the evolution of computer system architectures, e.g., introducing manycore accelerations, employing large on-chip caches, and increasing DRAM (or main memory) capacity. The growth of such hardware implementation makes a lot of optimization opportunities available to software developers. Unfortunately, we cannot expect sustainable transistor shrinking anymore, i.e. the end of Moore’s Law will come. Although still device and manufacturing technologies have been progressing, some researchers predict that transistor shrinking may stop at around 2025 to 2030 due to physical or economic reasons. The goal of this research is to open the door for post-CMOS ultra high-performance, low-power computing. Our approach stands on device/circuit/architecture level co-designs by targeting an emerging device called SFQ (Single-Flux-Quantum). This talk introduces the current status of our research for SFQ accelerator designs such as a successful demonstration of 32 GHz 6.5 mW gate-level-pipelined 4-bit microprocessor and an over 50 GHz neural network processing unit. Then, we discusses future directions of cryogenic computing platforms.