Press Release

Supercomputer Pegasus Marks Japan’s Highest Performance per Watt

Pegasus, the latest supercomputer from the University of Tsukuba, achieved the highest ranking in Japan (12th globally) on the November 2023 edition of the Green500 list, showcasing an impressive 41.12 billion floating-point operations per second (GFLOPS) per watt. The Green500 list is a semiannual compilation, globally recognized for evaluating supercomputers based on their energy efficiency.

In December 2022, University of Tsukuba’s newest supercomputer, Pegasus, was installed at the Center for Computational Sciences in Tsukuba, Japan. It integrates the world’s highest-performing graphics processing unit (GPU) and central processing unit, advanced nonvolatile memory, and the highest-performance interconnection network, making Pegasus the most advanced supercomputer to date.

The Green500 list, one of the world’s most advanced supercomputer rankings, is published annually in June and November. This competition is based on the level of computational performance per unit of power consumption (W), specifically for solving a simultaneous linear equation with a dense matrix. It has garnered attention as an indicator of “energy efficiency of supercomputers” from the perspective of energy issues, among other factors.

The latest Green500 list was announced at SC23, an international conference on supercomputer technology and computational science held in Denver, Colorado, USA, on November 13, 2023. Pegasus achieved an impressive 41.12 billion floating-point operations per second (GFLOPS) per watt, surpassing the previous record set in Japan and earning the title of the greenest supercomputer in Japan (12th in the world). Pegasus offers various usage programs in Japan and worldwide. This outstanding achievement highlights how the energy-saving technology of the University of Tsukuba, as an early adopter of state-of-the-art GPUs in Japan, contributes to various research projects.


Professor BOKU Taisuk
Director of the Center for Computational Sciences, University of Tsukuba